Anda di halaman 1dari 351

LEADING-EDGE ELECTRIC POWER RESEARCH

LEADING-EDGE ELECTRIC POWER RESEARCH

CIAN M. O'SULLIVAN
EDITOR

Nova Science Publishers, Inc.


New York

Copyright 2008 by Nova Science Publishers, Inc.

All rights reserved. No part of this book may be reproduced, stored in a retrieval system or transmitted in any form or by any means: electronic, electrostatic, magnetic, tape, mechanical photocopying, recording or otherwise without the written permission of the Publisher. For permission to use material from this book please contact us: Telephone 631-231-7269; Fax 631-231-8175 Web Site: http://www.novapublishers.com NOTICE TO THE READER The Publisher has taken reasonable care in the preparation of this book, but makes no expressed or implied warranty of any kind and assumes no responsibility for any errors or omissions. No liability is assumed for incidental or consequential damages in connection with or arising out of information contained in this book. The Publisher shall not be liable for any special, consequential, or exemplary damages resulting, in whole or in part, from the readers use of, or reliance upon, this material. Independent verification should be sought for any data, advice or recommendations contained in this book. In addition, no responsibility is assumed by the publisher for any injury and/or damage to persons or property arising from any methods, products, instructions, ideas or otherwise contained in this publication. This publication is designed to provide accurate and authoritative information with regard to the subject matter covered herein. It is sold with the clear understanding that the Publisher is not engaged in rendering legal or any other professional services. If legal or any other expert assistance is required, the services of a competent person should be sought. FROM A DECLARATION OF PARTICIPANTS JOINTLY ADOPTED BY A COMMITTEE OF THE AMERICAN BAR ASSOCIATION AND A COMMITTEE OF PUBLISHERS. LIBRARY OF CONGRESS CATALOGING-IN-PUBLICATION DATA Leading-edge electric power research / Cian M. O'Sullivan, Editor. p. cm. Includes index. ISBN-13: 978-1-60692-427-3 ISBN-10: 1-60021-981-0 (hardcover) 1. Electric power systems. I. O'Sullivan, Cian M. TK1005.L39 2008 621.31--dc22 2007033915

Published by Nova Science Publishers, Inc.

New York

CONTENTS
Preface Expert Commentary Commentary A Analysis and Characterization of Complex Inter-Area Oscillations from Measured Data: A Time-Frequency Perspective A. R. Messina, E. Barocio and M. A. Andrade Research and Review Articles Chapter 1 Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners for Wall-Fired Boilers Zhengqi Li Fuel Cell Combined Cycle Power Generation System Installed into Micro-Grid Shinya Obara Electricity from Renewable Energy Sources: A Multi-Criteria Evaluation Framework of Technologies Fausto Cavallaro Gas Turbines and Electric Distribution System Francisco Jurado Micro CCHP: Future Residential Energy Center R. Z. Wang and D. W. Wu Sensitivity Calculation in Real Time Transmission Network and Energy Markets Jizhong Zhu Wide-Area Monitoring and Analysis of Inter-Area Oscillations Using the Hilbert-Huang Transform A. R. Messina, M. A. Andrade and E. Barocio 5 vii

Chapter 2

127

Chapter 3

173 205 239

Chapter 4 Chapter 5 Chapter 6

265

Chapter 7

285

vi Chapter 8 Index

Contents Unconventional Problems in Power Systems Protection Mahmoud Gilany and Mohamed A. Mahmoud 317 331

PREFACE
This book presents new and significant research on electric power. The world is becoming increasingly electrified. For the foreseeable future, coal will continue to be the dominant fuel used for electric power production. The low cost and abundance of coal is one of the primary reasons for this. Electric power transmission, a process in the delivery of electricity to consumers, is the bulk transfer of electrical power. Typically, power transmission is between the power plant and a substation near a populated area. Electricity distribution is the delivery from the substation to the consumers. Due to the large amount of power involved, transmission normally takes place at high voltage (110 kV or above). Electricity is usually transmitted over long distance through overhead power transmission lines. Underground power transmission is used only in densely populated areas due to its high cost of installation and maintenance, and because the high reactive power gain produces large charging currents and difficulties in voltage management. A power transmission system is sometimes referred to colloquially as a "grid"; however, for reasons of economy, the network is rarely a true grid. Redundant paths and lines are provided so that power can be routed from any power plant to any load center, through a variety of routes, based on the economics of the transmission path and the cost of power. Much analysis is done by transmission companies to determine the maximum reliable capacity of each line, which, due to system stability considerations, may be less than the physical or thermal limit of the line. Deregulation of electricity companies in many countries has led to renewed interest in reliable economic design of transmission networks. Chapter 1 - The kind of swirl coal burners is given. Radial-biased-combustion and centrally-fuel-rich swirl coal combustion technology was developed. In the air and the airparticle test facilities, the single sensor hot-film and the anemometers were used to measure air and air-particle flows in the near-burner region of different swirl burners. Both cold air flow and reacting flow experiments were performed in the industrial 50, 220, 410, 670 and 1025 ton per hour boilers. On an air-particle test facility, the characteristics of the pulverizedcoal concentrator with cone vanes were investigated. The influence of structure parameters, such as run parameters such as swirling vane angle and burner cone angle and length, and run parameters, such as non-swirl secondary air, central air and air supply, and primary air flow type on divergent angles, diameter and length of the central recirculation zone, mixing characteristic of the primary air and the secondary air, in-situ gas temperature and NOx formation near the burner zone, carbon in ash and NOx emission of boilers was determined with the radial-biased-combustion burner. The difference characteristics of gas/particle flow

viii

Cian M. O'Sullivan

and coal combustion of the centrally-fuel-rich and dual register burners were obtained. The experimental results show that the two new burners simultaneously have the ability of high combustion efficiency, flame stability, low NOx emission and resistance to slagging and high temperature corrosion. The air-surrounding-fuel combustion theory was put forward. Chapter 2 - The introduction to urban areas of the micro-grid system has the following characteristics. (a) The distance between the heat-supply side and the heat-demand side is short, and effective utilization of exhaust heat is possible. (b) It is linked with the load leveling of the existing large-sized electric power facilities. (c) Since a facility suitable for the energy-demand characteristics of a region can be installed, energy efficiency may increase and facility costs may decrease. The micro-grid using a proton exchange membrane type fuel cell (PEM-FC) may greatly reduce environmental impact. However, when connecting an energy system to the micro-grid of a city area and operating, partial load operation occurs frequently and power generation efficiency falls. And, the electrode material (especially the catalyst material and the proton exchange membrane) of PEM-FC is expensive, and its system is complex. Consequently, it is necessary to connect two or more power generation systems to the micro-grid, and to design optimization of an operation plan for the purpose of maximization of power generation efficiency. Therefore, the methods of an improvement of the efficiency of the power generation system connected to the micro-grid installed into a city area are described. In this chapter, it consists of subjects of three studies on the micro grid. In these studies described in this chapter, the improvement of the subject of the micro grid is tried by combining fuel cell and other power equipment. Section 1 describes "Operation Plan of Micro Grid Using PEM-FC/Diesel Engine Generator Combined System." Section 2 describes "Carbon-Dioxide Emission Characteristic of Micro Grid Using PEMFC/Hydrogenation City gas-Engine Combined System." Section 3 describes "Dynamic Characteristics of Micro Grid Using PEM-FC/Woody Biomass Engine Combined System." Chapter 3 - The energy policy of many Western governments aims to diversify supply and reduce dependence on foreign sources and thus to maximise benefits from internal resources. Undoubtedly, the main strategy underlying this is one that seeks to optimise the use of renewable energy sources (RES). The development of these sources, as well as their market penetration, depends however not only on political will but also on sound management of energy demand in order to rationalise and stabilise energy consumption. In addition to fortifying the guaranteed energy supply, RES represent a potential that cannot be overlooked. This lies in their ability to reduce greenhouse gas emissions and thus to stem the growing trend of global warming, one which has accelerated particularly in recent years and which is due mainly to the use of fossil fuels for producing electricity. The use of RES for the production of electric power brings huge benefits both in terms of environmental protection as well as savings in non-renewable resources. Nevertheless, the very nature of RES raise technical and economic problems that create a considerable gap between their potential capacity and ways to feasibly exploit them. Their many different forms and the ways in which they may be used have to be carefully examined in order to evaluate the costs and other technical and environmental factors involved. The planning and appraisal of sustainable energy projects involve rather complex tasks. This is due to the fact that the decision making process is the closing link in the process of analysing and handling different types of information: environmental, technical, economic and social. Such information can play a strategic role in steering the decision maker towards one choice instead of another. Some of these variables (technical and economic) can be

Preface

ix

handled fairly easily by numerical models whilst others, particularly ones relating to environmental impacts, may only be adjudicated qualitatively. In many cases therefore, traditional evaluation methods and the chief economic and financial indicators are unable to deal with all the components involved in an environmentally valid energy project. Multicriteria methods provide a flexible tool that is able to handle and bring together a wide range of variables appraised in different ways and thus offer valid assistance to the decision maker in mapping out the problem. Chapter 4 - Lately, the use of gas turbines following the deregulation of the electricity supply industry has become greater quickly. The motivation for modeling the gas turbines and their controllers is determinant to the interpreting of their impacts on distribution systems. The model predictive control (MPC) is used to damp the oscillation when the power distribution system is subjected to a disturbance. MPC is selected because it can explicitly handle the nonlinearities, and constraints of many variables in a single control formulation. The IEEE 13 node power distribution system is employed to demonstrate the effectiveness of MPC to damp the oscillations of gas turbines. Among fossil fuels, gas is the most quickest, with a growth rate nearly double that of coal and oil. The electricity generation field is the leading market for gas. The natural gas business has a great interaction with the electricity market in terms of fuel consumption and energy conversion. On the other hand, the transmission and distribution activities are very similar with the natural gas transportation through pipelines. The power losses in gas and electric systems are compared. It is also demonstrated that the electricity system results more convenient for longer distances of gas wells from electricity consumption area. Chapter 5 - Combined cooling, heating and power (CCHP) system, as a distributed energy system, can work all the year and provide cooling/hot-water/power in summer, heating/hot-water/power in winter and hot-water/power in other seasons. In CCHP systems, the total energy efficiency increases to over 85%, while the average energy efficiency of conventional fossil fuel fired electricity generation systems is around 40%. The energy efficiency promotion of CCHP systems results in emission reduction compared to the conventional methods of generating heat and electricity separately. And as a distributed energy resource, CCHP systems also increase in the reliability of the energy supply. With the overall development of CCHP systems and related technologies, the utilization of micro CCHP systems in the residential sector is emerged as a growing potential. The article focuses on the micro CCHP systems for single-family applications (around 10 kW) and multi-family or residential district applications (under 200 kW). The status quo of micro CCHP systems is briefly presented and diverse combinations of technologies existing in applications or experimental units are listed through comprehensive literature review. Various technologies available or under development are introduced, such as reciprocating internal combustion engine, micro-turbine, fuel cell, Stirling engine, absorption chiller, adsorption chiller and so on. Afterward, the tendency and issues of micro CCHP systems are discussed. The review shows that micro-CCHP applications are entering into average families as a nextgeneration residential energy supply center. Chapter 6 - The calculations of the several sensitivities such as loss sensitivity, voltage sensitivity, generator constraint shift factor, and area based constraint shift factor become very important in energy management system (EMS) and energy markets. This chapter focuses on the analysis and implementation details of the above-mentioned sensitivities calculations in the practical transmission network and energy markets. The power operator

Cian M. O'Sullivan

uses them to study and monitor market and system behavior and detect possible problems in the operation. These sensitivities calculations are also used to determine whether the on-line capacity as indicated in the resource plan is located in the right place on the network to serve the forecasted demand. If the congestion or violation exists, the generation scheduling based on the sensitivities calculations can determine whether or not a different allocation of the available resources could resolve the congestion or violation problem. This chapter also comprehensively discusses how to compute and use the sensitivities under the different references such as the market-based reference, and the energy management system based reference. The calculation results of the several sensitivities are illustrated using the IEEE 14 bus system and AREVA T & D 60-bus system. Chapter 7 - Many transient processes in power systems involve phenomena that vary in time and space in complicated ways. Comprehensive monitoring of large-scale power systems by means of properly placed time-synchronized phasor measurement units (PMUs) provides the opportunity to analyze and characterize complex inter-area swing dynamics involving all or most of the power system. Wide-area real-time monitoring may prove invaluable in power system dynamic studies by giving a quick assessment of the damping and frequency content of dominant system modes after critical contingencies. Measured data, however, may exhibit quite different dynamics at each system location or exhibit abrupt changes, dynamic irregularities, or be complicated by nonlinear trends or noise. Traditional Fourier and Prony methods for system identification are unable to resolve the localized nature of these processes and hence provide little useful information concerning the nature of noisy, time-varying oscillatory processes. In this Chapter, a new method for analyzing the temporal dynamics of nonlinear and nonstationary inter-area oscillations using a local empirical mode decomposition (EMD) method and the Hilbert transform is presented. Two novel algorithms are developed to address nonlinear and non-stationary issues. The first method is a local implementation of the empirical mode decomposition technique. The second is an algorithm to compute the Hilbert transform using finite impulse response (FIR) filters. By combining these approaches, the method can be used to analyze complex signals for which the conventional assumptions of linearity and stationarity may not apply and can be implemented for on-line estimation of modal damping and frequency using synchronized wide-area measurement systems. The physical mechanism underlying nonlinear time-varying inter-area oscillations is investigated and methods to characterize the observed oscillatory phenomena in terms of physically meaningful modal components are proposed. Emphasis is placed on identifying modal content in the presence of noise and nonlinear trends. Issues concerning the implementation of the method and numerical considerations are also discussed. As specific applications, data obtained from PMU measurements from a real event in the northern systems of the Mexican interconnected system are used to examine the potential usefulness of nonlinear time series analysis techniques to characterize the spatio-temporal characteristics of the observed oscillations and to determine the nature and propagation of the system disturbance. The efficiency and accuracy of the method is demonstrated by comparison to other approaches. Chapter 8 - Of the numerous electric power faults an Electric Engineer comes across in a life time, only a few of these faults are memorable- the rest being routine ones. In this chapter, some of those unconventional faults, which are mainly related to power system protection, are presented. The chapter presents five case studies of actual field incidents

Preface

xi

rather than hypothetical scenarios. The objective of the chapter is to present a typical approach for analyzing the faults in power systems.

In: Leading-Edge Electric Power Research Editor: C. M. OSullivan, pp. 1-3

ISBN: 978-1-60021-981-8 2008 Nova Science Publishers, Inc.

Expert Commentary

ANALYSIS AND CHARACTERIZATION OF COMPLEX INTER-AREA OSCILLATIONS FROM MEASURED DATA: A TIME-FREQUENCY PERSPECTIVE
A. R. Messina, E. Barocio and M. A. Andrade
Large sparse power systems form an extremely complex dynamical system which usually possess many degrees of freedom and poses a challenge for simulation and analysis. Forced complex oscillations triggered by the loss of major system resources may manifest highly complex spatial and temporal dynamics and involve a large number of machines and take place over a great range of time and time scales. Proper understanding of the underlying dynamics causing these oscillations requires investigation of the various types of temporal nonlinear interactions involving the fundamental modes of the system. Such features may be obscured or distorted in the normal spectral analysis approach. The analysis of spatio-temporal dynamic patterns is important for many reasons. Nonlinearity causes the fundamental waves or temporal modes to interact, leading to frequency and amplitude modulation and to a phase relationship known as quadratic phase coupling between the frequency components involved. Mounting evidence suggest that these interactions can have a significant impact on system performance such as the modal content of the observed oscillations and may the design of controllers. Further, it is also possible that nonlinearity contributes to non-stationary behavior in the record. Characterization of non-stationary behavior is, on the other hand, required for both, detailed understanding of the mechanisms leading to instability and addressing the key questions of how the temporal oscillation modes evolve over time. Accurate tracking of temporal behavior allows replicating the events leading to the observed oscillations, and analyzing the specific condition, control action or device on modal content. With the recent development and application of sophisticated measurement systems, the detection and characterization of the temporal evolution of temporal oscillations is becoming ever more important. Non-stationary system behavior may result from the effects of sequential faults, control actions, and changes in system topology and operating conditions. The issue of stationary is particularly important in studying the system response to large and abrupt changes in system

A. R. Messina, E. Barocio and M. A. Andrade

topology or operating conditions, and in tracking the system response to sequential faults. This information may be used to understand the source mechanism of oscillations, to examine dynamic trends and phase relationships between key system signals, and to detect and recognize instability signatures in the dynamic activity of the system. Detecting and identifying sources of nonlinearity and nonstationarity in observed time series are difficult problems. In large, loosely connected power systems, the analysis and characterization of inter-area oscillations from measurements is a formidable challenge. Measured data is noisy, non-stationary and often of limited duration. Extracting the dynamics of interest from a limited and usually reduced number of measurements is a complex problem. This is particularly true in the study of measured, inter-swing dynamics in which local and inter-area motions may participate in the observed oscillations. The detection and characterization of temporal oscillations in measured and simulated data is greatly complicated by non-stationary variations in system dynamic behavior. Postmortem data from recent wide-area electrical disturbances shows that complex oscillatory processes are often accompanied by short-term, irregular, event-type features that make the on-line analysis and control of transient performance very difficult. Successful analysis of highly complex dynamic events requires analysis approaches with high levels of sophistication including the ability to treat nonlinear and nonstationary data, increased time and frequency resolution, accuracy, and ease of implementation among other features. In this contribution, a critical review of methods for the analysis, modeling and characterization of transient processes in power systems is presented with an emphasis on methods of analysis of nonlinear, time-varying time series. Attention is focused on the most promising advances and areas where knowledge is still insufficient or incomplete. Recent developments and future challenges are emphasized. Several time and frequency representations have been explored over the last few years to analyze dynamic processes that are characterized by nonlinear and non-stationary characteristics. Most of them have been developed for the analysis of linear system response and are available in many commercial packages. However, its effectiveness is often hampered by the presence of non-stationarity in the process and the validity is limited to processes showing linear behavior. Two traditional methods, Prony and Fourier-based spectral analysis methods are now used routinely as tools for investigation and characterization of inter-area oscillations. These methods do, however, suffer from a number of disadvantages, the most important of which is the assumption of stationarity that renders them invalid for many applications. A further complication in the application of these techniques is that power system signals are, at best, quasi-periodic and the periods may change over time. This can result in a degradation of resolution in spectral estimates. Ambient data from on-line measurements and recordings from real events are the best resource to investigate nonlinear, nonstationary effects on system behavior. Recorded data display a very rich structure indicative of a variety of intrinsic nonlinear activity in the system dynamics. The detection of temporal changes in the system response following a sequence of faults requires a monitoring technique that accurately represents the major relations among the process variables. Rapid advances and significant developments in communications technology have led to the fast evolution of wide-area measurements systems. Wide-area measurements provide the opportunity to analyze and characterize inter-area swing dynamics in complex interconnected power systems. This refined information can be used to advantage in system control and

Analysis and Characterization of Complex Inter-Area Oscillations

monitoring, and model validation. For many applications, however, a complete framework for dynamic security monitoring is still evolving. The significant improvement in the quality of power system stability data achieved during recent years has led to a variety of improved methods for estimating spatial and temporal characteristics of the observed oscillations. Methods currently used to predict nonlinear random response include higher order statistics, auto-regressive moving average techniques and joint time-frequency representations. Among such procedures, the continuous wavelet transform, empirical orthogonal analysis and the Hilbert-Huang technique have proved to be useful tools for analyzing and studying the time-varying modal characteristics of lightly damped nonlinear systems subjected to large perturbations where the limitations of Fourier-based techniques make the investigation of localized or time-varying features difficult or uninformative. These methods partially avoid some of the problems associated with conventional spectral analysis in the non-stationary setting above, but they still have some limitations. Application of time-frequency transformation tools has been successful in determining both, the interacting modes and the distribution of non-stationarity. These methods give a sharp description of the time-varying frequency content of the physical phenomena and can be implemented for on-line monitoring of transient processes. Of particular interest are applications where these techniques are used to extract modal information on an on-line basis or for real-time control of system behavior. This is a relatively neglected field for which further research is needed. As these techniques mature, however, their efficient implementation for the analysis of complex processes hinges upon their computational streamlining. Recent studies suggest that various levels of refinement are required according to the application. Also the considerable potential for combining time-frequency analysis with other analytical approaches, for prediction of spatio-temporal patterns, remains unrealized. Future applications could be focused now on the implementation of smart wide-area monitoring, protection and control systems based on sophisticated methods of analysis of time series. While much progress has been made, there are still some important issues that need to be addressed before such approaches can be realized for in-line monitoring and control of transient oscillations.

In: Leading-Edge Electric Power Research Editor: C. M. OSullivan, pp. 5-125

ISBN: 978-1-60021-981-8 2008 Nova Science Publishers, Inc.

Chapter 1

RADIAL-BIAS-COMBUSTION AND CENTRAL-FUELRICH SWIRL PULVERIZED COAL BURNERS FOR WALL-FIRED BOILERS
Zhengqi Li
School of Energy Science and Engineering, Harbin Institute of Technology, 92, West Dazhi Street, Harbin 150001, P. R. China

ABSTRACT
The kind of swirl coal burners is given. Radial-biased-combustion and centrally-fuelrich swirl coal combustion technology was developed. In the air and the air-particle test facilities, the single sensor hot-film and the anemometers were used to measure air and air-particle flows in the near-burner region of different swirl burners. Both cold air flow and reacting flow experiments were performed in the industrial 50, 220, 410, 670 and 1025 ton per hour boilers. On an air-particle test facility, the characteristics of the pulverized-coal concentrator with cone vanes were investigated. The influence of structure parameters, such as run parameters such as swirling vane angle and burner cone angle and length, and run parameters, such as non-swirl secondary air, central air and air supply, and primary air flow type on divergent angles, diameter and length of the central recirculation zone, mixing characteristic of the primary air and the secondary air, in-situ gas temperature and NOx formation near the burner zone, carbon in ash and NOx emission of boilers was determined with the radial-biased-combustion burner. The difference characteristics of gas/particle flow and coal combustion of the centrally-fuelrich and dual register burners were obtained. The experimental results show that the two new burners simultaneously have the ability of high combustion efficiency, flame stability, low NOx emission and resistance to slagging and high temperature corrosion. The air-surrounding-fuel combustion theory was put forward.

Tel.: +86 451 86 41 8854; Fax: +86 451 86 41 25 28; E-mail:lizhengqihit@hotmail.com

Zhengqi Li

1. INTRODUCTION
1.1. Problems of Coal Combustion Technology
In China, utility boilers consume about 27 % of the total coal production and generate about 70 % of the countrys electricity, and emit the majority of pollutants in cities. Chinas coals are mostly low-grade with low calorific value. They either have small amounts of volatile matter or high moisture and/or ash content. Generally the flame from these coals is not stable. The ashes also have low ash fusion points, thus a tendency to slag in the furnace. The power industry requires coal combustion techniques, which have flame stability, no slagging propensities, high combustion efficiency and which meet pollution control standards. It is difficult to attain these requirements simultaneously because they are often in conflict with each other. For example, the quality of coal provided to power plants often fluctuates and is usually low-grade. It is very difficult to keep a stable flame with this type of coal, especially when the load is low. This also lowers combustion efficiency. To achieve a stable flame, supporting fuel is often used which increases the operating cost [1]. Another example of these conflicting requirements is the attempt to decrease the emission of the pollutant NOx. An effective method is to raise the pulverized coal (PF) concentration and delay the mixing of air with the coal stream. However, this method has a propensity to form slag in the furnace and also decreases combustion efficiency [2-5]. Clearly there is a challenge in meeting these conflicting requirements of the industry.

1.2. Characteristics of Swirl Coal Combustion Technology


Swirl coal burners are applied widely on boilers with different capacities. Compared with the tangential firing boilers, boilers with swirl burners have the following main characteristics: (1) They can avoid imbalance of temperature of super heater in tangential firing boilers. (2) Swirl burners form their own coal flame independently and dont influence each other. (3) There is no strict demand for furnace shape and the furnace shape with swirl burners does not need to approach to square. It is convenient to the arrangement of convection heating facilities. (4) Swirl burners are arranged evenly on front-wall and back-wall. The distribution of quantity of heat imported to furnace is relatively uniform. It reduces the slagging trend resulted from excessive high gas temperature in the central region of furnace. (5) There is no need to increase the thermal capacity of a single burner when the unit capacity increases. We just increase the width of furnace and increase the number and rows of burners. Swirl burners take high gas temperature central recirculation zone as thermal source to make air/coal mixing ignite.

1.3. Types of Swirl Coal Burners


According to the types of the secondary air and the fuel concentration of primary air/coal mixing, swirl coal burners can be divided into three categories: general, air-staged combustion and fuel-bias-combustion types.

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners

1. General Type of Swirl Burners This type of burners is the burner whose secondary air is introduced to furnace without being divided into two parts and whose primary air/coal mixing is not concentrated. The following burners have the characteristics: double-volute swirl coal burner, tangential-register vane burner, axial-register vane burner and the volute burner with axial-register vanes. 2. Air-Staged Combustion Type of Swirl Burners This type of burners is the burner whose secondary air is divided into two or more parts and whose primary air/coal mixing is not concentrated, the following burners have the characteristics:
(1) The dual channel swirl burner [7]Without swirling, the primary air is ejected to the furnace. The secondary air is separated to two parts. Most of the secondary air passes through axially bent vanes, and without swirling, the rest secondary air is ejected to the furnace at a large velocity via another air channel. This type of burner has good regulating property. It is the first generation low NOx burner of BW Company. (2) The SM burnerThe primary air is non-swirling and the secondary air swirls via axial bent vanes. The primary and secondary airs account for 80-90% of the total air. The rest of air is introduced to furnace through four symmetrical-arrangement nozzles around the burner. This type of burner is suit to slag tapping and dry ash extraction boilers. (3) The RSFC burner: The primary air is non-swirling and the secondary air is introduced to furnace through three concentric nozzles. Each nozzle has a swirler. Any one or all of the nozzles can be used for the injection of externally recirculated flue gas through the burner. With the RSFC burners, radial stratified combustion is formed in the near burner region, the coal burnout decreases, the NOx emission is reduced, and high temperature corrosion is prevented. (4) Volute-vane swirl burner [7]: The primary air is introduced to furnace through volute. The secondary air is separated into two swirling streams. They are injected to furnace through axially bent vanes in the inner and outer channels respectively. (5) Primary Air Exchange(PAX) burner[9]: For coals with very low volatile matter, such as semi-anthracites and anthracites, further provisions are required to obtain satisfactory ignition performances. With such coals, volatile matter has diminished to such an extent that the heat it contributes to the burner ignition zone is insufficient to sustain ignition. Temperature in the ignition zone has to be increased by controlling heat loss, returning heat from char reactions to the ignition zone, and further preheating the air and fuel prior to its introduction to the burner. BandW developed the PAX burner for these applications. The PAX burner utilizes a device in the burner nozzle to vent off primary air and replace it with hot air to preheat the fuel immediately prior to ignition. The extracted primary air, accompanied by a small percentage of the coal, is vented into the furnace and uses the main burner as its ignition source. The dual register design (derived from the Enhanced Ignition-Dual Register burner) is used.

Zhengqi Li

3. Fuel-Bias-Combustion Swirl Burner This type of burners is the burner whose primary air/coal mixing is separated to increase the fuel concentration. As the fuel concentration increases, the flame stability improves. It has two categories. The primary air/coal mixing is separated into two streams: the fuel-rich and the fuel-lean streams. The fuel-rich stream is introduced to furnace through burner, and the fuel-lean one is introduced to furnace solely at a certain furnace position. This is a high fuel concentration swirl burner. The fuel-rich and fuel-lean streams are introduced to furnace through different channels of burner. This is the fuel-rich and fuel-lean combustion swirl burner. (1) High Fuel Concentration Swirl Burner
(1) Former Soviet Russias high fuel concentration swirl burner: Coal concentration combustion experiments were done on a 30105kw T-210A type boiler. The air/coal mixing was conveyed in a new tubule by compressed air and the fuel concentration of the mixing was as large as 40-50kg (coal)/kg (air). The primary air duct only conveyed air. Before arriving at the burner outlet, pulverized coal was injected to the primary air at a suitable position and a fuel concentration of 0.9 kg (coal)/kg (air) was gotten. The results showed that air/coal mixing ignited earlier and NOx emission sharply decreased.

Figure 1. a) NSZ burner with external fuel enrichment and b) NSW burner with internal fuel enrichment.

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners

(2) Low-NOx burner with external coal concentrator(NSZ burner, see Figure 1a)[10]: Swirl type concentrator is inserted between coal mill and burner. Enriched coal-air mixing flows through the burner fuel nozzle. In order to intensify mixing of fuel with hot internal recirculation gas, this enriched mixing is supplied via four separate ducts, parallel to the burner axis. Air and transporting gas from concentrators are released into the combustion chamber above the burners.

(2) The Fuel-Rich and Fuel-Lean Combustion Swirl Burner


(1) Dual Register(DR) burner and Enhanced Ignition-Dual Register(EI-DR) burner(see Figure 2) [9]: The EI-DR burner has axial vanes in the inner secondary air duct and tangential or axial vanes in the outer secondary air duct and the swirling direction of the inner and the outer secondary airs is identical. Pulverized coal is separated and concentrated into the wall zone of the primary air duct by the conical diffuser. Then, the coal rich zone is near the wall zone of the primary air duct and the coal lean zone is in the central zone of the primary air duct. The structures of DR and EI-DR burners are the same. The designed primary air velocity of EI-DR burner is less than that of DR burner, and The designed secondary air velocity of EI-DR burner is larger than that of DR burner. (2) Low-NOx burner with an internal coal concentrator (NSW burner, see Figure 1b) [10]: Such complex system of NSZ burners is a source of many problems, especially in boiler reconstruction. Usually there is no room available for fitting release ducts from concentrators, while many pipelines make access to the boiler difficult and complicate maintenance and repairs. It then proved necessary to design a new burner with an internal coal concentrator. Enriched coal-air mixing is directed towards the axis of flow in the form of several slightly swirled streams, while lean mixing is directed to the secondary air stream. When burner operates properly, ignition takes place in the zone of the finest released particles and thus a shield is formed, protecting the main combustion zone from the secondary air inflow. Secondary air is supplied via two coaxial ducts outside the main fuel nozzle, and fuel is mixed with the secondary air outside the rich combustion zone. Secondary air flow in both these ducts is controlled. (3) Dense dilute dual-channel burner (see Figure 3) [11]: Along the radial direction, from outside to inside, they are outer secondary air, inner secondary air, primary air in sequence. The inner and outer secondary airs become swirling after they pass through axial-register vanes. The swirling intensity of the two air streams can be regulated by pulling or pushing the axial-register vanes. The primary air/coal mixing becomes swirling through bent vanes in primary air duct. After the swirling coal-air mixing entering the four axial arrangement channels, the swirling of air is restrained and swirls weekly. Because of coal inertia, the pulverized coal gathered in near pulverized coal collector region and four fuel-rich and fuel-lean air/coal mixings are formed. Then fuel-bias combustion along the circle direction is gotten.

10

Zhengqi Li

Figure 2. EI-DR burner and the position of the monitor pipe (the dimensions are in mm): (1) particle deflector, (2) primary air duct, (3) inner secondary air duct, (4) outer secondary air duct, (5) watercooled wall, (6) axial vanes (the angle is 60), (7) tangential vanes (the angle is 25), (8) conical diffuser, (9) monitor pipe.

Figure 3. Dense dilute dual-channel burner: (1) cone, (2) axial vanes, (3) flame stabilization ring, (4) axial vanes, (5) pulverized fuel feeder, (6) throttle valve, (7) bent vanes, (8) central pipe, (9) primary air duct, (10) pulverized coal distributor, (11) pulverized coal collector, (12) inner secondary air duct, (13) outer secondary air duct.

1.4. The Prospect of Swirl Coal Combustion Technology


In early stage, it was thought that we should use large swirl number and central recirculation zone to have the coal flame stable. We ignored the factor of coal concentration.

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners

11

So, the general and air-staged combustion burners adopt various measures to increase the swirl number and central recirculation zone. Because the fuel concentration of primary air/coal mixing is little, especially with swirling primary air, most of the pulverized coal is threw off to the low gas temperature region near the secondary air under the effect of centrifugal force. There is a little coal near the high gas temperature central recirculation zone. There is no zone where both the fuel concentration and the gas temperature are high. The quality of coal provided to power plants often fluctuates and is usually low-grade. It is very difficult to have flame stable, especially when the load is low. Xu et al. [12] has experimentally studied the gas flow fields, temperature distribution and gas composition distribution in pulverized coal precombustion chamber, numerically calculated particle trajectories in combustion processes for different particle diameters and presented a new flame stabilization theory - three high zone principle. This principle says that if flow carrying over pulverized coal particles forms a high concentrated coal particle zone that is also a high-temperature and appropriate oxygen zone (called three-high zone ) at a local region near the burner mouth, this zone can stabilize pulverized coal flame. Three-high zone theory makes us have a good understand of the effect of high fuel concentration. It can give us the reason why the flame stability is bad when both the swirl number and central recirculation zone are large. Air-staged combustion burner can decrease the formation of NOx effectively. The outer secondary air segregates the water-cooled wall from reducing atmosphere in the burner center zone. It can prevent furnace wall and water-cooled wall from high temperature corrosion and slagging propensities, but the carbon-in-ash increases to some extent. High fuel concentration has good flame stability, especially at low load. Since the fuel-lean air/coal mixing carries off some air, the carbon-in-ash increases to some extent. Fuel-rich and fuel-lean combustion technology combines high fuel concentration technology and air staged combustion technology together. Meanwhile, with the fuel-rich and fuel-lean mixings introduced to furnace stratified, the burner intensifies further staged combustion. Some fuel-rich and fuellean combustion technologies have high combustion efficiency, flame stability, no slagging propensities and high temperature corrosion and low NOx emission. It is the prospect of the swirl combustion technology.

2. RADIAL BIASED COMBUSTION SWIRL COAL BURNER


2.1. Concept of Radial Biased Combustion Swirl Coal Burner
Qin proposed the radial biased combustion swirl coal burner in 1993 [13]. A fuel concentrator is installed inside the fuel-conveying duct which radially separates the primary air/coal mixing into two jets of different fuel concentrations (Figure 4). The fuel-rich primary air/coal mixing is in the inner annulus and the fuel-lean one is in the outer annulus. Surrounding them is a secondary air jet, which is also divided into two annular parts the inner swirling secondary air jet and the outer non-swirling secondary air jet. The swirler is made of axial vanes.

12

Zhengqi Li

Figure 4. The RBC burner: (1) wall, (2) non-swirl, (outer) secondary air duct, (3) Swirler, (4) swirl (inner) secondary air duct, (5) primary air duct, (6) central pipe, (7) flame igniter, (8) damper, (9) enricher, (10) fuel-lean primary air-coal mixture duct, (11) fuel-rich primary air-coal mixture duct.

2.2. Experimental Research on Pulverized-Coal Concentrator with Cone Vanes


2.2.1. Structure And Parameters of Pulverized-Coal Concentrator with Cone Vanes Figure 5 shows the structure of pulverized-coal concentrator with cone vanes[14]. Under the impact of cone vanes installed inside the primary air duct, the majority of the pulverized coal carried by the primary air is concentrated in the central zone of the primary air duct, with a fraction of coal in the peripheral zone of the primary air. Passing the cone vane, the fuelrich primary air/coal mixing is formed behind the concentrator and the peripheral primary air with a fraction of pulverized coal flows over the cone vanes. The fuel-lean primary air/coal mixing is formed outside the fuel-rich primary air/coal mixing. In Figure 5, R2 is the radius of the primary duct; R1 is the radius of the central pipe; and R is the radius of the outlet of the end vane. The principal parameters of the concentrator include a few of vanes n, the angle of vane , the setting angle of vane , the vane length L, the axial projection length of vane L1, the space between vanes L2, the coverage ratio of vanes , the blockage radio and the area ratio of the fuel-rich primary air, etc. The coverage ratio of vanes was calculated from the following equation:

= H1 H2

(1)

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners

13

where H1 is the radial projection length of overlap between two vanes, H2 is the radial projection length of vanes (see Figure 5), with m per unit. The blockage radio was calculated using the following equation:

= 1 f0
f

(2)

where f0 is the area of the end vane outlet (see Figure 5), f is the area of the primary air duct, with m2 per unit.

Figure 5. The enricher with cone vanes.

The area fraction of the fuel-rich primary air/coal mixing was calculated from the equation of

fr , where fr (m2 per unit) is the area of outlet of the fuel-rich primary air. f

The performance parameters of concentrator includethe air ratio Ra , the coal concentration ratio Rrl, the enriching ratio Rr, the resistance coefficient . The air ratio Ra is the ratio of the air feed in the fuel-rich primary air/coal mixing to the air fed in the fuel-lean primary air/coal mixing. The coal concentration ratio Rrl is the ratio of the coal concentration of the fuel-rich primary air/coal mixing to that of the fuel-lean primary air/coal mixing. The enriching ratio Rr is the ratio of the coal concentration of the fuel-rich primary air/coal mixing to that of the primary air/coal mixing. The resistance coefficient was calculated using the following equation:

= P uin 2
2g

(3)

14

Zhengqi Li

where P (Pa per unit) is the static pressure difference between the inlet and outlet of concentrator, (kg/m3 per unit) is the density of the primary air, and uin (m/s) is the inlet velocity of the primary air.

2.2.2. Impact and Rebound Phenomena of Particles on Wall of the Cone Vane Figure 6 shows the impact and rebound phenomenon of particles on wall of the cone vane. The velocity of a rebounding particle is determined using the restitution coefficients,

et (

V2t V ) and en ( 2 n ), that is measured by experiments. The rebound particle velocity V1t V1n

components, V2t and V2 n are then calculated by the following manner:

V2t = etV1t V2 n = etV1n

(5) (6)

Figure 6. Impact and rebound phenomena of particles on wall.

It should be noted here that the restitution ratios, which are determined experimentally, vary according to the flow velocity, and the combination of particles and target materials. The vane material generally is wear-resistant cast steel or ceramic so that the service life of concentrator is more than an overhaul period (always be four years). The performances of pulverized-coal impacting on wear-resistant cast steel or ceramic are being investigated by us. Results obtained by other researchers can be taken as reference. The expressions for the restitution ratios for sand particles impacting on the 410 stainless steel from Grant and Tabakoff [15] are as following:

et = en =

V2t = 1.0 2.12 1 + 3.077512 1.113 V1t V2 n = 1.0 0.4159 1 + 0.4994 12 0.29213 V1n

(7) (8)

The expressions for the restitution ratios for 157-177 m silica sand particles impacting on the target materials: 2024 AL, Ti 6-4 and INCO718 from Wakeman and tabakoff [16] are as following:

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners

15

et = en =

V2t = 0.953 0.000446 1 + 0.0000064813 V1t V2 n = 1.0 0.02111 + 0.0022812 0.00000087613 V1n

(9) (10)

2.2.3. Structure Optimization for Cone Vane of Pulverized-Coal Concentrator Experiments were carried out on a gas-particle two phase test facility to optimize the structure of pulverized-coal concentrator. The full industrial-scale concentrator studied in the experiments was designed for a 670-tph coal-fired boiler. A scale ratio of 1:3 was employed. Coal ashes from a power plant were used to be the particles. The ratio of the model size to the actual burner size is 1: 3. For the concentrator model, the value of 2 time of R2 is 165 mm, the value of 2 time of R is 65mm, n is 3. The particle concentration is from 0.26 to 0.64kg (coal)/kg(air). Table 1 shows the experiment results. The results indicate that with three vanes, the air ratio is up to 1; the coal concentration ratio is up to 2; the enriching ratio is up to 1.4; and the resistance coefficient is less. The industrial experiments of the concentrator show that with this coal concentration ratio the burner can keep a stable flame.
Table 1. Concentrator model structure parameters and experimental results [14]
case L1(mm) L2(mm) () () R(mm) Ra Rrl Rr 1 20 50 15 15 53.8 1.064 2.56 1.42 2.31 2 20 55 13.6 15 54 1.112 2.49 1.40 2.36 3 20 45 16.7 15 53.7 0.992 2.29 1.39 1.97 4 20 40 10 10 66.7 1.08 1.74 1.26 2.34 5 20 45 20 25 45.6 0.871 1.70 1.28 2.35 6 20 45 25 10 42 0.96 1.77 1.29 2.01

2.2.4. Improved Pulverized-Coal Concentrator with Cone Vanes In order to reduce NOx emissions greatly, the coal concentration ratio should be large, which need to be installed more vanes. With increasing of the number of vanes, resistance of concentrator increases, which lead to pulverized coal plugging in the concentrator and difficulty to set vanes. Structural improvement has been made to separate the end vane and isolation ring between the fuel-rich primary air/coal mixing and fuel-lean primary air/coal mixing. When the fuel-rich primary air/coal mixing passes through the zone between the end vane and isolation ring, the air in it can diffuse into the fuel-lean primary air/coal mixing along radial direction and the pulverized coal still gathers in the fuel-rich primary air/coal mixing under inertia effect. Thus, the improved concentrator achieves a larger concentration ratio with less resistance. Figure 7 shows the improved pulverized-coal concentrator with cone vanes, where x is the distance from the end vane to the isolation ring, H is the length difference between R2 and R1, and Rr is the radius of the inlet of isolation ring.

16

Zhengqi Li

Figure 7. The improved enricher with cone vanes.

2.2.5. Structure Optimization and Performance of the Improved Pulverized-Coal Concentrator Table 2 shows the experimental parameters.
Table 2. Structure parameters of the improved coal concentrator model [14]

case L1 ( mm )

1 50 20 3.0 0.39

2 50 25 3.0 0.48

3 50 35 3.0 0.65

4 50 40 3.0 0.73

5 50 30 3.0 0.56

6 55 30 3.0 0.62

7 60 30 3.0 0.68

8 65 30 3.0 0.73

9 50 30 2.0 0.56

10 50 30 2.5 0.56

11 50 30 3.5 0.56

()
L2 L1

1. Resistance Characteristics (1) Influence of Dip Angle of Vane on Resistance The influence of dip angle of vane on resistance was determined by experiments in cases 1-4. The expression for resistance coefficient was as following:

= 0.53 + 0.57

20
(11)

The expression indicates that with the increasing of dip angle of vane the resistance of concentrator and the angle of air flowing over vanes to the fuel-lean primary air/coal mixing increases and the value of

f0 f decreases. With the value of 0 decreasing, the air velocity f f

increases. It is the main reason of resistance increasing.

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners

17

(2) Influence of Vane Length on Resistance The influence of vane length L on resistance was investigated by experiment in cases 5-8. The expression for resistance coefficient was as following:

= 0.70 + 1.74

l 50

(12)

The expression indicates that with the vane length increasing the resistance of concentrator increases. With length of vane increasing, the radial projection length of the vanes increases and the flow area of the fuel-rich primary air/coal mixing decreases.

(3) Influence of the Distance between Vanes on Resistance The influence of the distance between vanes L2 on resistance was determined by experiment in cases 9-11. Figure 8a shows the relation between resistance coefficient and
normalized distance

L2 . Resistance coefficient decreases and then increases with the L1

normalized distance increasing. With the normalized distance increasing, the air/coal flow area of the fuel-lean primary air/coal mixing increases and the angle of air flowing to the fuellean primary air/coal mixing. It results in the resistance coefficient decreases. With the normalized distance increasing further, eddies behind the vanes becomes large. It results in the resistance increases. The experiment indicated the minimum resistance coefficient is obtained when normalized distance

L2 is 2.5. L1

Figure 8. Influence of normalized distance between vanes (a) and blockage ratio (b) on resistance coefficient [11].

(4) Influence of Blockage Radio on Resistance The blockage radio is a parameter which synthetically shows the influence of the angle of vane, the vane length, the number of vanes and air flow areas of the fuel-rich and fuel-lean

18

Zhengqi Li

primary air/coal mixings. Figure 8b shows the relation between resistance coefficient and blockage radio . The experimental result indicates that the blockage radio could express the variation of resistance well. In the cases that the blockage radio is less than 0.6, the resistance coefficient increases slowly while the blockage radio increases. In the cases that the blockage radio is more than 0.6, the resistance coefficient increases remarkably while the blockage radio increases. Thus, considering the reduction of resistance of the concentrator, it is recommended that the blockage radio for concentrator design should be less than 0.6. The experiment results indicate that the resistance of the improved coal concentrator reduces by 50% compared with the former concentrator. The influence of distance between the end vane and the isolation ring on distribution of air flow. In case 6, the distribution of air axial velocities at the inlet of the isolation ring with different x (see Figure 7) were measured. Figure 9 shows the results of measurement, where

U m is the axial velocity of air flow in the primary air duct.

Figure 9. Influence of distance from the end blade to the isolation ring on distribution of air axial velocities at the inlet of the isolation ring [11].

In the radial direction, the air axial velocities of the fuel-rich primary air/coal mixing is larger than that of the fuel-lean primary air/coal mixing. The distribution of air axial velocities becomes flat while x increases. At x=0.23H, partial region is in the eddy current zone.

fr X Figure 10 shows the influence of normalized area f and normalized distance H on


the air ratio Ra. For radial biased combustion burner, it is recommended that the air ratio Ra should be in the range from 0.75 to 1. The influence of distance between the end vane and the isolation ring on radial separation performance.

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners

19

Figure 10. Influence of fr/f and X/H on the air ratio [11].

Figure 11. Influence of fr/f and X/H on the coal concentration ratio [11].

fr X Figure 11 shows the influence of normalized area f and normalized distance H on X the coal concentration ratio. With H from 0 to 0.5, the minimum coal concentration ratio is X obtained. The coal concentration ratio increases and then decreases with H increasing. The

20

Zhengqi Li

X maximum coal concentration ratio is achieved while H is 1. The phenomenon is caused by


the inertia effect of the pulverized coal. In the region near the outlet of the end vane, the pulverized coal still move into the zone of the fuel-rich primary air/coal mixing under the

X inertia effect. As H is larger than 1, the coal concentration in the zone of the fuel-rich
primary air/coal mixing reduces because the pulverized particles diffuses into fuel-lean primary air zone in the radial direction.

fr X We should choose appropriate values of H and f to achieve the appropriate air ratio fr X and coal concentration ratio. When H is 1 and f is in the range from 0.3 to 0.4, the air
ratio is in the range from 0.7 to 1 and the coal concentration ratio is in the range from 5 to 8.

2.3. Effect of Structure Parameters on Gas/Particle Flow Near the Burner Region
2.3.1. Effect of Angle of Secondary Air Swirling Vane on Characteristics of Isothermal Flow Issuing from the Radial Biased Combustion Burner A number of axial fixed swirling vanes are installed in the secondary air duct of radial biased combustion burner. The vanes were curved according to a special profile line and had an angle of with burner axis. After passing through the vanes, the secondary air rotates and develops into a swirling jet at a certain swirling momentum at the secondary air outlet. For adjusting the swirl number of the secondary air near the burner outlet, the secondary air is divided into two flows, a non-swirling outer secondary air flow and an inner swirling secondary air flow. The swirl number can be adjusted by controlling the ratio of inner secondary air to outer secondary air. Some relations between vane profile and design parameters have been presented in document [17]. Some studies have shown that to keep the length of the straight section at the vane outlet and the bending radius of the vanes at appropriate values, the non-dimensional height of the vanes should range from 1.5 to 1.9. On the condition that the number, angle and cover ratio of vanes remain constant, an optimized vane profile can be obtained by choosing an appropriate non-dimensional vane height. And a reasonable design of the shape of the vane can make the secondary air flow out into the furnace at the angle that is the same as the angles of the vanes. The angle of vanes influences deeply on the characteristics of swirling jet in the near burner region. A cold flow experiments in a small-scale burner were carried out to investigate the effect of the angle of the vane (the experimental angles are 55, 58, 60 and 65) on the characteristics of the flow issuing from the burner [18]. The original burner is used on a coal-fired 670-tph utility boiler. The ratio of the small model to the original burner is 1: 3. An IFA300 constant-temperature anemometer system with a single-sensor hot-film probe was used to measure the mean velocities and turbulent parameters of swirling jet issuing from the burner by method of

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners

21

rotating the probe. Ribbons tied to a coordinate-frame were used to measure the airflow direction in the flow field. Table 3 shows the experimental parameters. Table 3. Experimental parameters for different swirling vane angles
Primary air ratio, % 19.5 Secondary air ratio, % 80.5 Inner Swirl secondary air ratio, % 85 Velocity of the primary air, m/s 8.5 Axial velocity of the secondary air, m/s Reynolds number at the burner outlet 1.6105

12.0

1. Swirl Numbers of Airflow Issuing from the Burner at Different Swirling Vane Angles Assuming that the airflow passing through the vanes is idealized and non-viscous, the angle between the airflow direction and the burner axis is . Thus the related expression for
the tangential velocity w and the axial velocity u was w = u tan . Ignoring the influence of the static pressure on the axial momentum, the swirl number of secondary air S can be calculated as follow:
2 2 u tan r dr 0 R

S=

u rd dr
2 0 0

= tan

u rd dr
2 0 0

0 R

u r dr
2 2

= tan f ( u, R, d 0 )

(13)

where R is the divergent radius of airflow at the burner outlet and r is the radius of the measurement point. In the swirling jet, the secondary air takes most of the part, thus the swirl number of swirling jet issuing from burner is determined by the tangent of swirling vane angle. As shown in Equation (12), the swirl number rises with the increase of . Comparison between the experimental data of the swirl number and tangent of swirling vane angle is shown in Figure 12, and the value of f ( u , R, d 0 ) is approximately set to 1.

2. Influence of the Swirling Vane Angle on the Flow Resistance to the Secondary Air Because the ratio of the non-swirling outer secondary air is relatively small (no more than 25%), the main resistance to the secondary air occurs when the swirling secondary air passes through the swirling vanes. The resistance to the inner secondary air passing through swirling vanes includes frictional resistance and local pressure loss. The frictional resistance h f 1 was calculated from the following equation:

h f 1 = k1

u 2
2g

(14)

22

Zhengqi Li

where k1 is the frictional resistance coefficient which is determined by surface roughness, length and air humidity of the vanes,

( kg / m3 ) is air density.

Figure 12. Effect of

on swirl number and resistance.

When the airflow passes through the swirling vanes whose angle is increases from u to

, the velocity

u , which leads to a local pressure loss. The local pressure loss h f 2 cos

can be calculated as follow:

h f 2 = k2

u 2 tan 2 2 = u k 2 2 g cos 2 2g

u2

(15)

where k2 is the coefficient. It stays constant in a certain vane structure. Then, the resistance to the inner secondary air passing through swirling vanes can be calculated as follows:

hf = hf 1 + hf 2 =
where

u 2
2g

( k1 + k2 tan 2 ) =

u 2
2g

(16)

is the resistance coefficient which can be calculated from the expression

= k1 + k2 tan 2 . When the swirling vane angle is 45, k1 is relatively small and can be ignored, then is positively proportional to the square of the tangent of . Assuming that the value of k2 is 1, the relation between and should follow the curve shown in Figure 12. The resistance coefficient is small when is 50. When ranges

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners from 50 to 65; it begins to increase. When

23

is above 70, the resistance coefficient rises sharply with increase of . So, the increase of not only enlarges the length of central
recirculation zone, but also increases the resistance to the secondary air simultaneously. It is suggested that should not be over 70 in the design of swirling vanes.

3. Influences of Swirling Vane Angle on the Flow Field Structure and the Mean Velocity Distribution of the Swirling Jet Issuing from the Burner Table 4 shows the length LCR 2 and maximum diameter dCR 2 of the central recirculation
zone of the swirling jet issuing from the burner. d is the diameter of the none-swirling secondary air cone. When

increases from 55 to 65, dCR 2 increases by 20% and LCR 2

increases by 30%. The ratio of the maximum mass flux of the recirculation air in the central recirculation zone to the mass flux of primary air (defined as the maximum normalized recirculation ratio) is shown in Table 4. The maximum normalized recirculation ratio can basically shows the mass flux of the recirculation flow in the central recirculation zone because the mass flux of primary air has no remain constant. While is 60, the maximum normalized recirculation ratio is up to 2, which shows that the recirculation gas can supply enough heat for the ignition of fuel-rich primary air/coal flow and satisfy the flame stable combustion for low-grade coal. Table 4. Influence of swirling vane angle on the length of central recirculation zone and maximum normalized recirculation ratio

LCR 2 / d dCR 2 / d
Maximum normalized recirculation ratio

55 1.49 0.8 1.160

58 2.0 0.86 1.261

60 2.0 1.0 1.909

65 2.0 1.0 1.931

The distribution of the mean axial and the tangential velocities at different

(55 or

60) of the swirling jet are shown in Figure 13, where x is the distance from the burner outlet to the velocity point, U0 is the mean velocity of the airflow at the outlet of the burner. As shown in Figure 13a, near the burner outlet ( x / d = 0.0 ), the axial velocities have two peak values along the radial direction. The higher one is the velocity peak of the secondary air; the lower one is the velocity peak of the primary air. In the central pipe of the burner, theres no air fed into the furnace, so the axial velocity is low near the outlet of the central core, which helps to form the front stationary point of central recirculation zone. The central core and the cone between the primary air and swirling secondary air (see Figure 4) have guide effect on the primary and secondary air. Also, large turbulent mass exchange occurs because of the great velocity gradient between primary air and secondary air. Thus, the primary air and the secondary air mix rapidly and the velocity peak of primary air disappears at the crosssection x / d = 0.25 , which shows that the primary air has well mixed with the secondary air. At a large swirling vane angle, the velocity peak of the jet decays rapidly.

24

Zhengqi Li

Figure13. Profiles of axial mean velocities (a) and tangential mean velocities (b) in the jet with different swirling vane angles.

When

is set to 55 , the swirl effect of the jet and the centrifugal force of fluid

micelles are weakened. Most of the air is concentrated in the central zone of the jet and the flare angle of the jet reduces. Therefore, the length of central recirculation zone became short. As shown in Figure13b, at different angles of (55 or 60), only one tangential velocity peak appears. The tangential velocity is low in the central recirculation zone and in the flow field near the primary air outlet because the primary air is non-swirling and the airflow in the central recirculation zone is mostly the axial backflow. Influenced by the swirling vanes, the maximum tangential velocity appears in the airflow field near the secondary air outlet. As the swirling vane angle increases, the maximum tangential velocity and the swirl number rise, and the tangential velocity decays rapidly. It shows that the primary and secondary air is mixed intensely. Furthermore, the tangential velocity decays more rapidly than the axial velocity and in highly swirling flow the tangential rotation disappears soon because of the large turbulent mixing. At the cross-sections x / d 1.0 , the tangential velocity of the jet falls down to zero, and the flow becomes wholly axial. The differences between the two radial velocities with different s are very few.

4. Turbulent Stress Distribution in Swirling Flow Field at Different Vane Angles Figure 14 shows the spatial distributions of normalized turbulence fluctuating root-meansquare (RMS) velocities and turbulence normal stresses at different vane angles. Figure 14a shows that the turbulent fluctuating RMS velocities are low in the region of central

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners

25

recirculation zone and jet boundary. It reaches a peak value in the interface of recirculation zone and the mainstream zone.

Figure 14. Profiles of normalized root - mean - square velocities and normalized turbulent stresses with different swirling vane angles.

26

Zhengqi Li

As the vane angle increases, the peak value of the turbulent fluctuating velocity moves outward along radial direction and the turbulent fluctuating velocitoes at the outlet increas but decay rapidly. At x / d 0.5 , the turbulent fluctuating velocities at a larger vane angle are already less than that at a smaller vane angle. Distribution of turbulence normal stress is nonhomogeneous. It is low in the region of central recirculation zone and the jet boundary and reaches a radial maximum value in the region near the boundary of recirculation zone and the region of the mainstream zone of secondary air.. However, , the turbulent normal stress in the region of swirl burner outlet is not the maximum in the whole flow field. As the flow expands downstream, the turbulent energy produces continually. At x / d = 0.25 0.5 , the turbulent normal stress reaches a maximum value and gradually declines afterwards. Thus, this region with intense turbulent fluctuation is advantageous for pulverized-coal combustion. In practical operation, the radial biased combustion swirl burner increases the pulverized-coal concentration in this region, which improves the ignition of air/coal mixing. It also improves the flame stability and combustion intensity. Turbulent normal stress is obviously anisotropic. The maximum values of axial normal stress u and tangential normal stress w are high, which indicates that the turbulent flow fluctuates more intensely in the two directions. Meanwhile, the normal stress v in the radial direction is smaller than stresses those in the other two directions near the burner outlet, and gradually increases downstream. As the swirling vane angle increases, the turbulent fluctuation intensity of airflow obviously rises at the outlet and the turbulent mixing is enhanced. The maximum turbulent normal stresses in the three directions are raised more than two times. The higher turbulent fluctuation intensity is advantageous for the burning of the air/coal flow. The intense mixing in the early stage of the jet quickens the dissipation of turbulent energy; so the turbulent energy becomes lower in the later stage of the jet. At x / d = 1.0 , u and w at the smaller.
'2 '2 '2 '2 '2

at larger vane angle are less than those

5. Influence of Swirling Vane Angle on the Mixing Characteristics of Swirling Jet Figure 15 shows the ratio distribution of the fuel-rich primary air flow (Rrp) and the whole primary air flow (Rp) at different vane angles by the temperature tracing method. The fuel-rich primary air ratio at a certain point is the ratio of the fuel-rich primary air mass flux to the total mass flux at the same point in the flow field. The Primary air ratio can be obtained in the same way. The results show that the peak of the fuel-rich flow ratio reaches 40% near the outlet of burner. The peak locates at the cross-section r / d = 0.25 which is close to the central recirculation zone near the boundary of central recirculation zone, the ratio of the fuelrich primary air flow remains high, which is advantageous for the heat and mass transfer between the fuel-rich primary air /coal mixing and the high temperature gas in the recirculation zone. The peak ratio of the fuel-rich primary air flow does not fall until the flow reaches the cross-section x / d = 0.5 . The high ratio in the early combustion of the fuel-rich primary air flow can improve the flame stability, and lower the emission of fuel-NO by forming a reducing atmosphere. As the vane angle increases, the peak ratio of the fuel-rich primary air flow moves outwards in the radial direction and the mixing of the fuel-rich primary air flow and the gas in central recirculation zone is weakened. This is because as the swirl number of secondary air

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners

27

increases, the divergent angle of jet rises and the primary air expands outward under the driving of secondary air.

Figure 15. The ratio distribution of fuel-rich primary air Rrp a and primary air R p b at different vane angles.

Figure 15b shows the ratio distribution of the fuel-rich primary air flow which shows the mixing of the primary and secondary air. In the region near the outlet ( x / d = 0.25 ), the influence of the vane angle on the fuel-rich primary air ratio is not significant, and the maximum ratio can reach about 80%. The increase of vane angle will enhance the mixing of primary air and secondary air in the downstream and uniform the distributions of primary air ratio uniform in the whole flow field. Figure 16 shows the axial decay of the maximum ratios of the fuel-rich primary air flow, Rmrp, at different vane angles. In the region ranging from x / d = 0.25 to 1, when the vane angle becomes larger, the mixing intensity of the primary and secondary air was enhanced and the maximum ratio of the fuel-rich primary air flow declines much more rapidly. Also the distance, at which the ratio of the fuel-rich primary air flow is kept high, is shortened. At the downstream of the cross-section x / d = 1.0 , the mixing of the primary and secondary airs is already uniform. The ratio of the mixing flow flux to the entrance flux is 14. It indicates that the mixing of the airflow at the outlet of the swirl burner is intense, and the distance for air-flow to be uniformly mixed is short. The increase of the swirl number can strengthen the mixing of primary and secondary air. The gas/particle two-phase experiment with PDA shows a remarkable phenomenon that the diffusion velocity of particle-phase is slower than that of the gas phase [19]. In the region near the recirculation zone, the diffusion

28

Zhengqi Li

velocity of pulverized-coal particle obtained from the two-phase experiment is slower than that of fuel-rich primary air ratio obtained from the single phase experiment.

Figure 16. The axial decay of the maximum ratio of the primary air Rmrp at different vane angles.

As mentioned above, as the vane angle increases, the length of central recirculation zone and the divergent angle of jet increases as well. The normalized backflow ratio and turbulent fluctuation of airflow increase as well. It is advantageous for the coal flame stability. But when the vane angle increases to a certain extent, the resistance to the secondary air become greater, and the primary air and secondary air mixed much earlier. The peak value of the fuelrich primary air ratio moves outwards in the radial direction. The particle load of the fuel-rich primary air declines near the recirculation zone. It is disadvantageous for the ignition of air/coal flow. So, the vane angle should be chosen in the range between 60 and 65 .

6. Conclusion Swirl number S is proportional to the tangent of vane angle . As the vane angle increases, the swirl number becomes higher. Resistance coefficient of the secondary air is proportional to the squared value of vane angle . When the vane angle is over 70 , the quick increase of local resistance coefficient of the vane will make the resistance to the secondary air too large. When the vane angle increases, the tangential velocity of the flow at the outlet rises. The axial velocity peak moves outward in the radial direction. The divergent angle of jet and the length of central recirculation zone increase. The recirculation rate also increases to supply enough heat for the ignition of the fuel-rich primary air/coal mixing.
When the vane angle increases, the turbulent fluctuation velocities increase also, and its maximum value moves outwards in the radial direction. Distribution of turbulent normal stress is non-homogeneous and anisotropic in the airflow near the outlet. At the interface of the mainstream and the recirculation zone which ranges from x / d = 0.25 to 0.5, the

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners

29

turbulent normal stress has a maximum value. The intense turbulent fluctuation in this region is advantageous for the combustion of pulverized-coal. When the vane angle increases, the turbulent normal stresses in the three directions obviously rises. The maximum value can be 2 times as large as before. It shows that the turbulent fluctuation and mixing are strengthened. And the mixing of jet also is also strengthened with the increase of vane angle.

2.3.2. Effect of Division Cone Angles between the Fuel-Lean Primary Air/Coal Mixing and the Swirling Secondary Air on Particle-Laden Flows Near the Burners Figure 17 shows burner cones. The cone between the fuel-lean primary air/coal mixing and the swirling secondary air has the influence on the secondary air flow direction, the secondary air velocity and the mixing of primary and secondary airs. A two-dimensional particle dynamics anemometer was used to measure gas/particle flow characteristics with division cone angles of 10, 22.5, and 30 [20]. The size ratio of the burner model to the utility burner in a 670-tph coal-fired boiler was 1: 6. The particles, up to 8m, were used to measure the airflow velocity and turbulence, whilst particles of diameter in the range from 10 to 100m were to represent the particle (solid) phase flow.

Figure 17. Burner cones: 1. Core, 2. the cone between the fuel-rich and the fuel-lean primary air/coal mixtures, 3. the cone between the fuel-lean primary air/coal mixture and the swirling secondary air, 4. the cone between the swirling and non-swirling secondary air, 5. non-swirling secondary air cone.

1. Effect of Cone Angles of Air and Particle Velocities Figures 18, 19 and 20 show profiles of mean axial velocities. With the increase of the cone angle, the positions of peak values of air and particle mean axial velocities move outward along the radial direction, the peak values reduce and the jet decay is faster. It shows that the increase of the cone angle strengthens the secondary air radial diffusion. Otherwise, with the increase of the cone angle, the divergent angle of the jet increases, and then the cross-section of the jet enlarges. It results in the decrease of air and particle mean axial velocities.

30

Zhengqi Li

Figure 18. Profiles of air and particle mean axial velocities with the cone angle of 10.

Figure 19. Profiles of air and particle mean axial velocities with the cone angle of 22.5.

Figure 20. Profiles of air and particle mean axial velocities with the cone angle of 30.

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners

31

Figures 21 and 22 show the effect of cone angles ( 1 ) on the maximum diameter of central recirculation zone ( dCR 2 ) and the air divergent angle, where r is the distance between the burner center line and the boundary of the jet. The air divergent angle of jet is the air semi-velocity divergent angle which is the angle formed by two air semi-velocity boundaries. The air semi-velocity boundary is the line where the mean axial velocity is half of the maximum mean axial velocity at a cross-section. With the increase of cone angle, dCR 2 and semi-velocity divergent angle increases. The reason is as follows: the larger the cone angle is, the more swirling secondary air diffuses outward and the larger the swirling secondary air radial momentum are. Furthermore, the larger the cone angles are, the larger the area of the fuel-lean primary air is, and the less the primary air velocity is. The increase of the central recirculation zone size increases the recirculation flux of the high-temperature gas. It is advantageous to coal combustion.

Figure 21. Effect of cone angles on the maximum diameter of central recirculation zone.

Figure 22. Effect of cone angles on air semi-velocity boundary.

Figures 23, 24 and 25 show profiles of mean tangential velocities. Profiles of air and particle mean tangential velocities with different cone angles are similar. With the increase of cone angle, the positions of air and particle mean tangential velocities move outward along

32

Zhengqi Li

the radial direction. The mean tangential velocities increase slightly because the increase of area of swirling secondary air outlet.

Figure 23. Profiles of air and particle mean tangential velocities with the cone angle of 10.

Figure 24. Profiles of air and particle mean tangential velocities with the cone angle of 22.5.

Figure 25. Profiles of air and particle mean tangential velocities with the cone angle of 30.

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners

33

Figures 26, 27 and 28 show profiles of air and particle axial Root Mean Square (RMS) velocities. Figure 29, 30 and 31 show profiles of air and particle axial RSM velocities. Air and particle axial RSM velocities are similar to air and particle tangential RSM velocities. The RSM velocities are less in the central recirculation zone. They are larger near the boundary of central recirculation zone and in the region of swirling secondary air where there are larger shear layers. With the effect of airflow diffusion and turbulent transport, the air and particle axial and tangential fluctuation velocities decrease. At x / d = 0.11~0.51, the air and particle axial and tangential fluctuation velocities with

1 =30 are larger than that with

1 =10. It shows that with the cone angle increasing, the air and particle velocity gradient
between primary air and secondary air becomes larger. The increase of air and particle turbulent transport ability strengthens the air and particle momentums and mass transport between air/particle flows. It can increase coal combustion and flame propagation velocities. So, it is advantageous to the ignition and combustion of pulverized coal.

Figure 26. Profiles of air and particle axial RMS velocities with the cone angle of 10.

Figure 27. Profiles of air and particle axial RMS velocities with the cone angle of 22.5.

34

Zhengqi Li

Figure 28. Profiles of air and particle axial RMS velocities with the cone angle of 30.

Figure 29. Profiles of air and particle tangential RMS velocities with the cone angle of 10.

Figure 30. Profiles of air and particle tangential RMS velocities with the cone angle of 22.5.

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners

35

Figure 31. Profiles of air and particle tangential RMS velocities with the cone angle of 30.

2. Effect of Division Cone Angles on Particle Size Distribution The particle mean diameter (d10) is the average of diameters of particles. Figure 32 shows profiles of particle mean diameter. The distributions of particle size with different cone angles are similar. Because the primary air is non-swirling, the large particles go downstream under the inertia effect, and the little particles are easy to diffuse into the secondary air and the central recirculation zone by the driving of airflow. Therefore, in the region x / d =0.11~0.51, the particle mean diameter is little in the central recirculation zone and near the wall, but is larger outside the central recirculation zone. With the jet developing, particles gradually mix each other. The distribution of particle mean diameter is to be uniform.

Figure 32. Profiles of particle mean diameter with different division cone angles.

3. Effect of Division Cone Angles on Distribution of Particle Concentration Figures 33 and 34 show profiles of number concentrations for particle size in the range from 0.5 to 100m, where Cn is the number concentration at a given point and Cmax is the largest number concentration in the same cross section.

36

Zhengqi Li

Figure 33. Profiles of particle number concentration with different division cone angles.

Figure 34. Profiles of particle number concentration with the cone angle of 22.5.

There is a peak value of the particle number concentration at every cross-section at different cone angles. The peak value of particle number concentration with the cone angle of 10 is closer to the central recirculation zone than that with the cone angle of 30. The particle number concentration with the cone angle of 10 is less than that with the cone angle of 30 near the wall at x / d = 0.11~0.51. This is because that the cone at the angle of 10 delays the radial diffusion of primary air and makes more particles stay at the central zone. The particle number concentration is large in the central zone. With the jet developing, the particle number concentration becomes uniform radially behind x / d = 1.

4. Effect of Division Cone Angles on the Mixing Characteristics of Primary Air and Secondary Air Measurements were performed on a single-phase test facility to investigate the mixing characteristics of primary air and secondary air [21]. Figure 35 shows profiles of the maximum ratio of fuel-rich primary air with different cone angles. When the cone angle is

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners less, for example

37

1 =31.0, the mixing of fuel-rich primary air and other airflows is delayed

and the maximum ratio of fuel-rich primary air is larger in the flow-field because the secondary air is driven outward. When the cone angle increases more, for instance 1 =37.1, the mixing of primary air and secondary air is weakened in the early stage but strengthened in the latter stage. Compared with other two cone angles, the maximum ratio of the fuel-rich primary air with the cone angle of 37.1 declines more quickly. It shows that the mixing of primary air and secondary air in the latter stage is larger. Behind x / d=1, the mixing of fuelrich primary air is over.

Figure 35. The axial profiles of the maximum ratio of fuel-rich primary air with different cone angles.

5. Conclusion
(1) With the increase of division cone angle, the secondary air is driven outward along radial direction; and the central recirculation zone becomes larger. Meanwhile, the positions of the peak value of fuel concentration move outward along radial direction, and the particle concentration reduces near the central recirculation zone. (2) With the increase of the division cone angle, air and particle axial and tangential fluctuation velocities become larger near the burner outlet. (3) With the increase of the division cone angle, the mixing of primary air and secondary air is delayed in the early stage.

2.3.3. Effect of The Length of the Division Cone between the Swirling Secondary Air and the Fuel-Lean Primary Air/Coal Mixing on Characteristics of Isothermal Airflows Near the Burner Region Experiments were carried out on a single-phase test facility to investigate the effect of the increase the length of the division cone between the swirling secondary air and the fuel-lean primary air/coal mixing on isothermal airflow issuing from the burner model [14]. The size

38

Zhengqi Li

ratio of the burner model to the utility burner in a 670-tph coal-fired boiler was 1: 3. While the other structural parameters of the burner modal is invariable, the L for experiments are 6.8mm, 13.5mm, 20.3mm and 27mm, namely the L / d are 0.018, 0.036, 0.055 and 0.073. The boundary of the air (particle) jet is defined as the edge at which the air (particle) axial mean velocity is 10% of the air (particle) maximum mean axial velocity .Up to a downstream jet distance of x / d =0.52, jet borders of the air and particle are almost a straight line [22]. In order to show the jet development, the divergent angle of the air (particle) jet is defined as the angle between the air (particle) jet borders at the section from x / d =0.1 to 0.52. Figure 36 shows the effect of the increase of the cone length on the size of central recirculation zone and the divergent angle of jet. The increase of the cone length enhances the outward diffusion of secondary air, and enlarges the size of central recirculation zone and the divergent angle of jet. While L is 27mm, namely L / d is 0.072, the jet issuing from the burner model shows instability: the close jet, which has a stable central recirculation zone, becomes an open jet, which has no central recirculation zone, with a little disturbance. Figures 37 and 38 show the effect of the increase of the cone length on the ratio and the maximum ratio of primary air. With the increase of the cone length, the maximum ratio increases at x / d =0.25-1. It shows that the mixing of primary air and secondary air weakens. At x / d >1.0, the increase of the cone length has a little influence on the mixing of primary air and secondary air. In the region of burner outlet, the large ratio of primary air can make the coal concentration of the primary air/coal mixing large. It is advantageous to the ignition of pulverized coal. And the delay of mixing of primary air and secondary air is advantageous to reducing the fuel NOx formation in the beginning of coal combustion. By consideration of the above results, L should be in the range from 0.018d to 0.055d.

Figure 36. Effect of the increase of the cone length on the size of central recirculation zone and the divergent angle of jet ( ).

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners

39

Figure 37. Effect of the increase of the cone length on the characteristics of mixing of primary air and secondary air.

Figure 38. Effect of the increase of the cone length on the maximum ratio of primary air.

40

Zhengqi Li

2.3.4. Effect of the Angle of Non-Swirling Secondary Air Cone on Characteristics of Isothermal Airflows Near the Burner Region The non-swirling secondary air cone is the outmost cone of the burner (see Figure 17). With angles of non-swirling secondary air cone ( nons ) of 0, 10 and 30, experiments were
carried out on a single-phase test facility [21]. Figure 39 shows the maximum diameter and maximum normalized air recirculation rate of the central recirculation zone. The maximum diameter increases linearly with the increase of nons . While nons changes from 0 to 30,

dCR 2 / d changes from 0.73 to 0.84, increasing by 15.6%. The maximum normalized air
recirculation rate ( qrp ) also increases with the increase of

nons . While nons changes from

0 to 30, qrp changes from 0.36 to 0.54, increasing by 47%. Figure 40 shows the air semivelocity lines of jet with different increases linearly with increase of

nons . The air semi-velocity divergent angle ( 1 2 )

nons . While nons increases from 0 to 30, the 1 2

increases by 30%. The linear expression was as following:


1 2 = 0.42 nons + 35.69

(17)

The increase of the cone angle enhances the radial guidance to the non-swirling secondary air and reduces the restriction of non-swirling secondary air against the swirling secondary air. It makes the swirling secondary air diffuse more rapidly along the radial direction under its centrifugal effect. The experiments for mixing characteristic of primary air and secondary air show the change of angles of non-swirling secondary air cone just has a large influence on the flow characteristic of secondary air, but has a little influence on the mixing of primary air and secondary air. It is suggested to increase the angle of non-swirling secondary air cone. It is advantageous to the low grade coal combustion with large central recirculation zone.

Figure 39. Effect of the angle of non-swirling secondary air cone on the maximum diameter and maximum normalized recirculation rate.

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners

41

2.3.5. Influence of Two Burner Core Geometries on Particle-Laden Flows Near Swirl Burners A threedimensional particle-dynamics anemometer (PDA) made by Dantec was used in this study. We get the total number concentration (number of particles per unit measurement volume) and particle volume fraction (the percent volume of dispersed phase to carrier phase). The overall uncertainties in measured values of the particle diameter and the particle concentration are typically 4 % and 30 %, depending on optical configuration; the mean velocity, typically 1 %. The measurable range for velocities is 500 m/s to 500 m/s, depending on optical configuration; for particle diameters, 0.5 to 1000 m, respectively. The experiments were done on a test facility [19]. Titanium dioxide powders were fed via the electromagnetic oscillating feeder into the main air duct which sent the central air, the primary air and the secondary air between the wind box and the blower, to trace the air flow. The particle mass flow rate was 0.12 kg/hr, and particle diameters were smaller than 10 m. Glass beads were fed via another feeder into the fuel-rich primary air/coal mixing duct. The particle density of the glass beads was 2500 kg/m3. The particle-size distribution obtained by the PDA is shown in Figure 41. The particles between the diameters from 0 to 140 m amounted to 98.4 %. Therefore, the particle size distribution was almost the same as that of pulverized coal. Particle density varies greatly with respect to coal types, and generally speaking is about 2200 kg/m3 for bituminous coal. The characteristics of glass beads are similar to those of pulverized bituminous coal. The principal idea in the present study is to use the phase information to distinguish between signals from seeding particles and dispersed-phase particles. A necessary condition for allowing seed measurements by the phase-Doppler method are that the seeding particles are spherical and the refractive index must be identical.

Figure 40. Air semi-velocity lines of jet with different angles of non-swirling secondary air cone.

Otherwise, the PDA may catch the particles passing through the measurement volume with low efficiency. Glass beads can meet these requirements. The spherical characteristic of

42

Zhengqi Li

the titanium dioxide particles is not the same as that of the glass beads, and the refractive index of the titanium dioxide particles is different from that of the glass beads. Because the smaller particles are lost during the experiment, it is difficult for the PDA to detect more particles, whose diameters are smaller than 10 m, to obtain the information of the gas-phase. We tried to use the titanium dioxide particles. Compared with the case without the titanium dioxide particles, the PDA caught smaller particles in the case with the titanium dioxide particles. So the titanium dioxide particles were used although they are not quite adequate for phase-Doppler measurements.

15

particle number %

10

0 0 83.79 167.58 251.37 335.16 418.95

Particle mean diameter (m)

Figure 41. Particle size distribution.

Figure 42 shows the burner model with the common core of which diameter is 63 mm. Figure 43 shows the sawtooth shaped core of which diameter is 57 mm and the height of the tooth is 10 mm. The following definition of the particle swirl number Sp and the air swirl number S are used:

Sp = S=
0

do / 2

p p Qr 2 dr d 0
1 2
do / 2

1 2

do / 2

p u p Qrdr

(18) (19)

do / 2

ur 2 dr d 0

u 2 rdr

where: do is test section furnace diameter, m; p is the glass bead material density, kg/m3; p, up are the mean tangential, axial velocities of the particles with the size distribution in the range from 0.5 to 100 m, m/s; Q is the dispersed particle volume flux with the size distribution in the rang from 0 to 100m in the measured location, m3/m2 s; and , u are the air tangential, axial velocities, m/s. Table 3 shows experimental parameters. During the experiment, the total secondary air mass flow rate and primay air parameters were kept constant. The air mass flow rate was controlled within an accuracy of 5 %. Burners with the sawtooth shaped and common cores have the same primary air exit area.

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners

43

Figure 42. The burner model (the dimensions are in mm). 1. central air 2. primay air and glass beads 3. swirling secondary air 4. non-swirling secondary air.

Figure 43. The structure of the sawtooth shaped core.

The particles (titanium dioxide and glass beads), up to 8 m, were used to measure the airflow velocity and turbulence whilst particles (glass beads) of the diameter in the range 10 to 100 m were to represent the particle (solid) phase flow. From the view point of modeling, particles whose diameters were between 0.5 and 100 m were used for particle volume flux and number concentration and volume fraction analysis. Particle volume flux was defined as particle volume crossing a unit area of the measurement volume per unit of time. Air/particle flow characteristics were measured in sections of x/d = 0.1, 0.22, 0.52, 1.02, 2.02, 3.32.

44

Zhengqi Li

1. Velocity Figure 44 shows profiles of the gas/particle mean axial velocities. With the sawtooth shaped and common cores in the cross sections from the burner jet to x/d = 1.02, there are two peak values in profiles of air/particle mean axial velocities. The peak zone near the burner center is the primay air and particle mixing flow zone, and another peak zone near the wall is the secondary air flow zone. In the x/d = 0.1 cross section, the peak value near the burner center is larger than that near the wall. With the primay air and particle mixing diffusing into the secondary air, the peak value near the burner center gradually reduces. On the other hand, the peak value near the wall gradually increases. As the jet stream developed, the secondary air diffuses into the wall zone, and the velocity peak value also gradually reduces. And its radial position also gradually moves toward the wall. The profiles of gas/particle mean axial velocities are almost independent of the cone structures. The central recirculation zone and near wall recirculation zone of the burner with the sawtooth shaped core are the about same as those of the burner with the common cone. The profiles of the gas/particle axial fluctuation velocities, the gas/particle mean tangential and fluctuation velocities are almost independent of the cone structures.

Velocity (m/s) a
-5 0 400 350 300 250 200 150 100 5 10 15 0 5 10 15 0 5 10 0 5 10 0 5 0 5

Radius (mm) b

50 0 -50 -100 -150 -200 -250 -300 -350 -400

x=17.7 mm x=39.4 mm x=92.5 mm x/d=0.1 x/d=0.52 x/d=0.22 particle (the common core) air

x=181 mm x=358 mm x=588 mm x/d=1.02 x/d=2.02 x/d=3.32 particle (the sawtooth shaped core)

Figure 44. Profiles of air (a) and particle (b) mean axial velocities with different cores.

Figures 45 and 46 show profiles of mean radial velocities and radial fluctuation velocities. Although the primary air/coal mixing was non-swirling, due to the angle of the cones, mean radial velocities were larger in the primay air and secondary air flow zone (the radius r > 28.2 mm) at the section of x/d = 0.1. They were smaller in the wall zone. There was one peak value in the profiles. With the jet development, the mean radial velocities gradually decreased. At the sections of x/d = 0.1 and 0.22, with the sawtooth shaped core the particle mean radial velocity is smaller than the air mean radial velocity in the primay air and

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners

45

secondary air main flow zone. That make lots of particle remain in the central zone of the burner.

Velocity (m/s) a
-10 0 10 20 30 -10 0 10 20 30 400 350 300 250 200 150 100 0 5 10 15 0 5 10 0 5 0 5

Radius (mm) b

50 0 -50 -100 -150 -200 -250 -300 -350 -400

x=17.7 mm x/d=0.1

x=39.4 mm x=92.5 mm x=181 mm x=358 mm x=588 mm x/d=0.52 x/d=1.02 x/d=2.02 x/d=3.32 x/d=0.22 particle (the common core) air particle (the sawtooth shaped core)

Figure 45. Profiles of air (a) and particle (b) mean radial velocities with different cores.

RMS velocity (m/s)


0 10 20 30 40 102030405060 10 20 30 40 5 10 15 20 25 5 10 15 20 25 5 10 15 20 25

400 350 300 250 200 150 100

Radius (mm) b

50 0 -50 -100 -150 -200 -250 -300 -350 -400

x=17.7 mm x=39.4 mm x=92.5 mm x/d=0.1 x/d=0.52 x/d=0.22 particle (the common core) air

x=181 mm x=358 mm x=588 mm x/d=1.02 x/d=2.02 x/d=3.32 particle (the sawtooth shaped core)

Figure 46. Profiles of air (a) and particle (b) radial fluctuation velocities with different cores.

46

Zhengqi Li

The air mean radial velocity appears positive and negative fluctuation near the zone that radius is about 100 mm. At the sections of x/d 0.52, the sawtooth shaped core has little influence on the mean radial velocities. With the sawtooth shaped core the air and particle radial fluctuation velocities are clearly larger than those with the common cone in the central recirculation zones and primay air and secondary air main flow zones at the sections of x/d = 0.1-0.52 and the zone that radius is less than 150 mm at the section of x/d = 1.02. In these zones, air/particle flow possesses larger radial turbulence transport capacity.

2. Particle Concentration Figures 47-49 show profiles of particle volume fluxes, normalized particle number concentration and normalized particle volume fraction for particle size in the range from 0.5 to 100 m with different cores, Cn is the particle number concentration at a given point and Cnmax is the largest particle number concentration in the same cross section, Cv is the particle volume fraction at a given point and Cvmax is the largest particle volume fraction in the same cross fraction. As a lot of particles spray into the test section from gaps of the sawtooth shaped core (see Figure5), in radially measured fields of cross sections of x/d = 0.1-1.02, the peak of the particle volume flux with the sawtooth shaped core is closer to the burner center than that with the common core. From the profiles of normalized particle number concentration and normalized particle volume fraction, we can see that at the sections of x/d = 0.1-1.02, the sawtooth shaped core make the central recirculation zone possess more normalized particle number concentration and normalized particle volume fraction than those with the common core.

Particle volume flux (10 m /(m s))


-1 0 1 2 3 4 5 6 -2 0 2 4 150 0 1 2 3 4 -0.5 400 350 125 300 100 250 200 75 150 50 100 50 25 0 0 -50 0.0 0.5 0.0 0.5 0.0 0.5

-4

Radius (mm)

x=17.7 mm x/d=0.1

x=39.4 mm x=92.5 mm x/d=0.52 x/d=0.22 the common core

x=181 mm x=358 mm x/d=1.02 x/d=2.02 the sawtooth shaped core

x=588 mm x/d=3.32

Figure 47. Profiles of particle volume fluxes with different cores.

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners

47

Cn/Cnmax
0.0 150 0.5 1.0 0.5 1.0 0.5 1.0 0.0 400 350 125 300 100 250 200 75 150 50 100 50 25 0 0 -50 0.5 1.0 0.5 1.0 0.5 1.0

Radius (mm)

x=17.7 mm x/d=0.1

x=39.4 mm x=92.5 mm x/d=0.52 x/d=0.22 the common core

x=181 mm x=358 mm x/d=1.02 x/d=2.02 the sawtooth shaped core

x=588 mm x/d=3.32

Figure 48. Profiles of number concentration with different cores.

Cv/Cvmax
0.0 150 0.5 1.0 0.5 1.0 0.5 1.0 0.0 400 350 125 300 100 250 200 75 150 50 100 50 25 0 0 -50 0.5 1.0 0.5 1.0

Radius (mm)

x=17.7 mm x/d=0.1

x=181 mm x=358 mm x=39.4 mm x=92.5 mm x/d=0.52 x/d=1.02 x/d=3.32 x/d=0.22 the common core the sawtooth shaped core

Figure 49. Profiles of normalized particle volume fraction with different cores.

The peak of the particle volume flux with the sawtooth shaped core is closer to the burner center than that with the common core. The more the position is close to the burner center, the higher the gas temperature becomes. The sawtooth shaped core makes the high pulverized coal concentration well match high temperature. At the same time, there are larger air radial

48

Zhengqi Li

fluctuation velocities and radial turbulence transport capacity near the border of the central recirculation zone. These factors make the pulverized coal easily heated, ignited and keep flame stable [12, 24, 25]. There is the larger pulverized coal concentration in the central recirculation zone with the sawtooth shaped cone. The central recirculation zone is a low oxygen concentration and reducing atmosphere zone. This zone can effectively control the NOx formation [50].

3. Particle Diameter Figure 50 shows profiles of particle mean diameters for particle size in the range from 0.5 to 100 m. The particle mean diameter is the arithmetic mean diameter. The profiles are almost independent of the cone structures.

Particle mean diameter (m)


20 400 350 300 250 30 40 50 30 40 50 30 40 50 40 50 60 20 30 40 50 30 40 50

Radius (mm)

200 150 100 50 0 -50

x=17.7 mm x/d=0.1

x=39.4 mm x=92.5 mm x/d=0.52 x/d=0.22 the common core

x=181 mm x=358 mm x/d=1.02 x/d=2.02 the sawtooth shaped core

x=588 mm x/d=3.32

Figure 50. Profiles of particle diameters with different cores.

4. Conclusion
(1) Profiles of air/particle mean axial velocities, mean tangential velocities, axial fluctuation velocities and tangential fluctuation velocities and the central recirculation zone are almost independent of as the sawtooth shaped and the common cores. (2) The peak of the particle volume flux of the burner with the sawtooth shaped core is closer to the burner center than that with the common core. The sawtooth shaped core make the central recirculation zone of burner possess more normalized particle number concentration and normalized particle volume fraction than the common core.

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners

49

(3) With the sawtooth shaped core the particle mean radial velocities are smaller than the air mean radial velocities and the air mean radial velocities appear clear fluctuation in the primay air and secondary air main flow zone in the near the burner region. In the central recirculation zone, the air fluctuation velocities are clearly larger than those with the common cone. (4) Profiles of particle mean diameters are almost independent of as the sawtooth shaped and the common cores.

2.3.6. Influence of Division Cone Angles between the Fuel-Rich and the Fuel-Lean Ducts on Particle-Laden Flows and Combustion Near Swirl Burners In order to prevent the loss of the enriching effect caused by mixing fuel-rich and fuellean primary air/coal mixings before they are ejected into the furnace, we usually install a division cone between the primary air/coal mixings. A three-dimensional particle-dynamics anemometer (PDA) is used to study gas/particle flow characteristics with division cone angles of 43.2 and 0 [26]. The uncertainties and the particle material used in the experiment are introduced in 2.3.5 Part. No enricher was mounted in the RBC burner model (see Figure 51), and glass beads were fed only into the fuel-rich primary air/coal mixing duct. This simulates the extreme case in which particles in the primary air are all concentrated into the fuel-rich primary air/coal mixing. Except for the division cone structure, the two burner models were the same. In Model A, the division cone angle was 43.2 , the division cone diameter was 83mm, and the distance from the division cone edge to the central cone edge was 13.9 mm. In Model B, the division cone was removed (equal to division cone angle of 0 ) and the diameter of the division annulus between the fuel-rich and the fuel-lean ducts was71mm. The distance from the division cone edge to the central cone edge was 33mm. Operational parameters for the two models were the same. The primary air velocity was 15.5m/s, and the swirl secondary air axial velocity was 21.4m/s. The non-swirl secondary air velocity and central air velocity were both 0 m/s. The fuel-rich primary air/coal mixing particle mass concentration, which is defined as the ratio of particle mass flow rate to air mass flow rate, was 0.20kg (fuel)/kg (air), and that of the fuel-lean one was 0 kg (fuel)/kg (air). 1. Gas/Particle Flow Characteristics Gas/particle flow characteristics were measured in sections of x/d=0.22, 0.37, 0.52, 1.02, 2.02, 3.32. Figure 52 shows the profiles of air/particle axial mean velocities, and Figure 53 shows the profiles of 0 ~ 100 m particle volume fraction. In Figure 54, refers to the division cone angle. u=0 refers to the points between the central recirculation zone and the main flow zone at which the air axial mean velocity is 0m/s, and Qmax refers to the peak value of particle volume flux at the point near the burner center. Profiles of gas/particle tangential and radial velocities were similar at the two cone angles. Figure 52 shows that there were two peak values in the profiles of gas/particle axial mean velocities at both division cone angles. The peak zone near the burner center was the primary air/coal mixing flow zone, and another peak zone near the wall was the secondary air flow zone. As the jet stream developed, theprimary air entered the central recirculation zone from the cross section at x/d=0.37 with a division cone angle of 43.2 (see Figure 51a).

50

Zhengqi Li

Figure 51. Burner models (dimensions are in mm). a. with division cone angle of 43.2 b. with division cone angle of 0 . 1. central air 2. fuel-richprimary air and glass beads 3. fuel-leanprimary air 4. swirl secondary air 5. non-swirl secondary air.

Figure 52. Profiles of axial mean velocities for gas (a) and particles (b) at different division cone angles. the division cone angle is 43.2 the division cone angle is 0 .

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners

51

Figure 53. Particle volume flux profiles at different division cone angles. the division cone angle is 43.2 the division cone angle is 0 .

Figure 54. Axial zero air velocity (u=0) line and Qmax points.

At x/d=1.02, theprimary air was completely incorporated in the central recirculation zone. The diameter of the central recirculation zone increased, and it continued to enlarge with the jet development. In the central zone, between the cross section of x/d=0.52 and that of x/d=1.02, axial mean velocities of the air were negative while those of particles remained positive. Therefore, numerous particles penetrated the central recirculation zone. This pattern of air/particle flow was also found in previous works [27-33]. With a division cone angle of 0 , there was no barrier from the division cone and the axial flow momentum was much greater. Therefore, the fuel-rich primary air/coal mixing mixed earlier with the fuel-lean one and, as a result, theprimary air flow zone near the center maintained a higher positive velocity and the diameter and length of the central recirculation zone stayed small (Figure 52a and Figure 54). It should be noted that, at the position of x/d=1.02 (x=181mm), the central recirculation zone completely disappeared and no particles penetrated the central recirculation zone. Figure 53 shows that profiles of particle volume fluxes were also similar at both division cone angles. Profiles of particle volume flux in cross sections from x/d=0.22 to 1.02 have two peak zones, resulting from burner structures and particle inertia. With the RBC burner, the

52

Zhengqi Li

diameter of the outer tube of the fuel-rich primary air/coal mixing duct was 71mm (see Figure 51). The central cone diameter was 63mm. When particles ejected into the test chamber from the fuel-rich primary air/coal mixing duct, particles in the outer annulus of the fuel-rich primary air/coal mixing duct were generally not influenced by the central cone and ejected directly into the test chamber. It is these particles that formed the peak zone of the particle volume flux near the jet axis. While particles in the inner zone of the fuel-rich primary air/coal mixing duct ejected into the test chamber at a certain angle to the burner central axis, due either to collision with the central cone or to the guidance of the central cone, these particles formed the other peak of the particle volume flux outside of the peak zone near the burner central axis. With jet development, the peak values were noticeable. Figure 54 shows that with a division cone angle of 43.2 , the fuel-rich and the fuel-lean primary air/coal mixings did not mix with each other before entering the test chamber through their own ducts. The position of the particle volume flux peak near the central axis is closer to the axis than that of the cone angle of 0 , and downstream from the cross section x/d=0.37(x=66mm) the particle volume flux peak zone is inside the central recirculation zone. In contrast, with a division cone angle of 0 , the fuel-rich and the fuel-lean primary air/coal mixings mix before entering the test chamber through their own ducts. Since particles in the fuel-rich primary air/coal mixing are easy to diffuse into the fuel-lean primary air/coal mixing, the position of the particle volume flux peak near the central axis is farther away from the central axis than that of the cone angle of 43.2 , and it is always in the outer part of the central recirculation zone.

2. Influence of Division Cone Angles on Combustion In boilers with swirl burners installed on walls, each swirl burner forms its own flame independently and generally does not influence the others. Using the central recirculation zone as the heat source, coal particles ignite on time and form stable flames. The size of the central recirculation zone has an important effect on ignition and combustion. With a division cone angle of 43.2 , the central recirculation zone is bigger and is expected to provide enough heat for PF ignition. The position of the particle volume flux peak near the central axis is closer to the central axis. Because the gas temperature is higher in the central recirculation zone, a zone of high temperature and high fuel concentration is formed. With the increase of PF concentration near the high temperature central recirculation zone, emissivity of the PF stream increases. Then radiative heat, absorbed from the central recirculation zone and the flame in the furnace, increases [34]. As a result, the temperature of the PF and air increases faster than usual. An experiment done by Horton et al (1977) [35] showed that increasing the PF concentration in a certain range could also increase the flame velocity. Downstream from the cross section at x/d=0.37 (x=66mm), the recirculation zone appears around the primary air flow zone, and the particle volume flux peak zone near the central axis gradually becomes incorporated in the central recirculation zone (see Figures 52 and 54). Therefore, a proper angle of division cone is advantageous for heating and ignition of PF particles. Coal quality frequently changes in Chinas power plants, and boilers are prone to flame extinction even with a rated load when burning low-grade coals. The appropriate division cone angle will curtail flame extinction and insure stable boiler operation. PF particles igniting at the proper position also provide good conditions for burnout. The early stage of coal combustion takes place in a dynamic combustion zone. When the temperature

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners

53

rises, combustion speed increases. These conditions are advantageous to the burnout of coal particles. With a division cone angle of 0 , the size of the central recirculation zone is too small to provide enough heat for PF ignition. The position of the particle volume flux peak near the axis is farther away from the central axis than that with the division cone angle of 43.2 . Furthermore, it is always in the outer region of the central recirculation zone where the temperature is lower. No zone of high temperature and high fuel concentration is formed. That is disadvantageous to flame stabilization and burnout, especially in the case of low-grade coals.

Figure 55. Zero velocity line ( ) of the burner central recirculation zone and the stream border ( ). a. with PA, without secondary air b. With bothprimary air and secondary air.

Many factors influence NOx formation, such as coal nitrogen, coal type, particle diameter, swirl number, stoichiometric ratio, primary air ratio, temperature and residence time. There are many measures to abate NOx emission. The low NOx burner is an inexpensive one. By ensuring the residence time of the coal particles in the fuel-rich zone or the reducing atmosphere zone, the burner decreases fuel NOx formation. When division cone angle is 43.2 , the central recirculation zone is large. Downstream from the cross section at x/d=0.37 (x=66mm), the particle volume flux peak zone near the axis is inside the central recirculation zone, which is a low oxygen zone where the atmosphere is reducing and can decrease the formation of the fuel NOx. When the division cone angle is 0 , the particle volume flux peak zone near the axis is farther away from the central recirculation zone. In this area, the atmosphere is oxidizing, the temperature is low, and ignition takes place further from the burner nozzles. The stronger the oxidizing atmosphere, the more advantageous it is to the formation of fuel NOx; the farther away from the burner nozzles that ignition takes place, the moresecondary air that mixes in [28-31, 33, 36-48].

3. In-Situ Industrial Experiments The 670t/h boiler of 67013.8545KT type was made in the former Soviet Union. It is fired with PF and synchronized to a 200MWe steam generator. It is a dry-ash type furnace with a division wall. The boiler burns lean coal (defined in China as a coal with 10-20% volatile, dry-ash free matter) with low volatility [Table 6].

54

Zhengqi Li Table 5. Experimental parameters


Central air Exit area (10-3 m2) Mass flow rate of particle (kg/s) Mass flow rate of air (10-2 kg/s) Particle loading Air temperature Particle swirl number Air swirl number 2.98 0.0 0.0 0.0 Swirling secondary air 11.41 0.0 36.91 0.0 16 0.92 0.26 Non-swirling secondary air 3.35 0.0 0.0 0.0

primary air

6.69 0.0124 12.08 0.1

Table 6. Design and burned coal composition


Proximate analysis, wt% (as received) Fixed Moisture Ash Volatility carbon Design coal Coal fired during test 4.38 5.60 34.36 27.46 17.0 13.20 43.81 53.74

Net heating value (kJ/kg) 19661 22600

Ultimate analysis, wt% (as received) Carbon Hydrogen Nitrogen Oxygen Design coal Coal fired during test 54.44 58.07 2.03 2.74 0.77 0.97 2.38 4.35

Sulfur 1.36 1.01

The boiler is of T type arrangement with dual furnaces, and uses a bin system with cold moistureladen exhaust air from pulverizers to convey the PF. The furnace is 17.76m in width and 8.8m in depth. Sixteen swirl burners are located on two opposite walls of the furnace, with eight burners on each side arranged vertically in two rows. The coal is low- grade and the quality changes frequently. The primary air/coal mixing temperature is 75, and the primary air ratio is 26%. Compared with the system in which PF is conveyed by hot air, the primary air/coal mixing temperature is lower, and theprimary air ratio is higher. Therefore, it is required that burners excel in flame stabilization. According to the above experimental results, burners with division cone angle of 43.2 are adopted.

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners

55

To verify the effectiveness of the division cone, we made in-situ measurements of the near-burner flow in the cold state. The uncertainty value in central recirculation zone size was 50mm. Figure 55a shows the flow structure without secondary air and with primary air only. The fuel-rich primary air velocity was 10.1m/s, and the fuel-lean primary air velocity was 10.8m/s. Figure 55b shows the flow structure with both secondary air and PA. Test velocities were determined by modeling. The fuel-rich primary air velocity was 10.1m/s, the fuel-lean primary air velocity was 10.8m/s, the swirl secondary air velocity was 12.1m/s, and the nonswirl secondary air velocity was 4.7m/s. db represents the outer diameter of the non-swirl secondary air nozzle, which was 1246mm. A central recirculation zone was formed with the guidance of the central cone and the division cone even though no swirl secondary air was ejected. The diameter and length of the central recirculation zone, and the divergence angle were 0.48db (600mm), 0.56 db(700mm) and 97 respectively [Figure 55a]. With secondary air given, the size of the central recirculation zone became much larger [Figure 55b]. The diameter of the central recirculation zone was between 1.61db (2000mm) and 1.77db(2200mm), and the length was 2.41db(3000mm). The divergence angle was 98 . Coal analysis is shown in Table 6. With a 200MWe load, the boiler operated stably, with parameters such as steam temperature, steam pressure, etc., within normal range. The PF ignited properly and combustion was successful. The carbon loss was 2.36%, and the boiler efficiency was 90.02%. NOx (O2=6%) was 762 mg/Nm3. The boiler operated stably at 100MWe during the low load test. Boiler steam parameters were in normal range. Furnace pressure fluctuation was 50Pa, which implies that combustion was stable in the furnace. The flame scanners showed a steady signal rather than an intermittent signal, and the boiler ran well. The low load test lasted four hours. After RBC burners were adopted, flame stability was greatly improved and no flame extinction occurred even with a wide variation in coal quality and boiler load.

4. Conclusion The influence of division cone angles between the fuel-rich and the fuel-lean ducts on gas/particle flow is significant. With an angle of 43.2 , the central recirculation zone is larger. The particle volume flux peak near the central axis is closer to the central axis than with an angle of 0 . Even at a certain distance from the burner, the particle volume flux peak zone is still in the central recirculation zone, and large masses of particles penetrate the central recirculation zone. Evidently, it is easy to form a high temperature and high fuel concentration zone. With a division cone angle of 0 , the central recirculation zone is smaller. The particle volume flux peak near the central axis is farther away from the central axis, and it is always in the outer part of the central recirculation zone. Therefore, it is not easy to form a high temperature and high fuel concentration zone. In-situ experiments on a 200MWe unit show that with a division cone angle of 43.2 , boiler efficiency is 90.02% when fired with lean coal, and NOx emission is 762mg/Nm3(O2=6%). The boiler operates stably with a load of 100MWe without auxiliary fuel oil.

56

Zhengqi Li

2.4. Influence of Run Parameters on Gas/Particle Flow


2.4.1. Influence of the Non-Swirling Secondary Air on Gas/Particle Flow and Coal Combustion 1. Gas/Particle Flow Characteristics Burner operational parameters have great effects on flow characteristics and burner properties [48-51]. The objective of our work is to investigate the influence of the nonswirling secondary air on gas/particle flow and coal combustion of radial biased combustion burners. To this end, the pilot and full scale experiments have been carried out. The data has value because of the support it lends to theoretical and numerical calculations [22]. A threedimensional particle-dynamics anemometer was used in this study. The uncertainties and the particle material used in the experiment are introduced in 2.3.5 Part. The non-swirling secondary air ratio is the ratio of the non-swirling secondary air mass flow rate to the total secondary airflow rate. The following definition of the total air swirl number St is used:

St

d0 / 2

( u + p p Q ) r 2 dr ( u 2 + p u p Q )rdr
(20)

d0 / 2

Figure 51a shows the radial biased combustion burner model. No enricher was mounted in it, and glass beads were fed only into the fuel-rich primary air/coal mixing duct. This simulates the extreme case in which particles in the primary air are all concentrated into the fuel-rich primary air/coal mixing. Table 7 shows experimental parameters. During the experiment, the total secondary air mass flow rate and primary air parameters were kept constant. The air mass flow rate was controlled within an accuracy of 5 %. Different nonswirling secondary air ratios were obtained by changing the non-swirling and swirling secondary air flow rates. The gas/particle flow characteristics were measured at sections of x/d = 0.1, 0.22, 0.37, 0.52, 1.02, 2.02, 3.32.

(1) The Total Air Swirl Number The total air swirl number, shown in Table 7, was calculated with the measured velocities at the section of x/d = 0.1. The total air swirl number decreased with the increase of nonswirling secondary air ratios. (2) Velocity The definition of the boundary of the air (the particle) jet and the divergent angle of the air (the particle) jet is shown in the 2.3.3 Part. Figure 56 shows jet borders with different nonswirling secondary air ratios, where B refers to the value of the jet boundary and Rns is the non-swirling secondary air ratio. Figure 56 shows that jet boundaries of the air and the particle were almost a straight line in the zone of x/d 0.52. Table 8 shows divergent angles of the air and the particle with different non-swirling secondary air ratios.

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners Table 7. Experimental parameters with different non-swirling secondary air ratio
Fuel-rich primary air 3.19 0.0124 0.2 Fuel-lean primary air 3.16 0.0 0.0 16 Mass flow rate of air ( 0.0 0.0 0.0 6.04 6.27 6.00 Swirling secondary air 11.41 0.0 0.0 Nonswirling secondary air 3.35 0.0 0.0

57

Central air Exit area (

103 m2)

2.98 0.0 0.0

Mass flow rate of particle (kg/s) Particle loading Air temperature Non-swirling secondary air ratio (%) 0 11.1 22.4

102 kg/s)
29.81 27.17 22.43 0.0 3.39 6.46

6.04 6.27 6.00

Total air swirl number 0.3 0.2 0.2

Table 8. Air and particle divergent angles


Non-swirling secondary air ratio Air divergent angle Particle divergent angle 0.0 % 11.1 % 22.4 %

100.8 109.0

45.5 88.6

44.3 68.8

Figure 56. Jet borders.

58

Zhengqi Li

Figures 57 and 58 show profiles of the decay of the gas/particle axial maximum velocities and the gas/particle mean axial velocities with non-swirling secondary air ratios. There were two peak values in the profiles of gas/particle mean axial velocities near the burner. The peak zone near the burner center was the primary air/coal mixing flow zone, and another peak zone near the wall was the secondary air flow zone. At the section of x/d = 0.1, the peak value near the burner center was larger than the peak value near the wall. However, with the primary air/coal mixing diffusing into the secondary air, the peak value near the burner center gradually reduced. On the other hand, the peak value near the wall gradually increased. With the jet development, the secondary air diffused into the wall zone, the peak value near the wall also gradually reduced and its radial position gradually moved toward the wall. The experimental results show that the gas/particle borders of the jet greatly moved to the burner centerline with the increase of non-swirling secondary air ratios. When the non-swirling secondary air ratios increase from 0 % to 22.4 %, the air divergent angles decreased from 108 to 44.3 and the particle divergent angles decreased from 109 to 68.8 . The central recirculation zone apparently decreased with increasing the non-swirling secondary air ratios. For non-swirling secondary air ratio equaled zero, the central recirculation zone diameters are 25.9~28.5 mm in the zone of x/d = 0.1~0.37, and they become larger with the development of the jet. As the jet developed, the primary air entered the central recirculation zone from the cross section at x/d = 0.37. At x/d = 1.02, the primary air was completely incorporated in the central recirculation zone. In the central zone between the cross section of x/d = 0.52 and that of x/d = 1.02, mean axial velocities of the air were negative, while those of particles remained positive. Therefore, numerous particles penetrated the central recirculation zone. The pattern of air-particle flow was also found in previous works [27-33]. In the case that the non-swirling secondary air ratio was 22.4 %, the central recirculation zone diameters were 14.2~20.0 mm at x/d = 0.1~0.37, and the central recirculation zone was closed at x/d = 0.52. In this case we observed the smallest central recirculation zone. In addition, the larger was the non-swirling secondary air ratio, the slower was the rate of air-particle diffusion towards the furnace wall.

Figure 57. Decays of air-particle maximum axial velocities.

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners

59

Figure 58. Profiles of (a) the air and (b) the particle axial mean velocities with different secondary air ratio.

Figure 59 shows profiles of axial fluctuation velocities. There were two peak values in the profiles of gas/particle root mean square (RMS) axial fluctuation velocities. The profiles became flat in the downstream region. Figure 60 shows profiles of mean radial velocities. Although the primary air/coal mixing was non-swirling, due to the angle of the cones, mean radial velocities were larger in the primary air and secondary air flow zone (the radius r > 28.2 mm) at the section of x/d = 0.1. They were smaller in the wall zone. There was one peak value in the profiles. With the jet development, the mean radial velocities gradually decreased. With the increase of the nonswirling secondary air ratio, the particle mean radial velocity and maximum mean radial velocity (Figure 60 (b)) decreased in the main flow field. In the same section, the position of particle maximum radial velocity in the case of the 0 % non-swirling secondary air ratio was closer to the wall than that in the case of 11.1 % and 22.4 % non-swirling secondary air ratios. Therefore, particles diffused faster into the wall zone when the non-swirling secondary air ratio was low. Figure 61 shows profiles of radial fluctuation velocities. At the sections of x/d = 0.1 to 1.02, the radial fluctuation velocities were larger in the central recirculation zone and the primary air and the secondary air main flow zones and smaller in the wall zone. At the sections of x/d = 2.023.32, the profiles of radial fluctuation velocities became flat as the profiles of the mean radial velocities became flat.

60

Zhengqi Li

Figure 59. Profiles of (a) the air and (b) the particle axial fluctuation velocities with different secondary air ratio.

Figure 60. Profiles of (a) the air and (b) the particle radial mean velocities with different secondary air ratio.

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners

61

Figure 61. Profiles of (a) the air and (b) the particle radial fluctuation velocities with different secondary air ratio.

Figure 62. Profiles of (a) the air and (b) the particle tangential mean velocities with different secondary air ratio.

62

Zhengqi Li

Figure 62 shows profiles of mean tangential velocities. Because there was no central air and the primary air was non-swirling, the mean tangential velocities were smaller in the burner central zone (r 55.5 mm) and larger in the secondary air at the section of x/d = 0.1. The profiles became flat in the down stream region. Without the non-swirling secondary air, the maximum mean tangential velocities were the largest among the three non-swirling secondary air ratios. The profiles were almost independent of the non-swirling secondary air ratios. Figure 63 shows profiles of tangential fluctuation velocities. At the sections of x/d = 0.1 to 0.52, there were two peak values in the central recirculation zone and the secondary air flow zone. At the section of x/d = 1.02 to 3.32, the profiles of the tangential fluctuation velocities became flat as the profiles of the mean tangential velocities became flat.

Figure 63. Profiles of (a) the air and (b) the particle tangential fluctuation velocities with different secondary air ratio.

(3) Particle Concentration Figure 64 shows profiles of particle volume fluxes for particle size in the range from 0.5 to 100 m at different non-swirling secondary air ratios. The profiles of particle volume fluxes show that they had two peaks in the sections of x/d = 0.1 to 1.02, resulting from burner structures and particle inertia. The diameter of the outer tube of the fuel-rich primary air/coal mixing duct was 71mm (see Figure 51a). The central cone diameter was 63mm. When particles ejected into the test chamber from the fuel-rich primary air/coal mixing duct, particles in the outer annulus of the fuel-rich primary air/coal mixing duct were generally not influenced by the central cone and ejected directly into the test chamber.

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners

63

Figure 64. Profiles of particle volume fluxes with different secondary air ratio.

It is these particles that formed the peak zone of the particle volume flux near the jet axis. While particles in the inner zone of the fuel-rich primary air/coal mixing duct ejected into the test chamber at a certain angle to the burner central axis, due either to collision with the central cone or to the guidance of the central cone, these particles formed the other peak of the particle volume flux outside of the peak zone near the burner central axis. At the sections of x/d = 2.02 and 3.32, the profiles became flat. The profiles of particle volume fluxes were almost independent of the non-swirling secondary air ratios. The maximum particle volume flux near the burner center, increased with increasing of the non-swirling secondary air ratio. Therefore, the concentration of particles in the burner central zone increased with the nonswirling secondary air ratio. Figure 65 shows profiles of number concentrations for particle size in the range from 0.5 to 100 m, where Cn is the number concentration at a given point and Cnmax is the largest number concentration in the same cross section. At the sections of x/d = 0.1 to 1.02, there were also two peak values in the profiles. However, at the sections of x/d = 2.02 and 3.32, the profiles became flat. The profiles of number concentrations were similar to that of particle volume fluxes. The profiles were almost independent of the non-swirling secondary air ratios.

(4) Particle Diameter Figure 16 shows profiles of particle mean diameters for particle size in the range from 0.5 to 100 m. The particle mean diameter is the arithmetic mean diameter. Profiles of the

64

Zhengqi Li

particle mean diameter had two peaks in the sections from x/d = 0.1 to 1.02, whose positions were same as the position of particle volume flux peaks.

Figure 65. Profiles of number concentrations with different secondary air ratio.

Particle in the fuel-rich primary air/coal mixing duct were separated into two particle jets, when ejected into the test chamber. The larger the particle was, the larger inertia it had. As the primary air/coal mixing was non-swirling, particles had the characteristic of keeping the nonswirling jet. The finer particles of two particle jets were liable to diffuse to the secondary air or the central recirculation zone. On the other side, the larger particles were liable to flow in their former flow direction due to their larger inertias. Then, two particle mean diameter peaks were formed. With the jet development, the larger particles were also transported into the wall zone or the central recirculation zone, and the profiles became flats. In the wall zone, particles near the burner jet came from the wall zone of the down stream through the outside recirculation. Particle diameters were larger in the wall zone of the down stream. As a result, the particle mean diameters were large in the wall zone. The profiles were almost independent of the non-swirling secondary air ratios.

2. Mixing Characteristics of Jet The mixing characteristics of jet from the burner were measured on a single-phase test facility [14]. Figures 67 and 68 show the ratio of primary air and the maximum ratio of primary air of every section with different non-swirling secondary air ratio. With the increase of non-swirling secondary air ratio, the ratio of primary air in the burner central zone of every section increase and the maximum ratio of primary air decreases more slowly.

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners

65

Figure 66. Profiles of particle mean diameters with different secondary air ratio.

Figure 67. Ratio of primary air with different secondary ratio.

66

Zhengqi Li

It shows that the mixing of primary air, gas of central recirculation zone and secondary air is greatly delayed. It is disadvantageous to the heating of primary air/coal mixing, coal ignition and flame stability. But it is advantageous to the air-staged combustion and reducing the fuel NOx formation.

Figure 68. Maximum ratio of primary air with different secondary air ratio.

3. Industrial Experiment (1) Experimental Boiler Experiments were carried out on the 410-tph and 220-tph boilers of a power plant. Both boilers were retrofitted from oil to coal - firing. Eight burners are located on the front boiler wall, and they are grouped in two rows. A common big wind box is used for the eight burners. The wind box is divided into the top and the bottom wind boxes by the division plate located between the top and the bottom burners. There are dampers at wind box entrances to balance the flow entering the top and the bottom wind boxes. There are also dampers at the burner swirling and non-swirling secondary air entrances. Figure 69 shows the 410-tph boiler furnace, platen superheater and burner index. The membrane water-cooled wall, whose tube diameter and pitch were 60 mm and 80 mm, was used. The tube diameter of the platen superheater was 42 mm. The transverse pitch and the number of transverse rows were 720 mm and 16. The longitudinal pitch and the number of longitudinal rows were 48 mm and 42. The working substance temperature was 321 in the water-cooled wall. The working substance temperature in the platen superheater was 352 at the entrance and 437 at the exit. Figure 70 shows the full industrial-size burner. The axially vanes were set at angles of 65 with the airflow direction. Table 9 shows the design parameters of the burner.

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners

67

Figure 69. The furnace, burner index, and swirling secondary air direction on the 410-tph boiler (dimensions are in mm).

Figure 70. The full industrial-size burner on a 410-tph boiler (dimensions are in mm) (1)central air (2) fuel-rich primary air/coal mixture (3) fuel-lean primary air/coal mixture (4) swirling secondary air (5) non-swirling secondary air.

68

Zhengqi Li

Table 9. Design parameters of the burner on the 410-tph boiler


Fuel-rich primary air 83.4 1.376 0.74 320 90 Fuel-lean primary air 83.4 0.459 0.25 90 320 320 Swirl number 0.0 2.206 0.48 0.36 Swirling secondary air 412.7 Nonswirling secondary air 98.8

Centr al air Exit area (

103 m2 )

102.3

Mass flow rate of coal (kg/s) Coal loading Air temperature () Non-swirling secondary air ratio (%) 0 20.0 0.0 0.0

Mass flow rate of air (kg/s) 1.865 1.865 1.865 1.865 11.030 8.824

Table 10. Design parameters of the burner on the 220-tph boiler


Fuel-rich primary air Fuel-lean primary air Swirling secondary air Nonswirling secondary air 69.2

Centr al air Exit area (

103 m2 )

86.0

45.2 0.739 0.67

45.2 0.246 0.22 95

203.3

Mass flow rate of coal (kg/s) Coal loading Air temperature () Non-swirling secondary air ratio (%) 0 20.0 0.0 0.0 300

95

300

300 Swirl number 0.0 1.215 0.39 0.31

Mass flow rate of air (kg/s) 1.106 1.106 1.106 1.106 6.074 4.859

Figure 71 shows the 220-tph boiler furnace, platen superheater and burner index. The membrane water-cooled wall, whose tube diameter and pitch were 60 mm and 80 mm, was used. The outside two tubes diameter of the platen superheater was 42 mm, and the other tubes diameter, 38 mm. The mean transverse pitch and the number of transverse rows were 766 mm and 12. The mean longitudinal pitch and the number of longitudinal rows were 49.3 mm and 38. The working substance temperature was 317 in the water-cooled wall. The

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners

69

working substance temperature in the platen superheater was 375 at the entrance and 445 at the exit. Figure 72 shows the full industrial-size burner. The axially vanes were set at angles of 65 with the airflow direction. Table 10 shows the design parameters of the burner.

Figure 71. The furnace, burner index, and swirling secondary air direction on the 220-tph boiler (dimensions are in mm).

Figure 72. The full industrial-size burner on a 220-tph boiler (dimensions are in mm) (1)central air (2) fuel-rich primary air/coal mixture (3) fuel-lean primary air/coal mixture (4) swirling secondary air (5) non-swirling secondary air.

70

Zhengqi Li

(2) Cold Flow Experimental Results Cold flow experiments were carried out on the 410-tph boiler. The similarity criteria are as follows: (1) self-modeling flows; (2) momentum ratios of the primary air coal-mixing to the secondary air maintained constant. The mass fluxes of different burners in the same wind box were kept equal. Burner swirling secondary air dampers were open, the non-swirling secondary air damper positions of the bottom burners (No. 1~4) were set to 0 %, and the nonswirling secondary air damper positions of the top burners (No. 5~8) were set to 100 %. Table 11 shows the experimental parameters. Figure 73 shows the measured flow field, where d0 is the burner diameter. The distance between two measurement traverses was 100 mm. We estimate that the uncertainties in establishing the location of the central recirculation zone border were 100 mm (0.10 d0). With the increase of the non-swirling secondary air ratio, the divergent angles decreased from 101 to 76 , and the central recirculation zone diameters decreased from 1.53 d0 to 1.21 d0.
Table 11. Cold flow experimental parameters on the 410-tph boiler
Non-swirling Secondary air damper (% open) Non swirling secondary air ratio (%) 3.7 20.3 Burner number Centra l air Fuel - rich primary air Fuellean prima ry air Swirling secondary air Nonswirling secondar y air Swirl number

0 100

1~4 5~8

Mass flow rate of air (kg/s) 0.095 1.850 1.850 0.095 1.723 1.723 Air temperature () 30 50 50

10.529 9.183 30

0.403 2.337 30

0.45 0.37

Figure 73. The jet border and the CRZ boundary on the 410-tph boiler.

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners

71

(3) Reacting Flow Experimental Results In the reacting flow experiment, the boiler operational parameters such as the load, the furnace exit O2, the coal feed rate, the central air and the primary air flow rates were kept constant. Only damper positions of the swirling and the non-swirling secondary air jets were changed. Table 12 shows the properties of the coal-fired.

Table 12. Burned coal composition


Proximate analysis, Wt % (as received) Moisture Ash 4.9 34.26 Ultimate analysis, Wt % (as received) Carbon Hydrogen 52.22 3.08 R200 PF fineness, % R90 46.09 3.0 Volatility 36.09 Oxgen 4.14 Fixed carbon 24.75 Nitrogen 0.77 Net heating value (kJ/kg) 19720 Sulfur 0.63

A thermocouple inside a water-cooled probe was used to measure the gas temperature in the burner zone. Gas temperatures of the No. 2 burner were measured along the burner axis in the 410-tph boiler. Figure 74 shows the results, where curves a and b corresponded to different positioning of the swirling and the non-swirling secondary air dampers respectively. In the 220-tph boiler, gas temperatures of the No. 4 burner were measured radially at x/d1 = 2.4 (x = 1920mm), where d1 = 798 mm was the burner diameter. Figure 75 gives the results. Figures 74 and 75 shows that in both boilers gas temperatures in the burner zone decreased with increasing the non-swirling secondary air ratio.

Figure 74. Temperature profiles on the burner axis.

72

Zhengqi Li

Figure 75. Radial temperature profiles in the section that is 2.4d (1920mm) away from the exit of burner.

In the 220-tph boiler, the swirling secondary air dampers of all the burners were set to 100 %, and the non-swirling secondary air dampers of all the burners were set in sequence to at 0 %, 50 % and 100 %. Along the axis of No. 2 burner axis, gas sample were taken using a water-cooled probe and subsequently analyzed. Measurement ranges for NO, NO2, CO and O2 of the instrument are 0-2000 ppm, 0-200 ppm, 0-4000 ppm and 0-21 %. The volume concentration of NOx is the sum of volume concentrations of NO and NO2. Resolutions of NO, NO2, CO and O2 are 1 ppm, 1 ppm, 1 ppm and 0.1 %. Figure 76 shows the measured O2, NOx and CO concentrations when the non-swirling secondary air was set to 0. The measures were taken inside the central recirculation zone. The O2 concentration was almost 21 % at the burner exit, since the flame had not been ignited and the cold air jet was given from the burner central tube. Further down zone the coal jet was heated and volatiles were given off. In the zone of x/d1 = 0.2~0.5, a large quantity of volatiles were given off and they burned rapidly. Consequently, oxygen was rapidly consumed. The volatile-N, given off during devolatilization, was oxidized to a greater, extended to NOx. The NOx formation rate rapidly increased, and it stayed relatively constant when x/d1 0.4 . When the swirling jet mixed rapidly with the coal jet, the volatiles had enough oxygen for rapid combustion. Therefore, the CO concentration was at a low level. Down jet, at the central recirculation zone closure, the O2 concentration graduately increased due to the diffusion of O2 from the secondary air jet. It assured a good char burnout. Figures 77, 78 and 79 show O2, CO and NOx profiles at the burner axis for different positions of the non-swirling secondary air damper. Pickett et al. (1999) found that the velocity profiles for reacting flow showed similar trends and patterns as were observed in cold flow experiments [48]. Our cold flow experiments show that when the non-swirling secondary air ratio was increased the mixing of the fuel jet and the primary air jet was decayed.. Thus, as the non-swirling secondary air dampers were graduately opened, the

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners

73

oxygen flowing into the jet center decreased, and the O2 concentration on the burner axis decreased (see Figure 77).

Figure 76. Profiles of in situ NOx, CO, and O2 concentrations on the burner axis.

Figure 77. Profiles of in situ O2 concentration on the burner axis.

74

Zhengqi Li

Figure 78. Profiles of in situ CO concentration on the burner axis.

Figure 79. Profiles of in situ NOx concentration on the burner axis.

Our gas/particle flow experiments show that fuel fluxes in burner central zone increased with the non-swirling secondary air ratio. In this way, the fuel-rich combustion was promoted, and the NOx reducing environment was created. The specific air-particle flow pattern inhibited the formation of the fuel-NOx. Our observations are in agreement with the previous experiments [2, 4, 31, 36-47, 52]. The NOx formation in the burner zone apparently decreased when the non-swirling secondary air ratios increased (see Figure 79).

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners

75

Table 13 shows the influence of the position of non-swirling secondary air dampers on NOx emission and char burnout of the 220-tph boiler. The economizer exit oxygen shows that the increase of non-swirling secondary air had a little effect on the boiler exit oxygen, because opening of these dampers had little effect on the total amount of air supplied to the boiler. NOx emissions (O2 = 6 %) decreased from 359 ppm to 331 ppm, with the average exit CO concentration at 18.7 ppm. Thus, the gaseous incomplete combustion loss is negligible. The carbon-in-ash decreased substantially from 2.98 % to 2.03 %. The swirl number influences the amount of oxygen and particles and the particles residence time in the burner central zone, which have effect on the particle burnout in the burner region. Experiments with a single annular orifice (SAO) burner and a single central orifice (SCO) burner in a large scale laboratory combustor [33, 53] emerges that the quality of the burnout in the burner region, are primarily influenced by the particles residence time in the burner inner recirculation zone. The maximum particle burnout for both burners occur at a certain swirl number.

Table 13. The boiler NOx emission and the carbon in ash content on the 220-tph boiler
Non-swirling secondary air damper, % open Economizer exit oxygen, % NOx (O2 = 6 %), ppm Carbon in ash content, % 0 3.5 359 2.98 50 3.5 337 1.96 100 4.0 331 2.03

Above and below this swirl number, the particle burnout starts to decrease, respectively. Increasing the swirl number creates a more compact and intense central recirculation zone within which combustion rates, excepting very high swirl values, are generally intensified. At very high swirl values, the particles will be centrifuged out of the central recirculation zone resulting in shorter residence times and a fall-out in burnout. With increase of the nonswirling secondary air ratios, the fuel fluxes increased in the burner central zone, where the gas temperature is higher than the gas temperature in the outside, resulting in a fall-out of carbon-in-ash.

4. Conclusion The radial bias swirl-stabilized burner for combustion of pulverized coal has been investigated by conducting both extensive measurement of gas and solid phase velocities, particle concentrations, particle diameters, and subsequent trials in industrial boiler. The following conclusions are drawn:
(1) Up to a down jet distance of x/d = 0.52 (d - burner diameter), the gas-phase flow pattern coincides with the solid-phase flow. Jet borders of the air and particle are almost a straight line. Further down jet, the gas jet and the solid phase jet diverge. (2) Low NOx combustion is achieved by injecting coal particles into the swirl-induced central recirculation zone that is a well-known method. However, a substantial improvement to char burnout is achieved by dividing the secondary combustion air jet into a non-swirling and a swirling part and the primary air-coal mixing into a fuelrich and a fuel-lean primary air-coal mixings. The non-swirling jet directed the particles towards the down jet part of the central recirculation zone. When 20 % of

76

Zhengqi Li the secondary air jet was provided through the non-swirling jet, the carbon in ash content decreased from 2.98 % to 2.03 %. (3) The ratio of mass flow rate of the non-swirling part to the flow rate of the total (swirling and non-swirling) secondary combustion air jet is an important burner design parameter. The dependence of the central recirculation zone properties and primary air diffusion and subsequent NOx and CO emissions on this parameter has been given in the paper.

2.4.2. Influence of the Central Air on Coal Combustion 1. Experimental Boiler The 670t/h boiler of 670-13.8-545KT type in a power plant was made in former soviet Union. It is fired with pulverized coal and synchronized to a 200 MWe steam generator. It is of dry-ash type furnace with a division wall. The boiler is of T type arrangement with dual furnaces, and it adopts bin systems with cold moisture-laden exhaust air from pulverizers conveying PF. The furnace is 17.76m in width and 8.8m in depth. Sixteen swirl burners are located on two opposed sidewalls of the furnace, with eight burners on each sidewall vertically in two rows. Volute burners were originally used. Eight burners of the bottom row were retrofitted with RBC burners before the industrial test [19]. 2. Cold Flow Experiment In cold flow experiment, co-axial airflow pattern was maintained to be in the second selfmodeling zone. Compared with reacting flow, momentum ratios were maintained same. In addition, the SA and the PA flowrates of each burner were kept to be equal, and dampers of the non-swirl SA were 100% open. Little tufts were used to trace flow near the burner. The uncertainty is 100mm (0.01db). Here, db represents outer diameter of the non-swirl SA nozzle of the burner, which is 1080mm. Figure 80 shows the flow pattern, where Rc is the central air ratio.

Figure 80. Reverse-flow and jet boundaries of a full-scale burner.

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners

77

As ratio of the central air is 1.2%, the velocity of the central air is so little that the boundary of the reverse flow maintained unchanged. As ratio of the central air is 6.2%, the part of the reverse flow zone at the burner exit in the central zone of the burner is blown away. Instead, a zone, where the axial velocity is positive, is formed like a taper cone. Little tufts also measured the jet boundary, where axial velocity is zero, and the divergent angle of the jet was got. As ratios of the central air are 1.2% and 6.2%, divergent angles of the jet are 90 and 89 . It has a little change.

3. Reacting Flow Experiment At rated load, in-situ species and temperature were measured on axis. They are shown in Figures 81 and 82. A naked thermoelectric couple protected by a water-cooled gun was used to measure the gas temperature. In-situ species were sucked out by the water-cooled gun and continuously measured by an analyzer. Resolutions of O2 and CO are 0.1% and 1ppm. CO2 is calculated continuously according to the measured O2 concentration. The coal fired is shown in Table 14. As ratios of the central air increase from 0.8% to 4.9%, velocities of the central air increase from 4.0m/s to 24.5m/s, and gas temperature on axis decreases about 263. Especially as ratio of the central air is 4.9%, the gas temperature was only 927 at x/db=0.46(x=500mm). As ratio of the central air is0.8%, with increase of distance, O2 decreases quickly, and becomes to be about 4.5%. Meanwhile, CO2 and O2 increase instantly, and CO2 becomes to be about 14.5%. This makes the zone of the reverse flow become the zone of reducing atmosphere. On the contrast, as ratio of the central air is 4.9%, O2 increases quickly, and CO and CO2 maintain relatively small. At x=800mm(x/db=0.74), O2 decreases to be 18.9%. This makes the zone of the central air become the zone of oxidizing atmosphere.

Figure 81. In-situ gas temperature along axis of burner.

78

Zhengqi Li

Figure 82. In-situ species along axis of burner.

Table 14. Fired coal composition on the 670-tph boiler


Proximate analysis, Wt % (as received) Moisture Ash Volatile matter 23.20 Oxygen 6.38 Fixed carbon 33.92 Nitrogen 0.92 Net heating value (KJ/kg) 18343 Sulfur 0.71

9.20 33.68 Ultimate analysis, Wt % (as received) Carbon Hydrogen 45.88 3.23

At 160MWe load, the furnace exit O2 kept to be constant. Then, NOx emission of boiler was measured with damper of the central air at different positions. Each case lasted about four hours. The coal fired is shown in Table 14. MSI compact analyzer made in German was used. Resolutions of NO and NO2 are 1ppm. The volume concentration of NOx is the sum of volume concentrations of NO and NO2. As position of the damper of the central air were changed from 0% to 100% open, NOx emission of boiler (Via O2=6%) increased from 216ppm to 226ppm. The increased value was only 10ppm. Effect of the central air on NOx formation can be explained as follows. With the increase of ratio of the central air, recirculated-mass flowrate decreases and gas temperature of the zone of the reverse flow decreases. It is advantageous to reduce the formation of thermal NOx. On the other side, with the increase of ratio of the central air, O2 increases and CO decreases. Thus, the oxidizing atmosphere is enhanced around the PF combustion zone. This is advantageous to the formation of fuel NOx. Therefore, the central air has a relatively small effect on NOx formation.

3. Conclusion
(1) The central air will change the type of flow pattern. When the ratio of the central air is small, the part of the reverse flow zone at the burner exit in the central zone of the

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners

79

burner is blown away. A zone, where the axial velocity is positive, is formed like a taper cone. (2) As ratio of the central air increase, gas temperature greatly decreases and the oxidizing atmosphere is enhanced in the central zone of burner. NOx emissions of boiler increase slightly.

2.4.3. Influence of Air Supply on Nox Formation and Coal Burnout 1. Effect of the Burner Primary Air Ratio on Flow And Mixing Characteristics of Jet Experiments were carried out on a single-phase test facility to investigate the effect of the burner primary air ratio on flow and mixing of jet. The size ratio of the burner model to utility burner in a 670-tph coal-fired boiler is 1: 3. During the experiments, the fuel-rich primary air flux, total air flux, ratio of swirling secondary air to non-swirling secondary air (80%:20%) are constant. The burner primary air ratio is the ratio of the burner primary air flux to the burner total air flux. It was changed by changing the proportion between primary air and secondary air in the experiments. The primary air was heated to 60 and the other airflows were at ambient temperature. The profiles of air temperature near the burner outlet were measured. The maximum ratio of primary air shows the mixing of primary air and secondary air. With the increase of the burner primary air ratio, the primary air velocity and the air axial momentum of jet increase, but the total swirl number decreases because the burner primary air of the radial biased combustion burner is non-swirling. Figures 83, 84 and 85 show experiment results, where r1 is the burner primary air ratio. With the increase of the burner primary air ratio, the secondary air ratio declines, the divergent angle of jet, the length and diameter of central recirculation zone decreases sharply. Those phenomenons also appear in the air cold experiment of the utility burner. Otherwise, with the increase of the burner primary air ratio, the ratio of primary air and the maximum ratio of primary air at the measured points increase. It shows that the mixing of primary air and secondary air decreases.

Figure 83. Effect of the burner primary air ratio on central recirculation zone and divergent angles of jet.

80

Zhengqi Li

Figure 84. Effect of the burner primary air ratio on characteristics of mixing of primary air and secondary air.

Figure 85. Effect of the burner primary air ratio on the maximum ratio of primary air.

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners

81

2. Effect of Stoichiometric Ratio on the Nox Formation and Coal Burnout (1) Experimental Boiler Experiments were carried out on a 220-tph boiler of a power plant (see 2.4.1 Section). In cold air experiments, the results of measured velocities at the burner outlet show that the secondary air fluxes of the burners from the same air box are uniform. The stoichiometric ratio of the burner is the ratio of air mass supplied by burner to the theoretical air mass which is used for coal to burn out completely. It can be calculated according to the excess air coefficient at the furnace outlet, the secondary air mass fluxes of up and bottom air boxes and the primary air mass flux, which were measured in the experiments. There is no central air during the experiments. While the primary air mass flux is constant, the secondary air supply to up and bottom burners is changed by adjusting the opening of dampers in the air boxes. The total air mass flux and stoichiometric ratio of a burner are changed respectively. In the reacting low experiment, the boiler operational parameters such as the load, the furnace exit O2, and the coal feed rate were kept constant. The boiler load was 195.5tph. The secondary air dampers of burners are 0% open. Table 15 shows the characteristics of the coal used in the experiments.
Table 15. Burned coal composition on the 220-tph boiler
Proximate analysis, Wt % (as received) Moisture Ash Volatile matter 20.43 Oxygen 3.62 Fixed carbon 39.99 Nitrogen 0.60 Net heating value (KJ/kg) 19881 Sulfur 0.50

3.20 36.38 Ultimate analysis, Wt % (as received) Carbon Hydrogen 52.55 3.16
Notes: R90=39.12%R200=5.35%.

(2) Effect of Stoichiometric Ratio on the Nox Formation of the Burner Near the burner region, gases were sampled by a water-cooled gun. Figures 86-89 show profiles of gas components on the central axis of the burner [19, 54], where db is the diameter, 798mm, and SR is the stoichiometric ratio of the burner. Cold air test in the boiler show that the central recirculation zone of jet with the length of 1.29 db and diameter of 0.65 db, begins at the central cone of the burner. In the experiments, components of gases were measured in the central recirculation zone. Figure 86 shows that with the jet developing, the Nox concentration on the central axis of the burner increases in the measured zone and rises sharply ahead of x=0.2 db where the fuel Nox is mainly formed. With SR=1.02, the NOx mean concentration is the largest and larger at the burner outlet. As SR is more or less than 1, the more or the less the SR is, the more the NOx mean concentration decreases. With SR<1, the coal combustion is the low-oxygen combustion near the burner region. Therefore, the fuel NOx formation from char-N is the minimum. The CO concentration is large in the high gas temperature recirculation zone. And the O2 concentration is less. Therefore, the oxidative atmosphere decreases and the reducing atmosphere increases in the near-burner region (see Figures 87 and 89).

82

Zhengqi Li

Figure 86. Profiles of in situ NOx concentration on No. 2 burner axis.

Figure 87. Profiles of in situ CO concentration on No. 2 burner axis.

Because the secondary air supplied to the burner decreases, the primary air ratio rises. The mixing of primary air and secondary air is weakened. The large fuel concentration of the fuel-rich primary air/coal mixture can makes the pulverized coal ignited in time. The gas temperature is large in the central recirculation zone. With the large gas temperature, the volatile can be released in time from coal. The volatile concentration is large. Thus, the NOx

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners

83

formation from volatile-N is less. The less the SR is, the less the NOx formation from volatileN and char-N is.

Figure 88. Profiles of in situ O2 concentration on No. 2 burner axis.

Figure 89. Positions of the minimum O2 concentration on No. 2 burner axis with different stoichiometric ratio.

84

Zhengqi Li

Figure 88 shows profiles of O2 concentration on the central axis of burner is a U curve. When the jet issues from the burner outlet, the fuel-rich primary air/coal mixture is not ignited. So the O2 concentration is large at the burner outlet. As the fuel-rich air/coal mixture burns, the oxygen consumes quickly. Then, the O2 concentration decreases sharply. As the fuel-lean primary air/coal mixture and secondary air diffuse into the fuel-rich primary air/coal mixture gradually, the O2 concentration increases graduately. Thus, the position of the minimum O2 concentration has a relationship with the position of coal ignition. With SR>1, the supplied secondary air is larger and the primary air ratio is less. The divergent angle of jet, the diameter and length of central recirculation zone increase. It is advantageous to the coal ignition in time. Figure 89 shows with the increase of SR, the distance from the position of the minimum O2 concentration to the burner decrease. Therefore, the position of coal ignition is closer to the burner outlet. It makes the temperature of the fuel-rich primary air/coal mixture rise fast. The rapid increase of coal temperature makes the volatile release fast. Then, the large volatile concentration is formed. Because the fuel-rich primary air/coal mixture is non-swirling and the particle inertia is large, the quantity of pulverized coal entering the central recirculation zone increases with the increase of central recirculation zone diameter. The combustion of those pulverized coal reduces the O2 concentration in the central recirculation zone and increases the CO concentration (see Figs. 87 and 89). It is advantageous to reducing the fuel-volatile NOx [28]. Furthermore, the rapid ignition of pulverized coal and the larger central recirculation zone increase the time of pulverized coal in the reducing atmosphere. It reduces the fuel NOx formation. So, the NOx mean concentration decreases in the central zone of the burner.

Figure 90. Profiles of NOx concentration of No. 4 burner with different SR at x =2.4 db (1920mm).

Figure 90 shows profiles of NOx concentration, where x is the distance from the measured points to the burner outlet along the jet flow direction. The NOx concentrations of No. 4 burner were measured by a water-cooled gun through the monitor port at the side wall near

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners

85

the burner. In Figure 90, r is the distance from sampling point to the burner axis. As SR increases from 0.79 to 1.02 or decreases from 1.66 to 1.02, the NOx concentration declines. The results are in concordance with the above analysis. In conclusion, when the stoichiometric ratio of burner is more or less than 1, the oxidative atmosphere decreases and the reducing atmosphere increases in the center of burner region so that the NOx formation reduces.

(3) Effect of Air Supply of Up and Bottom Burners on Nox Emission and Coal Burnout With different openings of dampers in the up and bottom air boxes, the O2 concentration at the outlet of economizer is at the range from 3.0% to 3.5%, the furnace gas temperature is 1291-1359 at the elevation of 6100mm and 1374-1443 at the elevation of 8300mm. It shows that the different air supply of up and bottom burners has no influence on the total air supply and normal coal combustion in the furnace. Figure 91 shows the effect on the unburned combustible loss (q4) and the NOx emission in flue gas, where SR1 is the stoichiometric ratio of up burners and SR2 is the stoichiometric ratio of bottom burners. As the SR1 and the SR2 are farther from 1, the NOx emission is less and the unburned combustible loss is larger. The unburned combustible loss with SR1>1 and SR2<1 is less than that with SR1<1 and SR2>1. The reason is as follows: the NOx formation of a burner is less as the stoichiometric ratio of a burner is farther from 1. Then, the NOx emission of boiler decreases.

Figure 91. Unburned combustible losses (q4) and NOx emissions with different stoichiometric ratio of up and bottom burners.

86

Zhengqi Li

3. Conclusion Experiments were carried out on a small-scale cold air test facility and a 220-tph bituminite-fired boiler with radial biased combustion burners in a power plant. The results are as follows:
(1) With the increase of primary air ratio, the divergent angle of jet, the length and the diameter of the central recirculation zone and the mixing of primary air and secondary air decrease. (2) With the primary air mass flux constant, the stoichiometric ratio of burner is changed by changing the secondary air supply of burner. As the stoichiometric ratio of burner is farther from 1, the oxidative atmosphere decreases and the reducing atmosphere increases in the central zone of the burner. The NOx formation of burner decreases. As the stoichiometric ratio of up and bottom burners are farther from 1, the NOx emission of the boiler decreases and the unburned combustible loss increases.

2.5. Characteristics of Gas/Particle Flow and Coal Combustion of Radial Biased Combustion and Volute Burners
2.5.1. Effect of Primary Air Flow Types on Particle Distributions in the Near Swirl Burner Region Primary air flow types of swirl burner have two kinds of swirl type and non-swirling one. Burners whose primary air is swirl are widely used, and they are main volute burners in China. A three-component particle-dynamics anemometry was used to measure the characteristics of gas/particle two phase flows with two swirl burners with different primary air flow types on a gas/particle two phase test facility [13, 19]. The uncertainties and the particle material used in the experiment are introduced in 2.3.5 Part. The radial biased combustion burner model is shown in Figure51a. The primary air swirler of volute burner is volute. The models primary air channel is long and narrow. If the volute were used, the resistance would be too large. Therefore axial bent vanes are used to make the primary air and particles swirl, and the other structure of the volute burner model (see Figure 92) is the same as that of the radial biased combustion burner. Test parameters are given by the way of the approximate modelling. The particle mass concentration is the ratio of glass bead mass flow rate to the air mass flow rate in the primary air. Two burners velocities are the same, the primary air velocity is 15.5m/s, the swirl secondary air axial velocity being 21.4m/s, the non-swirling secondary air velocity and central air velocity being all 0m/s, the fuel-rich primary air/coal mixing particle mass concentration of the radial biased combustion burner being 0.20kg (fuel)/kg (air), the fuel-lean primary air/coal mixing particle mass concentration being 0.0kg (fuel)/kg (air), and the primary air particle mass concentration of the volute burner being 0.10kg (fuel)/kg (air). 1. Velocity Swirl numbers are calculated with the measured data in the profile whose distance from the burner jet is 17.7mm. The air swirl number of the volute burner, calculated by the

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners

87

Formula 19, is 0.46, and that of the radial biased combustion burner is 0.30. The air swirl number of the volute burner is about 1.5 times as much as that of the radial biased combustion burner. This is because the primary air swirling increases the whole stream swirling. The particle swirl number of the volute burner, calculated by the Formula 18, is 3.98, which far exceeds its own air swirl number, that of the radial biased combustion burner is 0.09, it is about 44.2 times as much as that of the radial biased combustion burner. The main reason is as follows: the volute burner particles swirl with the primary air swirling, with the centrifugal stress action, particles fast go into the zone which is far distant from the burner center (see below measured results), the particle swirl momentum is larger. On the contrary, the radial biased combustion burner primary air does not swirl in the primary air channel, particle phase becomes swirl after being brought by the air near the burner zone, particles mostly distribute near the burner center zone, and the moment of the particle swirl momentum is relatively less.

Figure 92. The volute burner model (the dimensions are in mm).

Figures 93 and 94 show the profiles of the maximum mean tangential and radial velocities of the two burners. The maximum mean tangential velocities with the volute burner are relatively larger, and they are about 2 times as much as that of the radial biased combustion burner in the measured zone. At the beginning, the volute burner has much high air/particle maximum mean radial velocities; when x/d1air/particle maximum mean velocities are about 2 times as much as that of the radial biased combustion burner. When x/d1, air/particle maximum mean velocities with the radial biased combustion burner gradually reduce to 0, with the volute burner keeping about 8m/s. In 8 cross sections from x/d=0.1 to 3.32, measured mean radial velocities shows: with the volute burner, in the cross sections from the burner jet to x/d=0.37, air/particle two phase both have larger mean radial velocities in the burner center zone and the main stream zone, air/particle two phase mean radial velocities are above 2 times as much as that with the radial biased combustion burner in the main stream zone.

88

Zhengqi Li

12

velocity m/s

8 4 0

RBC burner volute burner gas particles

0.5

1.5 2 x/d

2.5

3.5

Figure 93. Declines of maximum tangential velocities with the radial bias combustion burner (RBC) and the volute burner.

30 25
RBC burner volute burner gas particles

velocity m/s

20 15 10 5 0

0.5

1.5

2 x/d

2.5

3.5

Figure 94. Declines of maximum radial velocities with the radial bias combustion (RBC) burner and the volute burner.

In the other cross sections, compared with the radial biased combustion burner, the volute burner also keeps larger mean radial velocities, and values becomes even after x/d=1.0. Figure 95 shows the profiles of air/particle mean axial velocities of the two burners. With the radial biased combustion burner in the cross sections from the burner jet to x/d=1.02, there are two crest values in profiles of mean axial velocities for air and particles, the crest zone near the burner center is the primary air and particle mixing flow zone, and another crest zone near the wall is the secondary air flow zone. In the x/d=0.1 cross section, the crest value near the burner center is larger than that near the wall, with the primary air and particle mixing diffusing into the secondary air, the crest value near the burner center gradually reduces. On the other hand, the crest value near the wall gradually increases. With the stream development, the secondary air diffuses into the wall zone, the velocity crest value also gradually reduces, and its radial position also gradually moves toward the wall. In the crest zone near the burner center, the mean axial velocity of the particle is obviously larger than that of the air. In the x/d=1.02 cross section, when the mean axial velocity of the air is negative value, that of particles remains positive value. That is to say that particles pass

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners

89

through the central recirculation zone. With the volute burner, the crest zone near the center zone is very narrow in the radial direction, and in the x/d=1.02 cross section it already disappears. This shows that the primary air and the secondary air mix fast, and the particles do not pass through the central recirculation zone.
velocity m/s
-10 -5 0 5 10 15 20 -5 0 -400 350 300 5 10 15 20 -5 0 5 10 15 20 -5 0 5 10 -5 0 5 10 0 3 0 3

diameter mm

250 200 150 100 50 0 -50

x = 17.7 mm x / d = 0.1

x = 39.4 mm x = 66 mm x / d = 0.22 x / d = 0.37

x = 92.5 mm x = 181 mm x / d = 0.52 x / d = 1.02

x=358 mm x/d=2.02

x = 588 mm x / d = 3.32

Figure 95. Profiles of mean axial velocities for air and particles with radial bias combustion burner ( gas, particles) and volute burner (gas, particles).

2. Particle Volume Fluxes and Concentrations Figures 96, 97 and 98 show the profiles of particle volume flux, particle volume fraction and particle number concentration of the size distribution in the range from 0 to 100m in the different cross sections with two burners. With the radial biased combustion burner, in radially measured fields of cross sections from x/d=0.1 to 1.02, profiles of particle volume flux, particle volume fraction and particle number concentration have two crest zones, resulting from burner structures and the particle inertia. With the radial biased combustion burner, the fuel-rich primary air/coal mixing channel outer tube diameter is 71mm (see Figure 51a). The central diffuser cone diameter is63mm, when particles spray into the test section from the fuel-rich primary air/coal mixing channel, particles in the outer zone of the fuel-rich channel are not influenced by the central diffuser cone and directly spray into the test section. Then near the burner center they form the crest zone of the particle volume flux, the particle volume fraction and the particle number concentration. Particles in the inner zone of the fuel-rich primary air/coal mixing channel come into collision with the central diffuser cone; as a result, they spray into the test section in one angle with the burner center. Another possibility is as follows: although they do not come into collision with the central diffuser cone, and with the guide of the central diffuser cone, they also spray into the test section at one angle with the burner center line. Then, the other crest zone of the particle volume flux, the particle volume fraction and the particle number concentration out the crest zone near the burner center is formed. So crest values of particle volume flux, particle volume fraction and particle number concentration are formed near the edge of the central recirculation zone, and with the stream development crest values always being larger. ln x/d=0.1 cross section the maximum particle volume flux of the radial biased combustion burner is 25 times as much as that of the volute burner. The particle volume fraction and the particle volume concentration are also larger in the central recirculation zone with the radial biased combustion burner (see Figure 9 and 10).

90

Zhengqi Li

Figure 96. Particle volume flux profiles with the radial bias combustion burner (a) and the volute burner (b).

Figure 97. Particle normalized volume fraction profiles with the radial bias combustion burner (--) and the volute burner (--).

In two cross sections of x/d=0.52 and 1.02, the particle volume flux crest zone near the burner center is the zone where the air mean velocities are negative. It is as much as to say that a great number of particles pass through the central recirculation zone. With the development of the stream, particles diffuse to the wall zone. In two cross sections of x/d=2.02 and 3.32, there is only one particle volume flux crest zone, and the position of the crest zone is near the wall. Meanwhile, particle volume fractions and particle number concentrations become large in the wall zone. Roughly speaking, particles are concentratively distributed to the burner central zone with the radial biased combustion burner. With the volute burner, the primary air and particle mixing itself is swirl, and particle swirl number is relatively larger. The particle radial stirring motion transport ability is stronger. Major

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners

91

particles are distributed to the wall zone which is far from the burner center. In the burner central zone particle volume flux, particle volume fraction and particle number concentration are all very little. On the other hand, in the zone far from the central recirculation zone, the particle volume flux is relatively larger, and the particle volume fraction and the particle number concentration crest value zones appear.

Figure 98. Particle normalized number concentration profiles with the radial bias combustion burner (-) and the volute burner (--).

2.5.2. Influence of Gas/Particle Flow Characteristics on Coal Combustion With the radial biased combustion burner, the particle concentration is larger near the edge of the central recirculation zone, where the gas temperature is high, and then a high temperature and high fuel concentration zone is formed. Some experiments shows: increasing the coal concentration in a defined range can reduce ignition temperature, the ignition time and the coal stream ignition heat, and increase the flame propagation velocity. With the coal concentration increasing, the coal stream degree of blackness increases, the radiation heat increases and is absorbed from the high temperature recirculation zone and the high temperature flame of the furnace by the fuel-rich coal stream [34]. They all improve the flame stability. With the volute burner, particle volume flux, particle volume fraction and particle number concentration are little in the high temperature central zone, and it is disadvantageous to the flame stability. With the radial biased combustion burner, large particles mainly gather in the burner central zone, where the temperature is high. It improves the coal burn out because the coal combustion stage is in the dynamic combustion range. Good flame stability prolongs the coal combustion time, which is advantageous to coal burn out. With the volute burner, major particles whose mean diameters are large gather in the low temperature secondary air which is far away from the central recirculation zone, it is disadvantageous to coal burn out. In the central zone of the radial biased combustion burner, particle volume fluxes and particle volume fractions are all large, and in some cross sections it appears that a great number of particles pass through the central recirculation zone. The central zone, especially the central recirculation zone, is the low oxygen zone, in which the reducing atmosphere is strong. It can reduce the formation of the fuel NOx [28, 30, 33, 47, 49, 55]. There are a great number of particles in the secondary air of the volute burner, coal early burns in the highly oxidizing atmosphere. It can increase the formation of the fuel NOx.

92

Zhengqi Li

With the radial biased combustion burner, particles have less tangential and radial velocities, major particles burn gathering in the burner central zone, it is advantageous to the formation of oxidizing atmosphere near the water-cooled wall, which increases the ash fusion point, it is advantageous to the resistance to slagging and high temperature corrosion. With the volute burner, particles have larger tangential and radial velocities; radial transport ability is strong. The phenomenon is liable to appear that a great number of particles are thrown to and pasted on the water-cooled wall. When a great number of particles burn near the watercooled wall, the reducing atmosphere is liable to be formed, it results in the water-cooled wall slagging and high temperature corrosion. Thus, the characteristics of gas/particle flow of radial biased combustion burner is advantageous to flame stability, high efficiency combustion, low NOx emission and resistance to slagging and high temperature corrosion.

2.5.3. Industrial Experiments with Radial Biased Combustion Burners The first industrial application of radial biased combustion burners was in December 1994. And so far, the radial biased combustion burners were used in 15 lean and bituminous coal-fired utility boilers, whose capacities were at the range from 50 to 670-tph, in eight power plants. Tables 16 and 17 show some representative industrial applications. The operational results are as follows: Firstly, compared with the former burners, the boilers with the radial biased combustion burners have less unburned combustible loss. Secondly, the flame stability improves. Thirdly, the NOx emission of boiler is less. Finally, the problems of high-temperature corrosion and slagging were solved. 2.5.4. The Air-Surrounding-Fuel Combustion Theory Based on the study of swirl coal burners, the air-surrounding-fuel combustion theory is put forward as follows: in coal ignition zone, high temperature and high coal concentration are formed, meanwhile, outside the ignition zone-near the water cooled wall zone, oxidizing atmosphere zone in which air is the major composition is also formed. Conforming to the theory, the measures can realize flame stability, slagging resistance, high temperature corrosion, pollution control and high combustion efficiency. 2.5.5. Conclusion With the radial biased combustion swirl coal burner, particle volume fluxes and the particle volume fractions and the particle number concentrations are larger near the edge of the central recirculation zone, particle volume fractions and the particle number concentrations are also larger in the central recirculation zone, In some cross sections it is found that a great number of particles pass through the central recirculation zone. With the volute burner, particle volume fluxes and particle volume fractions and particle number concentrations are less near the edge of and in the central recirculation zone, and they are larger in the wall zone. The industrial experiments show that radial biased combustion burner has the performance of high flame stability, high efficiency combustion, low NOx emission and resistance to slagging and high temperature corrosion.

Table 16. Industrial boilers and major problems before retrofitting


Boiler index Boiler capacity Type and layout of burners Sixteen volute burners with tangential register vanes were located in two rows on opposite walls of the furnace. Pulverized coal is conveyed by hot air.

No.

Power plant

Coal

Major problems The minimum load without auxiliary fuel oil were 180~190MWe. As the burner resistance was large, the excess O2 concentration at the furnace exit was less than 1. The carbon-in-ash was more than 20%. Serious high-temperature corrosion appeared on the front water-cooled wall. The NOx emission at rated load was 390ppm (via O2=6%). The milling system is the direct system of pulverized fuel preparation with two medium-speed pulverizers. Each pulverizer is used to supply pulverized coal to the burners in a row, and there was no spare pulverizer. With two pulverizers in service, the boiler could not operate stably without auxiliary fuel oil at the 70% rated load. When the burners in one row run with only one pulverizer in service, the auxiliary fuel oil must be used to keep the coal flame stable. The quantity of fuel oil was large. Carbon in ash was large.

Huangdao Power Plant

670-tph

Lean coal with low volatility

410-tph

Xinhua Power Plant 3 1 220-tph

Eight dual-register burners were located in two rows on the front wall of the furnace.

Bituminite

Branch Power Plant in Jiamusi Paper Factory

50-tph

Four dual-volute burners were located in one row on two opposite walls of the furnace.

Bituminite

No atomizing oil gun was installed for flame stability at low rate. Slagging appeared on the front and the rear walls.

Table 17. Experimental results after retrofitting with RBC burners


Coal fired during test Volatility, Ash, wt% wt% (as dry (as received and ash free basis) basis) 12.20 36.42 2 3 4 43.28 36.09 Bituminite 28.62 21.26 25.92 34.26

No.

Net heating value, (kJ/kg) 20550 21702 20095 19720

The ratio of the minimum boiler load without auxiliary fuel oil to the rated load of the boiler, % 48 At rated load 40. The burners in only the up or the bottom row run. 40. The burners in only the up or the bottom row run. The minimum load at which the coal flame was stable was 30tph.

Combustion efficiency, %

NOx, ppm (via O2=6%)

slagging

High-temperature corrosion Without hightemperature corrosion Without hightemperature corrosion Without hightemperature corrosion Without hightemperature corrosion

96.1* 98.1 98.2 98.1* /

310* / 310* /

Without slagging Without slagging Without slagging Without slagging

Notes: 1. The number is corresponding to that in Tab. 1; 2. All the burners of the above boiler were retrofitted by radial bias combustion burners. 3. * means the results were obtained at rated load; 4. After burners retrofitting, the 50-tph boiler was successfully preheated by a single radial bias combustion burner without auxiliary fuel oil when the boiler set up.

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners

95

The air-surrounding-coal combustion theory is given as follows: in pulverized coal ignition zone, high temperature and high pulverized coal concentration are formed, mean while, outside the ignition zonenear the water cooled wall zone, oxidizing atmosphere zone in which air is the major composition is also formed. Conforming to the theory, the measures can realize flame stability, slagging resistance, high temperature corrosion, pollution control and high combustion efficiency.

2.6. Erosion of the Central Core of the Burner with Bituminite Fired
2.6.1. Temperature Distribution on the Central Core Wall Because the coal enricher with cone vanes is installed inside the radial biased combustion burner and distant to the high-temperature gas of the furnace, the enricher temperature is low. It is certain that the enricher made of ceramic can be used for four years without failure in the utility burner. The temperature of the cone between swirling secondary air and the fuel-lean primary air/coal mixture is large. It is certain that the life of the cone made of high temperature and abrasion-resistant steel is more than four years with the fuel-lean primary air/coal mixture wearing. The central core is heated by the high-temperature gas of the furnace and simultaneously worn by the fuel-rich primary air/coal mixture. So, the erosion damage of central core is a serious problem. The tests and results of central core erosion are the discussed topic in this section. Experiments were carried out on a 410-tph boiler of a power plant. The introduction of the boiler is in 2.4.1 Section. The Nos. 5-8 burners are in the up row; the Nos. 1-4 burners are in the bottom row. The coal-fired (see Table 11) was bituminous. Wall temperature profiles of the central core along the axial flow direction of the jet were measured by the platinum-rhodium-platinum thermocouples embedded in the wall (see Figure 99).

Figure 99. Measurement for wall temperature of the core.

Figure 100 shows wall temperature profiles of central cores of the burners in the up row with the boiler at 50% rated load rate [56]. With the boiler at 50% rated load, the burners in the up row were out of service, and the burners in the bottom row were all in service with 100% open of their dampers of non-swirling and swirling secondary airs. Figure 101 shows

96

Zhengqi Li

wall temperature profiles of central cores of the burner in the up row with the boiler at 100% rated load.

Figure 100. Wall temperature profiles of the central cores of Nos. 6 () and 7 () burners with the boiler at 50% rated load.

Figure 101. Wall temperature profiles of the central cores of Nos. 6 () and 7 () burners with the boiler at 100% rated load.

With the boiler at 50% rated load, the burners of up row were all out of service. So, there was no cooling air to cool the central cores which were heated by high-temperature gas of the furnace simultaneously. The central core wall temperature was up to 1090 at the tip. With the boiler at 100% rated load, it was up to 1000 at the tip. It shows that compared with the case of 50% rated load, the temperatures of central cores of up burners decreased in the case of 100% rated load. The reason is that the central cores of up burners were cooled by the primary air/coal mixture when the burners were in service. Figure 102 shows wall temperature profiles of the central cores of Nos. 2 and 3 burners in the bottom row. With the boiler at 50% rated load, the four burners in the bottom row were in service, the maximum temperature at the tip is 900. With the boiler at 100% rated load, the maximum temperature at the tip is 950, and the temperature in the position which is a

Wall temperature

Wall temperature

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners

97

distance of 120mm to the tip is about 700. It shows that with the increase of load, the change of wall temperatures of central cores of the bottom burners was not significant, namely the effect of load on it is small.

Figure 102. Wall temperature profiles of the central cores of the burner in the bottom row. --The boilers was at 50% rated load . The four burners in up row were out of service; the four burners in bottom row were in service. -- The boilers was at rated load. All burners were in service.

In conclusion, no matter that the load is high or low, the wall temperatures of central cores of the up and bottom burners were large and they were all more than 900 at the tip.

2.6.2. Erosion of the Central Core of the Burner with Bituminite Fired 1. The Service Life of Central Core with Bituminite Fired The central core temperature was large and simultaneously the core was worn by the fuelrich primary air/coal mixture. So, the erosion damage of central core is a serious problem. Oxidizing wearing is the chief erosion, and impact wearing is the another erosion. When bituminite is fired, the velocity of primary air is always as high as 25m/s. Because the erosion rate is proportional to the cube of primary air velocity, the erosion of central core in bituminite-fired boiler is faster, and the service life of the central core made of Cr24Ni14Si2 is generally about one year. When lean-coal is fired, the velocity of primary air is always as low as 15 m/s. The erosion of central core is slower, and the service life of central core made of Cr24Ni14Si2 is more than four years. Thus, the service life of central core can satisfy the demand of lean-coal-fired boiler. To solve the problem, we tried to change structure and material of the core. Firstly, structural improvements were performed. The multi-layer structure was applied. The inner layer, facing the high-temperature gas of the furnace, is made of high temperature-resistant steel, and the outer layer, worn by the fuel-rich primary air/coal mixture, is made of high temperature and abrasion-resistant steel. The core diameter decreased. Industrial experiments were carried out on a 410-tph boiler with the new structure. The results show that the service life of the core with bituminite fired is about two years. And with lean-coal fired, it is more than six years. Secondly, the high temperature and abrasion-resistant material was improved.

Wall Temperature

98

Zhengqi Li

An alloying metal with better oxidizing resistance was used. Its chemical components include C 0.18%, Si 2.0%, Mn 0.90%, Cr 28.0%, Ni 7.8%, W 0.89%, Mo 1.0%, and proper amount of N and Re. According to the metal phase diagrams of Fe-C-Cr and Fe-Cr-Ni, this alloying metal has characteristics of + . It is a typical multi-phase metal [56]. The service life of central core made of the alloying metal is about two and a half years. The SiC was used as the material of central core in experiments. But the central core made of SiC broke up and fell off as soon as the boiler set up. So, it was failed.

2. Effect of Central Core Erosion on Performance of the Burner A two-component particle-dynamics anemometer was used to measure the characteristics of gas/particle two-phase flow with different division core angles of the radial biased combustion burner model [20]. The results, shown in Figure 103, show that with the decrease of the central core angle, the central recirculation zone diameters of jet reduce. Industrial experiment with the burner without central core was carried out on a 670-tph utility boiler. After all the central cores were taken away from the burners, the minimum load of boiler without auxiliary fuel oil increased from 100 to 140MW, and the carbon-in-ash increased. Thus, the central core was necessary to the radial biased combustion burner and the cores were installed again. In operation, with the erosion of central core, the flame stability of radial biased combustion burner declines and the minimum load of boiler without auxiliary fuel oil increases significantly.

Figure 103. Effect of central core angle on the diameter of central recirculation zone.

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners

99

3. THE CENTRALLY-FUEL-RICH SWIRL COAL COMBUSTION BURNER


3.1. Concept of the Centrally-Fuel-Rich (CFR) Swirl Coal Combustion Burner
In 2003, Professor Zhengqi Li proposed a new burner, the central-fuel-rich swirl coal combustion burner (see Figure 104), based on the radial biased combustion burner and enhanced ignition-dual register burner (see Figure 2). As both the inner and the outer secondary airs are swirling airs, the burner can have a large central recirculation zone with a large volume of outer secondary air. The centrally-fuel-rich burner has no central pipe. The primary air-coal mixing duct is in the center of the burner. The primary air is non-swirling air. Cone separators are installed in the primary air-coal mixing duct to concentrate the pulverized coal into the center of the burner. Compared with the radial biased combustion burner, the centrally-fuel-rich burner solves the abration of the central core. At the same time, the centrally-fuel-rich burner reduces the primary air resistance and is easily installed and repaired.

Figure 104. The centrally-fuel-rich swirl coal combustion burner. (1) primary air duct, (2) monitor pipe, (3) cone separators, (4) axial vanes, (5) inner secondary air duct, (6) tangential vanes, (7) outer secondary air duct.

3.2. Gas-Particle Flow and Coal Combustion Characteristics of the Centrally-Fuel-Rich Burner and Enhanced Ignition-Dual Register Burner
3.2.1. Gas-Particle Flow Characteristics of the Centrally-Fuel-Rich Burner and Enhanced Ignition-Dual Register Burner A three-component particle-dynamics anemometer is used to measure, in the near-burner region, the characteristics of gas-particle two phase flows with a central-fuel-rich swirl coal combustion burner and enhanced ignition-dual register burner, on a gas-particle two phase test facility [51].

100

Zhengqi Li

The full industrial-scale centrally-fuel-rich and enhanced ignition-dual register burner studied in the experiments was designed for a 1025 t/h coal-fired boiler (see Figure 2). A scale ratio of 1:7 was employed for two model burners. No enricher was mounted in the centrally-fuel-rich burner model (Figure 105) and glass beads were fed only into the fuel-rich duct. This simulates the extreme case in which particles in the primary air are all concentrated into the central zone of the burner. Mass flows of two burners are the same. The primary air mass flow is 0.073 kg/m3, the inner secondary air mass flow is 0.094 kg/m3, and the outer secondary air mass flow is 0.138 kg/m3. The primary air particle mass concentration is 0.20 kg (fuel)/kg (air), and that of the fuel-lean concentration was 0 kg (fuel)/kg (air). Gas/particle flow characteristics were measured in sections of x/d=0.1, 0.3, 0.5, 0.7, 1.0, 1.5, and 2.5.

The centrally-fuel-rich burner

The enhanced ignition-dual register burner Figure 105. Detail of the model burner jets (the dimensions are in mm). 1. primary air and glass beads 2. inner secondary air 3. outer secondary air.

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners

101

1. Velocity Figure 106 shows profiles of the gas/particle mean axial velocities for two burners. Two burners all have back flows near wall. From the burner jet to the x/d=0.7 cross-section, there are two peaks in the profiles of the mean axial velocities for the gas and particles for centrally-fuel-rich burner. From the burner jet to the x/d=1.0 cross-section, there are two peaks in the profiles of the mean axial velocities for the gas and particles for enhanced ignition-dual register burner. One peak near the center is the primary flow zone for the gas/particle mixing and the near the wall is the secondary airflow zone. The peak near the center is always greater than that near the wall. With diffusion of the primary gas/particle mixing into the secondary air and diffusion of the secondary air towards the wall, both peaks gradually decrease and the peak values near wall move towards the wall. For the x/d=1.0 cross-section, the peak near wall diminishes, which indicates that the primary and secondary air mix comparatively quickly. The central recirculation zone formed by the centrally-fuel-rich burner is earlier than that by the enhanced ignition-dual register burner. The central recirculation zone for the centrallyfuel-rich burner is formed in the x/d=0.3 cross section, and that for enhanced ignition-dual register burner is formed in the x/d=0.5 cross section. From the burner jet to x/d=2.5 cross section, the central recirculation zone of the centrally-fuel-rich burner are larger than that of the enhanced ignition-dual register burner. The gas-particle mean axial velocities of the enhanced ignition-dual register burner are always larger than that of the centrally-fuel-rich burner in the region of chamber axis.
velocity m/s
-5 350 300 250 200 150 100 50 0 -50 0 5 10 15 -5 0 5 10 -5 0 5 0 5 0 5 0 5 0 5

radius mm

x=17.6 mm x/d=0.1

x=52.8 mm x/d=0.3

x=88 mm x/d=0.5

x=123.2 mm x/d=0.7

x=176 mm x/d=1.0

x=264 mm x/d=1.5

x=440 mm x/d=2.5

the CFR burner (

particle - gas)

the EI-DR burner (

particle ... gas)

Figure 106. Profiles of mean axial velocities for gas and particles with the centrally-fuel-rich (CFR) and enhanced ignition-dual register (EI-DR) burners.

Near the center of the test chamber, there is a slip velocity between the gas and particle velocities, and the particle axial velocity is greater than the air axial velocity. This means that the particle velocity lags behind the gas velocity. Near the center of the test chamber, the

102

Zhengqi Li

mean axial velocities for gas and particles for enhanced ignition-dual register burner is always positive, that indicates that part of gas-particles mixing of primary flow zone completely penetrate the central recirculation zone and the residence time inside the central recirculation zone is short.. Figure 107 shows the profiles of the gas/particle RMS axial velocities for two burners. In cross-sections of x/d=0.11.5, the profiles show two peaks for two burners. In regions inside and outside the peak zone near the test chamber center, RMS axial velocities are comparatively high. The region inside the peak zone is the mixing zone between the primary gas/particle mixing and the reverse flow. The region outside the peak zone is the mixing zone between the primary gas/particle mixing and the secondary air. The peak near the wall is the secondary air diffusion zone and the RMS axial velocities are comparatively high. The two peaks indicate that there is comparatively high axial turbulent diffusion in these regions .With jet development, the two peaks for RMS axial velocities gradually decrease, the peak near the wall diffuses towards the wall and the profiles become flat. The peak near the test chamber center is always greater than the peak near the wall. In the x/d=0.1 cross section, in the radius range from 24 mm to 350 mm, gas-particle axial RMS values of the centrally-fuel-rich burner is higher than that of the enhanced ignition-dual register burner, which indicates that axis turbulent diffusion of centrally-fuel-rich burner is higher than that of enhanced ignition-dual register burner in this region. Figure 108 shows profiles of the gas/particle mean radial velocities for the two burners. From the burner jet to the x/d=0.7 section, the profiles for two burners show two peaks: the peak near the center of the chamber is the primary gas/particle mixing flow zone, and the other zone near the wall is the secondary air flow zone. The peak near the wall is always greater than the peak near the burner center. With diffusion of the primary gas/particle mixing into the secondary air and diffusion of the secondary air towards the wall, the two peaks of tow burners move towards the wall. With jet development, the profiles of the gas/particle mean radial velocities for two burners become flat. Near the center of the chamber from the burner jet to the x/d=0.7 cross-section, the mean radial velocities for centrally-fuel-rich burner are negative. This means that the primary gas/particle mixing flows towards the test chamber center, which enhances the particle volume flux near the center. Near the center of the chamber, from the burner jet to x/d=0.5 cross section, mean radial velocities for the enhanced ignition-dual register burner is close to 0 m/s; in the x/d=0.7 and 1.0 cross sections, mean radial velocities are negative this indicates the primary gas-particle mixing flows towards the test chamber center. Figure 109 shows shows profiles of the gas/particle RMS radial velocities for two burners. In the cross-sections from x/d=0.1 to 0.5, there is a peak for the RMS radial velocities for two burners, which indicates that there is comparatively high radial turbulent diffusion in this region. With jet development, the secondary air diffuses towards the wall, the peak gradually decreases and profiles of the gas/particle RMS radial velocities for two burners become flat. In the x/d=0.1 and 0.30 cross sections, gas-particle radial RMS values of the enhanced ignition-dual register burner are higher than that of the centrally-fuel-rich burner near the center of the chamber.

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners


velocity m/s
0 350 300 250 200 5 5 5 5 5 5 4

103

radius mm

150 100 50 0 -50

x=17.6 mm x/d=0.1

x=52.8 mm x/d=0.3

x=88 mm x/d=0.5

x=123.2 mm x/d=0.7

x=176 mm x/d=1.0

x=264 mm x/d=1.5

x=440 mm x/d=2.5

the CFR burner (

particle - gas)

the EI-DR burner (

particle ... gas)

Figure 107. Profiles of axial fluctuation velocities for gas and particles with the centrally-fuel-rich (CFR) and enhanced ignition-dual register (EI-DR) burners.

velocity m/s
-5 0 5 10 15 350 300 250 200 0 5 10 0 5 0 5 0 5 0 5 0 5

radius mm

150 100 50 0 -50

x=17.6 mm x/d=0.1

x=52.8 mm x/d=0.3

x=88 mm x/d=0.5

x=123.2 mm x/d=0.7

x=176 mm x/d=1.0

x=264 mm x/d=1.5

x=440 mm x/d=2.5

the CFR burner (

particle - gas)

the EI-DR burner (

particle ... gas)

Figure 108. Profiles of mean radial velocities for gas and particles with the centrally-fuel-rich (CFR) and enhanced ignition-dual register burners

104

Zhengqi Li
velocity m/s
0 350 300 250 200 150 100 50 0 -50 5 10 5 10 5 10 5 10 5 10 5 10 5 10 15

radius mm

x=17.6 mm x/d=0.1

x=52.8 mm x/d=0.3

x=88 mm x/d=0.5

x=123.2 mm x/d=0.7

x=176 mm x/d=1.0

x=264 mm x/d=1.5

x=440 mm x/d=2.5

the CFR burner (

particle - gas)

the EI-DR burner (

particle ... gas)

Figure 109. Profiles of radial fluctuation velocities for gas and particles with the centrally-fuel-rich (CFR) and enhanced ignition-dual register (EI-DR) burners.

velocity m/s
-5 350 300 250 200 150 100 50 0 -50 0 5 10 15 0 5 10 0 5 0 5 0 5 0 5 0 5

radius mm

x=17.6 mm x/d=0.1

x=52.8 mm x/d=0.3

x=88 mm x/d=0.5

x=123.2 mm x/d=0.7

x=176 mm x/d=1.0

x=264 mm x/d=1.5

x=440 mm x/d=2.5

the CFR burner (

particle - gas)

the EI-DR burner (

particle ... gas)

Figure 110. Profiles of mean tangential velocities for gas and particles with the centrally-fuel-rich (CFR) and enhanced ignition-dual register (EI-DR) burners.

Figure 110 shows profiles of the gas/particle mean tangential velocities for two burners. Because the primary air is non-swirling, the mean tangential velocities for two burners are relatively small in the x/d=0.1 cross-section (r50 mm). In the x/d=0.3 cross-section, the

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners

105

distribution of the mean tangential velocities for the centrally-fuel-rich burner is a Rankinetype vortex, which is a combination of a solid-body rotational core and a free vortex. Downstream from the x/d=0.5 cross-section, the peak mean tangential velocities for the centrally-fuel-rich burner move towards the center of the test chamber, which indicates that the gas/particle mixing near the center region begins to swirl, driven by secondary air. With jet development, profiles of the gas/particle mean tangential velocities become flat. In seven cross sections from in x/d=0.1 to 2.5, in the radius range from 0 mm to 50 mm, the mean tangential velocities for the enhanced ignition-dual register burner is smaller than that for the centrally-fuel-rich burner, which is because that the inner secondary air swirler of the centrally-fuel-rich burner is made of 16 axial bent vanes and that of the enhanced ignitiondual register burner is made of 8 axial vanes. Figure 111 shows profiles of the gas/particle RMS tangential velocities for two burners. In the x/d=0.1 cross-section, there are two peaks for the RMS tangential velocity for the centrally-fuel-rich burner in the inner and outer secondary air zones. This indicates that there is large turbulent diffusion ability in these two zones. In the x/d=0.1 cross-section, in the radius range from 24 mm to 84 mm, gas/particle RMS tangential velocities for centrally-fuelrich burner is higher than that of the enhanced ignition-dual register burner. With diffusion of the primary gas/particle mixing into the secondary air and the secondary air towards the wall, profiles of the gas/particle RMS tangential velocities for two burners become flat.
velocity m/s
0 350 300 250 200 150 100 50 0 -50 5 5 5 5 5 5 5

radius mm

x=17.6 mm x/d=0.1

x=52.8 mm x/d=0.3

x=88 mm x/d=0.5

x=123.2 mm x/d=0.7

x=176 mm x/d=1.0

x=264 mm x/d=1.5

x=440 mm x/d=2.5

the CFR burner (

particle - gas)

the EI-DR burner (

particle ... gas)

Figure 111. Profiles of tangential fluctuation velocities for gas and particles with the centrally-fuel-rich (CFR) and enhanced ignition-dual register (EI-DR) burners.

2. Particle Volume Flux Figure 112 shows profiles of the particle volume flux in the range from 0 to 100 m in different cross-sections of the centrally-fuel-rich and enhanced ignition-dual register burner,

106

Zhengqi Li

all of which have back flows near wall. In four cross-sections from x/d=0.1 to x/d=0.7, the particle volume flux shows two peaks for two burners.
particle volume flux 10 m m s
0 10 20 30 40 50-1 0 1 2 3 4 5 0.0 0.2 0.4 0.6 0.0 0.1 0.2 0.0 350 300 250 200 150 100 50 0 -50 0.1
-6 3 -2 -1

0.2 0.0 0.1 0.2 0.0

0.1

radius mm

x=17.6 mm x/d=0.1

x=52.8 mm x/d=0.3

x=88 mm x/d=0.5

x=123.2 mm x/d=0.7

x=176 mm x/d=1.0

x=264 mm x/d=1.5

x=440 mm x/d=2.5

the CFR burner

the EI-DR burner

Figure 112. Particle volume flux profiles with the centrally-fuel-rich (CFR) and enhanced ignition-dual register (EI-DR) burners.

With movement of the secondary air towards the wall, the two peaks gradually decrease and the peak near the wall move towards the wall. The peak zone near the center of the test chamber is the primary gas/particle mixing flow zone and that near the wall is the secondary airflow zone. Compared with the enhanced ignition-dual register burner, in the same cross section, the particle volume flux peak value near the jet axis of the centrally-fuel-rich burner is much closer to the chamber axis. In the cross section x/d=0.5, the maximum particle volume flux of the centrally-fuel-rich burner is 3 times as much as the of the enhanced ignition-dual register burner. In six cross sections from x/d=0.3 to 2.5, the particle volume flux in the central recirculation zone of the centrally-fuel-rich burner is much larger than that of the enhanced ignition-dual register burner. In three sections from x/d=0.3 to 0.7, the maximum particle volume flux in the central recirculation zone of the centrally-fuel-rich burner is 5 times as much as the of the enhanced ignition-dual register burner. In four cross-sections from x/d=0.1 to x/d=0.7, the particle volume flux shows two peaks and two troughs for the centrally-fuel-rich burner. The peak zone near the center of the test chamber is the primary gas/particle mixing flow zone and that near the wall is the secondary airflow zone. The peak near the center is much greater than the peak near the wall, and the absolute value of the trough near the center is also greater than that of the trough near the wall. Resulting from the burner structure and particle inertia (Figure 104), particles in the fuel-rich primary air duct are ejected directly into the center zone and form the peak for the particle volume flux near the jet axis. With jet development, the particle volume flux near the jet axis gradually decreases. In the x/d=1.0 cross-section, the peak near the wall diminishes and the two troughs combine into one. As particles have comparatively high tangential and

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners

107

radial velocities in the center of test chamber, with jet development in cross-sections from x/d=1.0 to x/d=2.5 in the radius range from 0 to 248 mm, the particle volume flux is negative; therefore, there is a peak for particle volume flux near the wall. In seven cross sections from x/d=0.1 to 2.5, in the radius range from 0 mm to 20 mm particle volume flux for the enhanced ignition-dual register burner is small resulting from conical diffuser and the particle inertia. From the burner jet to x/d=1.0 cross section, there are two peak zone, resulting from burner structures and the particle inertia. When particles ejected into the primary air duct, due either to collision with the conical diffuser or to the guidance of the conical diffuser, these particles formed a peak zone of the particle volume flux near the jet axis. With particles mixed with the secondary air, some particles are taken by the secondary and formed the other peak zone outside of the peak zone near the jet axis. Compared with the enhanced ignition-dual register burner, in the same cross section, the particle volume flux peak value near the jet axis of the centrally-fuel-rich burner is much closer to the chamber axis. In the cross section x/d=0.5, the maximum particle volume flux of the centrally-fuel-rich burner is 3 times as much as the of the enhanced ignition-dual register burner. In six cross sections from x/d=0.3 to 2.5, the particle volume flux in the central recirculation zone of the centrally-fuel-rich burner is much larger than that of the enhanced ignition-dual register burner. In three sections from x/d=0.3 to 0.7, the maximum particle volume flux in the central recirculation zone of the centrally-fuel-rich burner is 5 times as much as the of the enhanced ignition-dual register burner.

3. Particle Diameter Figure 113 shows mean diameter profiles for particle size in the range 0100 m. The mean particle diameter (d10) is the arithmetic mean diameter. In two cross-sections from x/d=0.1 to x/d=0.3, in the radius range from 20to 350 mm, the mean diameter for the enhanced ignition-dual register burner is almost larger than that of centrally-fuel-rich burner. In four cross-sections from x/d=0.1 to x/d=0.7, the particle diameter for the centrally-fuel-rich burner near the center of the test chamber is greater than those in other positions. This is because particles are ejected from the center of the fuel-rich primary air duct, and the larger a particle is, the greater is its inertia. Small particles can easily be transferred to the central recirculation zone by air. Particles near the center have small tangential velocities and axial velocities. With jet development, the positions of larger particles show little change. With further jet development, the particle diameter distribution gradually approaches a uniform value. Particle diameters for the enhanced ignition-dual register burner change little near the chamber axis which results from the radial and tangential velocities is small in these positions. 3.2.2. Influence of Gas-Particle Flow Characteristics on Coal Combustion The particle residence time in the burner central zone affects particle burnout in the burner region. For the centrally-fuel-rich burner, the primary gas/particle mixing partially penetrates the central recirculation zone and is then deflected radially. The residence time of particles in the reducing atmosphere is prolonged. From the burner jet to x/d=2.5 crosssection, the central recirculation zone of the enhanced ignition-dual register burner is smaller than that of the centrally-fuel-rich burner. Gas-particle mixing of the enhanced ignition-dual

108

Zhengqi Li

register burner completely penetrates the central recirculation zone, leaving a relatively small annular region of reverse flow.
diameter d10 m
40 350 300 250 200 50 60 40 50 60 40 50 60 45 50 55 60 45 50 55 60 45 50 55 40 45 50 55

radius mm

150 100 50 0 -50

x=17.6 mm x/d=0.1

x=52.8 mm x/d=0.3

x=88 mm x/d=0.5

x=123.2 mm x/d=0.7

x=176 mm x/d=1.0

x=264 mm x/d=1.5

x=440 mm x/d=2.5

the CFR burner

the EI-DR burner

Figure 113. Profiles of particle mean diameter with the centrally-fuel-rich (CFR) and enhanced ignition-dual register (EI-DR) burners.

The residence time of particles of the enhanced ignition-dual register burner inside the central recirculation zone is shorter than that of the centrally-fuel-rich burner. Experiments and numerical simulation [33, 58, 59] indicated that burnout quality in the burner region is primarily influenced by the particle residence time in the inner recirculation zone. Compare to the enhanced ignition-dual register burner, the centrally-fuel-rich burner is advantageous to burn-out. Some experiments showed that increasing the fuel concentration within a defined range can increase the flame velocity [34, 35, 41, 60]. Pulverized coal is concentrated in the central zone of the centrally-fuel-rich burner and thus a zone of high temperature and high fuel concentration is formed. With increasing fuel concentration, the degree of blackness of the fuel stream increases. Then, when radiation heat, which is absorbed from the hightemperature flame of the furnace, increases, the gas temperature increases. In the center region of the centrally-fuel-rich burner, there is a large particle volume flux, high gas temperature, and a zone with large RMS axial velocities, which exhibits intense heat convection with the central recirculation zone. This is advantageous for coal heating, firing and flame stability. The particle volume flux and particle diameter are large in the center of the test chamber, and large particles are mainly resident in the burner central zone, where the temperature is high. In each cross-section, particle volume flux of the enhanced ignition-dual register burner is small. Part of particles is taken by the secondary and formed a peak zone outside of the peak zone near the jet axis. The particle volume flux in the central recirculation zone of the enhanced ignition-dual register burner is smaller than that of the centrally-fuelrich burner. The residence time of particles of the enhanced ignition-dual register burner

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners

109

inside the central recirculation zone is shorter than that of the centrally-fuel-rich burner. Thus, the performance of flame stability and burnout for is better than that of the enhanced ignitiondual register burner. The particle volume flux is large in the centrally-fuel-rich burner central recirculation zone. The primary gas/particle mixing partially penetrates the central recirculation zone and is then deflected radially. The residence time of pulverized coal in the reducing atmosphere is prolonged. The central recirculation zone consists of hot burned gases with a low amount of O2. Devolatilization takes place here and hydrocarbons compete with nitrogen for the available substoichiometric amount of O2. In this reducing environment, NO formation is low and most of the reactive nitrogen is converted to N2. This inhibits the formation of fuelNOx. Staged mixing of the secondary air decreases the temperature in the near-burner region, which reduces the amount of thermal NO simultaneously with fuel nitrogen-derived NO [41]. The particle volume flux in the central recirculation zone of the enhanced ignition-dual register burner is smaller than that of the centrally-fuel-rich burner. The residence time of particles of the enhanced ignition-dual register burner inside the central recirculation zone is shorter than that of the centrally-fuel-rich burner. Thus, the performance of flame stability and burnout for is better than that of the enhanced ignition-dual register burner. So the enhanced ignition-dual register burner is disadvantageous to inhibit the formation of the fuel-NOx. Particles are ejected from the center of the centrally-fuel-rich burner and have small tangential and radial velocities, with major particles gathering in the burner central zone. This is advantageous for the formation of an oxidizing atmosphere near the water-cooled wall, which increases the ash fusion point and for resisting slagging and high-temperature corrosion. When particles are ejected into the primary air duct of the enhanced ignition-dual register burner, due either to collision with the conical diffuser or to the guidance of the conical diffuser, particles mixed with the secondary air, therefore, some particles are taken by the secondary and formed a particle volume flux peak zone near the wall. When a great number of particles burn near the water-cooled wall, the reduced atmosphere is liable to be formed, it results in the water-cooled wall slagging and high temperature corrosion.

3.2.3. Combustion Characteristics and Nox Emissions of Centrally-Fuel-Rich and Enhanced Ignition-Dual Register Burners in Utility Boilers 1. Reacting Flow Experiments in a Bituminous Coal-Fired 1025-tph Boiler (1) Utility Boiler The BandW B-1025/16.8-M type boiler with a 300-MWe unit was made by Babcock and Wilcox Beijing Co. Ltd. The opposite-wall-fired, pulverized-coal boiler with a dry-ash type furnace was equipped with 20 enhanced ignition-dual register burners. There are 12 enhanced ignition-dual register burners arranged in three rows on the front wall of the furnace. The other eight burners are arranged in two rows on the rear wall, opposite to the eight burners in the top and bottom of the three rows on the front wall. Five medium-speed mills and a positive-pressure direct-fired system are used to supply pulverized to the burners. To compare the combustion characteristics and NOx emission of the enhanced ignitiondual register and centrally-fuel-rich burners, experiments were carried out in the utility boiler. The design parameters for both burner types were uniform (Table 18).

110

Zhengqi Li

Table 18. Design parameters of the enhanced ignition-dual register and centrally-fuelrich burners in the utility boiler
Primary air 0.25967 8.27 75 Inner secondary air 0.4979 3.57 369.6 Outer secondary air 0.6677 8.37 369.6

Exit area (m2) Air mass flow rate (kg s-1) Air temperature (C)

(2) Measurements of Gas Temperature, Gas Species Concentration and Char Burnout in the Burner Region Data were obtained for the gas temperature, gas species concentration and char burnout measured in the region of burner no. 4, which was in the bottom on the rear wall and close to the right-hand side wall. Gases were sampled using a water-cooled stainless steel probe and analyzed online on a Testo 350M instrument. Gas temperature was measured using a nickel chromium-nickel silicon thermocouple placed inside a water-cooled stainless steel probe. Char sampling was also performed using a water-cooled stainless steel probe. Data were measured at positions along the jet near the burner region using a water-cooled stainless steel probe inserted through monitoring ports (Figs. 104 and 2) in the burner. The initial measurement position along the jet was located on the rear wall. Data were measured at positions along the direction from the side wall to the burner were measured through a monitoring port in the side wall, as shown in Figure 114. Measurements were made in the enhanced ignition-dual register burner region when the four burners in the bottom row on the front wall were not operating. Thus, to maintain experimental conditions consistent throughout the measurements, these four burners were also stopped when measurements were performed in the centrally-fuel-rich burner region.
Central line of water-cooled wall on the rear wall

No. 4 burner

Central line of water-cooled wall on the side wall

Monitoring port on the side wall

Figure 114. Positions of No. 4 burner and the monitoring port on the side wall (the dimensions are in mm).

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners Char burnout data were calculated from the following equation:

111

= [1 (k / x )]/(1 k )

(21)

where is the char burnout, is the ash weight fraction, and the subscripts k and x refer to the ash content in the input coal and the char sample, respectively. The percentage release of components (C, H and N) was calculated using the following equation:

= 1 [(i x / i k )( k / x )]

(22)

where i is the weight percentage of the species of interest, is the ash weight percentage and the subscripts k and x refer to different content in the input coal and char sample, respectively [61]. During the experimental campaign, the utility boiler was operated stably with a full load. The negative pressure of the furnace was stable and operating parameters such as the main steam pressure, main steam flow, etc., all met the design requirements. The characteristics of the coal used in the experiments are shown in Table 19. Table 19. Characteristics of the coal used in the experiments
Quantity Proximate analysis (as received, wt.%) Ash Volatiles Fixed carbon Moisture Net heating value (kJ kg-1) Ultimate analysis (as received, wt.%) Carbon Hydrogen Nitrogen Sulfur Oxygen Enhanced ignition-dual register burner 25.18 32.26 39.78 16.1 16920 47.05 2.29 0.62 0.82 7.94 Centrally-fuel-rich burner 27.13 33.15 40.82 11.8 17790 48.05 2.51 0.54 1.23 8.74

Figure 115 shows profiles of the gas temperature near the burner region for the two burner types. For both burners, the gas temperature along the jet first increased and then decreased, with high increase rates in the early stage. The rate of gas temperature increase was 2.38 C/mm at positions between 0 and 200 mm from the centrally-fuel-rich burner and 1.45 C/mm at positions between 0 and 200 mm from the enhanced ignition-dual register burner. Thus, the gas temperature and its rate of increase were higher in this region for the centrally-fuel-rich burner. For both burners, gas temperatures increased sharply and then remained at a high level at a distance from the burner because of rapid combustion of the pulverized coal in the high-temperature gas. At positions away from the burner, the gas

112

Zhengqi Li

temperature gradually decreased due to fuel consumption and mixing of the primary and secondary airflows.
1100

1000

Gas temperature

900

800

700

600

500 0 200 400 600 800 1000 1200 1400 1600 1800

Distance from the back wall (mm)

1100

1000

Gas temperature

900

800

700

600

500 0 200 400 600 800 1000 1200 1400 1600 1800

Distance from the side wall (mm)

Figure 115. Profiles of gas temperature measured (a) along the jet and (b) in the radial direction near the burner. Hollow symbols, enhanced ignition-dual register burner data; solid symbols, centrally-fuel-rich burner data.

The profiles in Figure 115a show that the centrally-fuel-rich burner can maintain a gas temperature up to 1000 C at a distance of 200900 mm from the burner, but the enhanced ignition-dual register burner could only maintain a gas temperature up to 1000 C at a distance of 300400 mm from the burner. Thus, the centrally-fuel-rich burner has a larger high-temperature zone, which is more advantageous for pulverized coal combustion and

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners

113

burnout. As observed in Figure 115b for both burners, gas temperature increased from the side wall to the burner center. Thus, gas close to the high-temperature central recirculation zone is at a higher temperature than that near the water-cooled side wall. The gas temperature near the water-cooled side wall was higher for the enhanced ignition-dual register burner than for the centrally-fuel-rich burner (Figure 115b). Figure 116 shows O2 concentration profiles near the burner region for the two burner types. For both burners, the O2 concentration along the jet first decreased sharply and then slowly decreased (Figure 116a).
22 20 18 16

O2 concentration (%)

14 12 10 8 6 4 2 0 0 200 400 600 800 1000 1200 1400 1600 1800 2000

Distance from the back wall (mm)


22 20 18

O2 concentration (%)

16 14 12 10 8 6 4 2 0 0 200 400 600 800 1000 1200 1400 1600

Distance from the side wall (mm)

Figure 116. Profiles of O2 concentration measured (a) along the jet and (b) in the radial direction near the burner. Hollow symbols, enhanced ignition-dual register burner data; solid symbols, centrally-fuelrich burner data.

114

Zhengqi Li

The minimum O2 concentration along the jet was 0.84% for the centrally-fuel-rich burner at 400 mm from the rear wall and 8.51% for the enhanced ignition-dual register burner at 600 mm from the rear wall. Thus, the O2 concentration in the central zone was lower for the centrally-fuel-rich burner than for the enhanced ignition-dual register burner. The reason for first sharp decrease in O2 concentration along the jet is that the pulverized coal combusts rapidly and consumes a great deal of oxygen. At positions away from the rear wall, the O2 concentration increases because oxygen in the secondary air is supplied to the primary air. As observed in Figure 116b, the O2 concentration from the side wall to the burner changed slightly for both burners. Furthermore, the O2 concentration near the side wall was higher for the centrally-fuel-rich burner than for the enhanced ignition-dual register burner. Figure 117 shows CO concentration profiles near the burner region for the two burner types. For both burners, CO concentrations along the jet first increased and then decreased (Figure 117a).
45000 40000 35000

CO concentration (ppm)

30000 25000 20000 15000 10000 5000 0 0 200 400 600 800 1000 1200 1400 1600 1800 2000

Distance from the back wall (mm)

700

600

CO concentration (ppm)

500

400

300

200

100

0 0 200 400 600 800 1000 1200 1400 1600

Distance from the side wall (mm)

Figure 117. Profiles of CO concentration measured (a) along the jet and (b) in the radial direction near the burner. Hollow symbols, enhanced ignition-dual register burner data; solid symbols, centrally-fuelrich burner data.

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners

115

The concentration along the jet first increases because combustion of the pulverized coal leads to a rapid decrease in O2 concentration. The CO concentration then decreases because the fuel burns out and the O2 concentration increases as a result of mixing of the primary and secondary airflows. The maximum CO concentration along the jet was 39448 ppm the centrally-fuel-rich burner at 500 mm from the rear wall and 20685 ppm for the enhanced ignition-dual register burner at 600 mm from the rear wall. Thus, the CO concentration in the central zone was higher for the centrally-fuel-rich burner than for the enhanced ignition-dual register burner. As observed in Figure 117b, the CO concentration from the side wall to the burner first increased and then decreased for the enhanced ignition-dual register burner; the CO concentration was very low for the centrally-fuel-rich burner. Furthermore, the CO concentration near the side wall was lower for the centrally-fuel-rich burner than for the enhanced ignition-dual register burner. Figure 118 shows CO2 concentration profiles near the burner region for the two burner types. The CO2 concentration along the jet increased and remained at high level for both burners (Figure 118a).
18 16

CO2 concentration (%)

14 12 10 8 6 4 2 0 0 200 400 600 800 1000 1200 1400 1600 1800 2000

Distance from the back wall (mm)

18 16 14 12 10 8 6 4 2 0 0 200 400 600 800 1000 1200 1400 1600

CO2 concentration (%)

Distance from the side wall (mm)

Figure 118. Profiles of CO2 concentration measured (a) along the jet and (b) in the radial direction near the burner. Hollow symbols, enhanced ignition-dual register burner data; solid symbols, centrally-fuelrich burner data.

116

Zhengqi Li

The CO2 concentration in the central zone was higher for the centrally-fuel-rich burner than for the enhanced ignition-dual register burner. The CO2 concentration first increases because combustion of the pulverized coal leads to a rapid decrease in O2 concentration; the CO2 concentration then remains at a high level while the CO concentration gradually decreases. As observed in Figure 118b, for both burners the CO2 concentration changed slightly because the O2 concentration also changed slightly. Furthermore, the CO2 concentration in the region near side wall was higher for the enhanced ignition-dual register burner than for the centrally-fuel-rich burner. Figure 119 shows char burnout near the burner region for the two burner types. The average char burnout measured along the jet was higher for the centrally-fuel-rich burner than for the enhanced ignition-dual register burner (Figure 119a). Char burnout measured from the side wall to the burner was high near the side wall for both burners (Figure 119b).
100 90 80 70

Char burnout(%)

60 50 40 30 20 10 0 400 600 800 1000 1200 1400 1600 1800

Distance from the back wall (mm)


100 99 98

Char burnout(%)

97 96 95 94 93 92 91 90 0 200 400 600 800 1000 1200 1400 1600

Distance from the side wall (mm)

Figure 119. Char burnout measured (a) along the jet and (b) in the radial direction near the burner. Hollow symbols, enhanced ignition-dual register burner data; solid symbols, centrally-fuel-rich burner data.

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners

117

Figure 120 shows the release of C, H and N from coal near the burner region for the two burner types. Hydrogen release was fastest and carbon release was slowest, which agrees with the laboratory [62, 63] and the industrial [61] results. As observed in Figure 9a, release of C, H and N from coal along the jet was faster for the centrally-fuel-rich burner than for the enhanced ignition-dual register burner, indicating that the coal combustion rate was higher for the centrally-fuel-rich burner. The release of C, H and N from coal measured from the side wall to the burner wall was high near the side wall for both burners (Figure 120b).
100 90 80

C,H,N release(%)

70 60 50 40 30 20 10 0 400 600 800 1000 1200 1400 1600 1800

C H N

Distance from the back wall(mm)

100

98

C,H,N release(%)

96

94

92

C H N
0 200 400 600 800 1000 1200 1400 1600

90

Distance from the side wall (mm)

Figure 120. Release of C, H and N from the coal measured (a) along the jet and (b) in the radial direction near the burner. Hollow symbols, enhanced ignition-dual register burner data; solid symbols, centrally-fuel-rich burner data.

For the enhanced ignition-dual register burner, under the influence of the particle deflector and conical diffuser, pulverized coal carried by the primary air mainly gathered in the region close to the primary air tube wall, with a fraction in the central zone of the primary air. A fraction of the pulverized coal carried by the central primary air entered the central recirculation zone near the burner, which is disadvantageous for ignition and burnout of the

118

Zhengqi Li

pulverized coal. Thus, the local mean O2 concentration near the enhanced ignition-dual register burner was higher compared with the centrally-fuel-rich burner, while local mean concentrations of CO and CO2, char burnout, and release of C, H and N were lower. For the centrally-fuel-rich burner, pulverized coal was concentrated in the central zone of the primary air. The majority of the coal carried by the primary air entered the center of the central recirculation zone near the burner, so the concentration of pulverized coal in the central recirculation zone increased and the residence time of the coal in the central recirculation zone was prolonged. This led to a high-temperature and fuel-rich atmosphere in the central recirculation zone, which increased the gas temperature, accelerated pulverized coal ignition, and increased the coal combustion rate, which is advantageous for burnout. Compared with the enhanced ignition-dual register burner, local mean concentrations of CO and CO2, char burnout, and release of C, H and N were higher near the centrally-fuel-rich burner, but the local mean O2 concentration was lower. Such gas concentration profiles prevent high-temperature corrosion. A fraction of the pulverized coal was carried to the side wall and since the ash fusion temperature is higher in an oxidizing atmosphere, slagging was reduced [64, 65]. Figure 121 shows NOx concentration profiles near the burner region for the two burner types. The NOx concentration along the jet first increased and then decreased near the centrally-fuel-rich burner. For the enhanced ignition-dual register burner, the NOx concentration along the jet first increased and then remained above 1100 mg/m3 at 6% O2. The NOx concentration in the center of the burner region was much higher than for the centrally-fuel-rich burner region. From the the side wall to the burner, the NOx concentration increased for both burners, with fluctuations. The NOx concentrations for both burners were almost the same. There was a peak in NOx concentration in the near-burner region along the jet of the centrally-fuel-rich burner. This is because rapid burning of the pulverized coal increases the formation of NOx and then the rapid decrease in O2 concentration forms an intense reducing atmosphere, which decreases the NOx concentration. For the enhanced ignition-dual register burner, the rapid burning of pulverized coal is also the reason for the first increase in NOx concentration. Owing to low pulverized coal and CO concentrations, high O2 concentration and the absence of an intense reducing atmosphere, the NOx concentration remained at a high level. Compared with the enhanced ignition-dual register burner, NOx concentration in the center of the centrally-fuel-rich burner region was much lower. This is because there were high concentrations of pulverized coal and CO, a low O2 concentration and an intense reducing atmosphere, which is advantageous for restraining NOx formation [66]. With increasing distance from the side wall to the burner center in the high-temperature central recirculation zone, the gas temperature and the pulverized coal concentration increased, resulting in an increase in NOx concentration for both burners.

(3) Comparison of Boiler Performance for Two Burners With eight centrally-fuel-rich burners on the bottom row, the boiler could run stably at a load of 300 MWe. The test showed that the negative furnace pressure, main steam pressure, main steam temperature, reheated steam pressure and reheated steam temperature all met the design requirements. Compared with the enhanced ignition-dual register burners, combustible material content decreased from 6.54% to5.86% in the fly ash and from 3.19% to 2.87% in the slag. The efficiency of pulverized coal combustion increased from 96.73% to 97.09%.

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners


1600 1400 1200 1000 800 600 400 200 0 0 200 400 600 800 1000 1200 1400 1600 1800 2000

119

NOx contcentration (mg/m @6%O2)

Distance from the back wall (mm)

700

NOx concentration (mg/m @6%O2)

680 660 640 620 600 580 560 540 520 500 0 200 400 600 800 1000 1200 1400 1600

Distance from the side wall (mm)

Figure 121. Profiles of NOx concentration measured (a) along the jet and (b) in the radial direction near the burner. Hollow symbols, enhanced ignition-dual register burner data; solid symbols, centrally-fuelrich burner data.

With eight centrally-fuel-rich burners on the bottom and the boiler operating at the rated load of 300 MWe, NOx emission was measured at the outlet of the air preheater. NOx emission was 843.55 mg/m3 at 6% O2 for the enhanced ignition-dual register burner and 727.67 mg/m3 at 6% O2 for the centrally-fuel-rich burner, a decrease of 115.88 mg/m3 (13.74%). A low-load experiment was performed with random coal using eight centrally-fuel-rich burners on the bottom row. Proximate analysis of the experimental coal yielded Vdaf=35.12%, Mar=15.07%, Aar=19.9%, and Qnet,ar=19080 kJ/kg. With only eight centrally-fuel-rich burners on the bottom row running, the load stabilized at 110 MWe (36.7% rated load) for 2 h, the furnace negative pressure stabilized well, flame monitoring images were normal and the

120

Zhengqi Li

overall operating condition of the boiler was good. For the enhanced ignition-dual register burners, the minimum load was up to 180 MWe (60% rated load) without auxiliary fuel oil.

2. Reacting Flow Experiments in a Lean Coal-Fired 1025-tph Boiler The 1025-tph BandWB-1025/18.3-M type boiler was made by Babcock and Wilcox Beijing CO. LTD. It is fired with pulverized coal and synchronized to a 300 MWe steam generator. The boiler was equipped with 24 EI-DR burners. It is a dry-ash type furnace. The boiler was equipped with a coal pulverizing storage system in which pulverized coal is conveyed by hot air. When burning the low grade coal, the boiler with 24 EI-DR burners has difficulty maintaining a stable flame. Table 20 shows proximate analysis of the low-grade coal. As we can see, the power plant had to burn other coals which are farther away from the power plant; increasing the operating cost. During four days (June 1-4, 2002) we determined the statistical proximate analysis of coals used in the boiler. We determined that the coal was bituminous coal (see Table 20). Even when burning bituminous coal, the boiler only operates stably at 170 MWe without auxiliary fuel oil.
Table 20. Proximate analysis of coals (as received)
Low grade coal 11.02 6.00 21.82 61.16 23 874 The bituminous coal 15.16 4.7 28.57 51.57 22 049

Volatile matter (%) Moisture (%) Ash (%) Fixed carbon (%) Net heating value(kJ kg-1)

After eight burners at the bottom row were retrofitted to the CSCC burners, the boiler runs stably at the load of 300 MWe, the carbon in ash content decreases, and the thermal efficiency is 92.06%. The boiler operates stably at a load of 140 MWe without auxiliary oil. Before retrofitting, the thermal efficiency of the boiler was 90.6%. The boiler had to use auxiliary oil to achieve flame stability under a load of 170 MWe. After CSCC burners were adopted, flame stability is greatly improved and no flame extinction occurs even with a wide variation in coal quality and boiler load. At the rated load, with the boiler operated in normal run mode and without adjusting any parameters, when compared with the boiler with 24 EI-DR burners, the NOx emissions of the boiler with the CSCC burners decrease 8%.

3.2.4. Conclusion
(1) Compared with the enhanced ignition-dual register burner, in the same cross-section, the particle volume flux peak value for the centrally-fuel-rich burner is much closer to the chamber axis and much larger near the chamber axis. In the cross section x/d=0.5, the maximum particle volume flux of the centrally-fuel-rich burner is 3 times as much as the of the enhanced ignition-dual register burner. In six cross sections from x/d=0.3 to 2.5, the particle volume flux in the central recirculation zone for the centrally-fuel-rich burner is much larger than that for the enhanced ignition-

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners

121

(2)

(3)

(4)

(5) (6)

dual register burner. In three sections from x/d=0.3 to 0.7, the maximum particle volume flux in the central recirculation zone for the centrally-fuel-rich burner is 5 times as much as the for the enhanced ignition-dual register burner. For the centrallyfuel-rich burner, most of larger particles are resident in the center and the residence time is prolonged. For the centrally-fuel-rich burner, particles penetrate the central recirculation zone partially, and are then deflected radially. For the enhanced ignition-dual register burner, particles completely penetrate the central recirculation zone. In x/d=0.3 cross section, the distribution of mean tangential velocities for the centrally-fuel-rich burner is Rankine type vortex configuration. The mean tangential velocities and RMS velocities for the centrally-fuel-rich burner are larger than that for the enhanced ignition-dual register burner. The results indicate that the local mean concentrations of CO and CO2 gas species, the gas temperatures and their rate of increase, char burnout and the release of C, H and N measured at positions along the jet near the centrally-fuel-rich burner region were all higher than the values obtained near the enhanced ignition-dual register burner region. The mean concentrations of O2 and NOx measured at positions along the jet near the centrally-fuel-rich burner region were lower. Compared with the enhanced ignition-dual register burners, at positions near the side wall, the mean O2 concentration was higher and the gas temperature and mean concentrations of CO and CO2 were lower for the centrally-fuel-rich burners. The analysis results for C, H and N release indicate that hydrogen release is most rapid and carbon is released slowly. The reacting flow experiments in the 1025-tph boilers have been done. The NOx emission of the boilers and the minimum load without auxiliary fuel oil with enhanced ignition-dual register burners was larger than that with centrally-fuel-rich burners. The combustion efficiency with enhanced ignition-dual register burners was less than that with centrally-fuel-rich burners.

ACKNOWLEDGEMENT
I am grateful to Professor Yukun Qin, who is an academician of Chinese Academy of Engineering, for reviewing this chapter. This study of the radial biased combustion swirl coal burner was sponsored by the Ministry of Science and Technology via Key Projects in The 9th Five-Year-Plan of China (Contract NO. 96A19010212/31), the State Education Commission of China Via a Doctoral Program Foundation of Institutions of Higher Learning (Contract NO. 97921307) and the State Key Laboratory of Clean Combustion of Coal at Tsinghua University. This study of the centrally-fuel-rich combustion swirl coal burner was supported by the Ministry of Education of China via the 2004 year New Century Excellent Talents in University (Contract No. NECT-04-0328), Heilongjiang Province via 2005 Key Projects (Contract No. GC05A314), State Key Laboratory of Coal Combustion at Huazhong University of Science and Technology and the National Basic Research Program of China (Contract No. 2006CB200303).

122

Zhengqi Li

REFERENCES
[1] [2] [3] [4] [5] [6] [7] [8] Xu, MH; Yuan, JW; Han, CY; Zheng, CG. Fuel, 199574, 1913-1917. Pohl, JH; Sarofim, AF. 16th Symp. (Int.) on Combust., 1976, 491-501. Ree, DP; Smoot, LD; Hedman, PO. 18th Symp. (Int.) on combust., 1981, 1305-1311. Asay, BW; Smoot, LD; Hedman, PO. Combust. Sci. and Tech., 198335, 15-31. Maiter, H; Spliethoff, H; Kicherer, A; Fingerie, A; Hein, KRG. Fuel, 199473, 14471452. Qin, YK; Li, ZQ; Wu, SH. Journal of Engineering for Thermal Energy and Power, 199712, 241-244(in Chinese). He, PA. Power System Engineering, 19884, 4-19(in Chinese). Ber, JM. Low NOx burners for boilers furnaces and gas turbines; Drive Towards the Lower Bounds of NOx Emissions. Third International Conference on Combustion Technologies for a Clean Environment, 3-6 July, 1995, Centro Cultural Delm Lisbon Portugal, 1-9. LaRue, AD; Cioffi, PL. Modern Power Systems, 19888, 42-47. Wrblewska, V; Golec, T; Swirski, J; Szymezak, J. Low NOx combustion of polish hard coals-achievements and problems; Drive Towards the Lower Bounds of NOx Emissions. Third International Conference on Combustion Technologies for a Clean Environment, 3-6 July, 1995, Centro Cultural Delm Lisbon Portugal, 29-36. Yu, ZY; Jiang, HL; Tan, HZ; Wei, TS; Dou, WY; Hui, SE; Xu, TM. Journal of Engineering for Thermal Energy and Power, 199914, 455-457 (in Chinese). Xu, XC; Chen, CH; Qi, HY; He, R; You, CF; Xiang, GM. Fuel Processing Technology, 200062, 153-160. Li, ZQ; Sun, R; Chen, LZ; Wan, ZX; Wu, SH; Qin, YK. Fuel, 200281, 829-835. Ma, CY. Experimental study on Radial biased combustion Swirl pulverized coal burners. Ph. Dissertation, Harbin Institute of Technology, Harbin, P.R. China, 1996 (in Chinese). Grant, G; Tabakoff, W. Journal of Aircraft, 197512:471-478. Wakeman, T; Tabakoff, W. Measured particle rebound characteristics useful for erosion prediction. Report No.82-Gt-170. The American society of Mechanical engineers, New York, 1982. Ma, CY; Xu, X; Wu, SH; Li, ZQ; Qin, YK. Power System Engineering, 199612, 36-41 (in Chinese). Sun, R; Li, ZQ; Sun, SZ; Ma, CY; Wu, SH; Qin, YK. Power System Engineering, 199915, 48-53 (in Chinese). Li, ZQ. Study on characteristics of gas-particle two phase flow with Radial biased combustion swirl pulverized coal burner and their applications. Ph. Dissertation, Harbin Institute of Technology, Harbin, P.R. China, 1997 (in Chinese). Wang, L. Effect of structure and operational parameters on gas-particle two phase flow field of the Radial biased combustion swirl burner. Ph. Dissertation, Harbin Institute of Technology, Harbin, P.R. China, 1999 (in Chinese). Sun, R. Experimental study and numerical simulation on flow properties of the Radial biased combustion swirl burner. Ph. Dissertation, Harbin Institute of Technology, Harbin, P.R. China, 1998 (in Chinese).

[9] [10]

[11] [12] [13] [14]

[15] [16]

[17] [18] [19]

[20]

[21]

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners

123

[22] Li, ZQ; Sun, R; Wan, ZX; Sun, SZ; Wu, SH; Chen, LZ. Combustion Science and Technology, 2003175, 1979-2014. [23] Li, ZQ; Sun, SZ; Sun, R; Wan, ZX; Zhang, Z; Chen, LZ; Wu, SH; Qin, YK. Journal of Engineering Thermophysics, 199920, 525-528. [24] Li, ZQ; Yang, LB; Qiu, PH; Sun, R; Chen, LZ; Sun, SZ. International Journal of Energy Research, 200428, 511-520. [25] Wei, XL; Xu, TM; Hui, SE. Energy Conversion and Management, 200445, 725-735. [26] Li, ZQ; Wan, ZX; Sun, SZ; Chen, LZ; Wu, SH; Qin, YK. Energy, 200227, 1119-1130. [27] Truelove JS, Wall TF, Dixon TF, Stewart IMC. Flow, mixing and combustion within the quarl of a swirled, pulverized-coal burner.In: Coal Combustion Techniques, Nineteenth Symposium (International) on Combustion /The Combustion Institute, 1982, Haifa, 1181-1187. [28] Smart, JP; Knill, KJ; Visser, BM; Weber, R. Reduction of NOx emissions in a swirled coal flame by particle injection into the internal recirculation zone. In: CombustionGenerated NOx and SOx, Twenty-Second symposium (International) on Combustion/The combustion Institute, 1988, Seattle, 1117-1125. [29] Costa, M; Costen, P; Lockwood, FC; Mahmud, T. Detailed measurements in and modelling of an industry-type pulveized-coal flame. In: Coal Boilers/Furnaces, Twenty-Third Symposium (International) on Combustion/The Combustion Institute, 1990, Orleans, 973-980. [30] Abbas, T; Costen, P; Lockwood, FC. Combustion and Flame, 199191, 346-363. [31] Smart, JP; Morgan, DJ; Roberts, PA. The effect of scale on the performance of swirl stabilized pulverized coal burners. In: Furnaces and Burners, Twenty-Fourth Symposium (International) on Combustion/The Combustion Institute, 1992, Sydney, 1365-1372. [32] Abbas, T; Costen, P; Lockwood, FC; Romo-Millares, CA. Combustion and Flame, 199393, 316-326. [33] Abbas, T; Costa, M; Costen, P; Gody, S; Lockwood, FC. Fuel, 199473, 1423-1436. [34] Godoy, S; Hiriji, K; Ismail, M; Lockwood, FC. Journal of the Institute of Energy, 198659, 38-44. [35] Horton, MD; Goodson, FP; Smoot, LD. Combustion and Flame, 197728, 187-195. [36] Pershing, DM; Wendt, JOL. Pulverized coal combustion: the influence of flame temperature and coal composition on thermal and fuel NOx. In: Colloquium on Combustion and Related Processes in Energy Production from Coal, Sixteenth Symposium (lnternational) on Combustion/The Combustion Institute, 1976, Massachusetts, 389-399. [37] Wendt, JOL; Pershing, DW; Lee, JW; Glass, JW. Pulverized coal combustion: NOx formation mechanisms under fuel rich and staged combustion conditions. In: Colloquium on Coal Combustion, Seventeenth Symposium (lnternational) on Combustion/The Combustion Institute, 1979, Leeds, 77-87. [38] Song, YH; Blair, DW; Siminski, VJ; Bartok, W. Conversion of fixed nitrogen to N2 in rich combustion. In: Combustion Generated Pollution, Eighteenth Symposium (International) on Combustion/The Combustion Institute, 1981, Waterloo, 53-63. [39] Fenimore, CP; Fraenkel, HA. Formation and interconversion of fixed-nitrogen species in laminar diffusion flames. In: Combustion Generated Pollution, Eighteen Symposium (International) on Combustion/The Combustion Institute, 1981, Waterloo, 143-149.

124

Zhengqi Li

[40] Chen, SL; Heap, NP; Pershing, DW; Martin, GW. Influence of coal composition on the fate of volatile and char nitrogen during combustion. In: NOx in Coal Combustion, Nineteenth Symposium (International) on Combustion/The Combustion Institute, 1982, Haifa, 1271-1280. [41] Song, YH; Pohl, JH; Beer, JM; Sarofim, AF. Combustion Science and Technology, 198228, 31-39. [42] Okazaki, K; Shishido, H; Nishikawa, T; Ohtake, K. Separation of the basic factors affecting NO formation in pulverized coal combustion. In: Coal Combustion, Twentieth Symposium (International) on Combustion/The Combustion Institute, 1984, Ann Arbor, 1381-1389. [43] Miller, JA; Branch, MC; Mclean, WJ; Chandler, DW. The conversion of HCN to NO and N2 in H2O2HCNAr flames at low pressure. In: Reaction Kinetics, Twentieth Symposium (International) on Combustion/The Combustion Institute, 1984, Ann Arbor, 673-684. [44] Hill, SC; Smoot, LD; Smith, PJ. Rediction of nitrogen oxide formation in turbulent coal flames. In: Coal Combustion, Twentieth Symposium (International) on Combustion/The Combustion Institute, 1984, Ann Arbor, 1391-1400. [45] Kranlich, JC; Randall, W; Gary, S; Samuelsen, S. Fuel, 198867, 1182-1189. [46] Afonso, R; Dusatko, GC; Pohl, JH. Combustion Science and Technology, 199393, 4145. [47] Smart, JP; Morgen, DJ. Fuel, 199473, 1437-1442. [48] Pickett, LM; Jackson, RE; Tree, DR. Combustion Science and Technology, 1999143, 79-107. [49] Schnell, U; Kaess, M; Brodbek, H. Combustion Science and Technology, 199393, 91109. [50] Shi, R.X; Chehroudi, B. Journal of Fluids Engineering, 1994116, 685-693. [51] Sloan, DG.; Smith, PJ; Smoot, LD. Combustion Science and Technology, 198612, 163250. [52] Nazzer, WA; Pickett, LM; Tree, DR. Combustion Science and Technology, 1999143, 63-77. [53] Gody, S; Hassan, MA; Ismail, MA; Lockword, FC. Combustion Science and Technology, 199180, 137-150. [54] Li, ZQ; Wu, SH; Sun, R; Chen, LZ; Sun, SZ; Qiu, PH; Sun, E; Qin, YK. Power Engineering, 199717, 27-31, 11. [55] Kramlich, JC; Seeker, WR; Samuelsen, GS. Fuel, 198867, 1182-1189. [56] Niu, HF. Study on wearing resistance at parts of the radial biased combustion swirl burner at high temperature. Ph. Dissertation, Harbin Institute of Technology, Harbin, P.R. China, 2002 (in Chinese). [57] Chen, ZC. Studies on characteristics of gas/particle two phase flows with centrally-fuelrich swirl coal combustion burner and its application. Ph. Dissertation, Harbin Institute of Technology, Harbin, P.R. China, 2007 (in Chinese). [58] Godoy, S; Hassan, MA; Ismail, MA; Lockwood, FC. Combustion Science and Technology, 199180, 137-150. [59] Gu, MY; Zhang, MC; Fan, WD; Wang, L; Tian, FG. Fuel, 200584, 2093-2101. [60] Zhou, H; Cen, KF. Powder Technology, 2006170, 94-107.

Radial-Bias-Combustion and Central-Fuel-Rich Swirl Pulverized Coal Burners

125

[61] Costa, M; Silva, P; Azevedo, JLT. Combustion Science and Technology, 2003175, 271 289. [62] Ismail, M. Char burnout and flame stability in pulverized fuel furnace. Ph. Dissertation, University of London, 1989. [63] Smoot, LD; Hedman, PO; Smith, PJ. Progress in Energy and Combustion Science, 198410, 359441. [64] Huffman, GP; Huggins, FE; Dunmyre, GR. Fuel, 198160, 585597. [65] You, C; Zhou, Y. Energy and Fuels, 200620, 18551861. [66] Van der Lans, RP; Glarborg, P; Dam-Johansen, K. Progress in Energy and Combustion Science, 199723, 349377.

In: Leading-Edge Electric Power Research Editor: C. M. OSullivan, pp. 127-171

ISBN: 978-1-60021-981-8 2008 Nova Science Publishers, Inc.

Chapter 2

FUEL CELL COMBINED CYCLE POWER GENERATION SYSTEM INSTALLED INTO MICRO-GRID
Shinya Obara
Department of Mechanical Engineering, Tomakomai National College of Technology. Tomakomai 059-1275, Japan

ABSTRACT
The introduction to urban areas of the micro-grid system has the following characteristics. (a) The distance between the heat-supply side and the heat-demand side is short, and effective utilization of exhaust heat is possible. (b) It is linked with the load leveling of the existing large-sized electric power facilities. (c) Since a facility suitable for the energy-demand characteristics of a region can be installed, energy efficiency may increase and facility costs may decrease. The micro-grid using a proton exchange membrane type fuel cell (PEM-FC) may greatly reduce environmental impact. However, when connecting an energy system to the micro-grid of a city area and operating, partial load operation occurs frequently and power generation efficiency falls. And, the electrode material (especially the catalyst material and the proton exchange membrane) of PEM-FC is expensive, and its system is complex. Consequently, it is necessary to connect two or more power generation systems to the micro-grid, and to design optimization of an operation plan for the purpose of maximization of power generation efficiency. Therefore, the methods of an improvement of the efficiency of the power generation system connected to the micro-grid installed into a city area are described. In this chapter,
E-mail: obara@indigo.plala.or.jp; Phone and FAX: +81-144-67-8010; 1985 Hakodate National College of Technology 1987 Bachelor of Engineering Nagaoka University of Technology 1989 Master of Engineering, Nagaoka University of Technology 2000 Doctor of Engineering Hokkaido University 1989 Engineer, Takasago Thermal Engineering Co., Ltd. 1991 Researcher, Equos Research Laboratory, Aisin AW Co., Ltd. 2001 to date Associate Professor, Department of Mechanical Engineering, Tomakomai National College of Technology. Research field : Optimal planning of energy system, Operation plan of fuel cell system, Radiation, Bio-ethanol reformer, Fuel cell co-generation, Micro-grid.

128

Shinya Obara
it consists of subjects of three studies on the micro grid. In these studies described in this chapter, the improvement of the subject of the micro grid is tried by combining fuel cell and other power equipment. Section 1 describes "Operation Plan of Micro Grid Using PEM-FC/Diesel Engine Generator Combined System." Section 2 describes "CarbonDioxide Emission Characteristic of Micro Grid Using PEM-FC/Hydrogenation City gasEngine Combined System." Section 3 describes "Dynamic Characteristics of Micro Grid Using PEM-FC/Woody Biomass Engine Combined System."

NOMENCLATURE
CP

E Ec
EF
Eg

G G
I NE

NR

n
P Pc

Pg

Pga Qh

Q
R

t W

system interconnection device power kW generation capacity kW power of the inverter outlet kW production of electricity kW CO2 emission g/s CO2 emission g/(s kW) integration parameter installed number of NEG number of grid routes grid route proportionality parameter interconnection power with other grids kW production of electricity of the generating equipment of the system interconnection grid kW pressure gauge amount of fuel supply kg/s heat quantity kW load factor % sampling time s objective function power generation efficiency % total power generation efficiency % boiler representative day diesel engine generator city-gas engine generator (NEG) exhaust gas of SEG cooling water of SEG heat radiation of SEG reforming gas PEM-FC

Greek Symbols

total Subscripts
B

Day
DEG
E Ex Ey

Ez Er

FC

Fuel Cell Combined Cycle Power Generation System Installed into Micro-Grid
l m n p
R S SEG

129

house model with PEM-FC system and NEG house model with NEG house model without PEM-FC system and NEG city-gas engine reformer reformer burner stirling engine

INTRODUCTION
The introduction to a urban area of a micro-grid 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 micro-grid, the scale of equipment for distributing electricity is small [13]. Furthermore, (d) Connecting renewable energy considering regionality is expected as an advanced system in micro-grid technology. The proton exchange membrane type fuel cell (PEM-FC) 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 PEM-FC to the micro-grid 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 largecapacity fuel cell. As technology to solve this subject, a fuel cell is divided into smallcapacity units, and there is the method of increasing the load factor of each unit [4]. However, by this method, the number of fuel cell units increases greatly, and facility cost increases. In Section 1, the base load of a micro-grid is supplied using a diesel engine generator (DEG), and how to install and interconnect two or more PEM-FC grids is examined. The compound grid of DEG and PEM-FC is interconnected in this study. This micro-grid is described as CIM (Compound Interconnection Micro-grid). 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 paper 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 micro-grid 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 maximum efficiency point is examined. On the other hand, the dynamic characteristics at the time of load fluctuation and carbon dioxide emissions of PEMFC are good compared with DEG [5]. PEM-FC linked to CIM is controlled to operate

130

Shinya Obara

corresponding to the load fluctuation of the grid. However, the load factor of PEM-FC changes with the power-demand patterns of each building linked to CIM in this system. At present, the method of a micro-grid interconnecting with commercial power, etc. is investigated (interconnect micro-grid) [3]. However, in order to achieve the advantages of (a) to (d) described above, it is necessary to operate a micro-grid independently. The subjects of the independent micro-grid 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 micro-grid. An improvement in power-generation efficiency is expected from the independent micro-grid 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 micro-grid, 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. 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 [6-9]. The power-generation system using a city gas engine with generator (NEG) is cheap compared with the fuel cell. So, in the Section 2 in this chapter examines the powergeneration efficiency and the carbon dioxide emissions when connecting NEG and the PEMFC to the independent micro-grid. Authors investigated until now about the operating method that connects distributed PEM-FC in a power network, and cooperates [4, 5]. Although generation efficiency of the PEM is high, greenhouse gas discharges by the reforming reaction of city gas. On the other hand, micro-combined heat and power (micro-CHP) using a small-scale Stirling engine generator (SEG) is examined in U.K. as an energy system for individual houses [10, 11]. 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 [5, 12, 13]. However, compared with an internal combustion engine or a fuel cell, the conventional SEG has subjects in generation efficiency, volume efficiency, equipment cost, etc. So, in the Section 3, the dynamic characteristics of the power of the independent micro-grid using hybrid cogeneration (PWHC) of PEM-FC 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 [14, 15]. 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 micro-grid, it is necessary to clarify the dynamic characteristics of the power with load fluctuation. So, in this study, the dynamic characteristics of the PWHC micro-grid are clarified by a numerical analysis using the results of an investigation of the SEG test machine and PEM-FC.

Fuel Cell Combined Cycle Power Generation System Installed into Micro-Grid

131

SECTION 1
Compound Interconnection Micro-grid Composed from PEM-FC and Diesel Engine Generator
1.1. Compound Interconnect Micro-Grid (1) Micro-Grid Model This Section examines two or more PEM-FC grids and the independent micro-grid built by a diesel engine generator (DEG) operated according to a base load. Figure 1 shows the compound interconnection micro-grid model (CIM) that introduced seven PEM-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 differ. Therefore, the power load pattern of each PEM-FC grid changes with the route of the building linked to the grid. Consequently, as shown in Figure 1 (a), the building linked to each PEM-FC grid is selected and arranged with the object of maximizing power generation efficiency. Figure 1 (b) shows the model of the PEM-FC grid (PEM-FC grid A to G) in Figure 1 (a), and power supply-and-demand is possible for each grid through the system interconnection device (CP1 to CP7). The system interconnection between 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. (2) CIM Model Figure 1 (c) shows the model of the FC grid and the interconnection device shown in Figures 1 (a) and (b). In CIM, DEG of with capacity of PDEG is installed and PEM-FC with capacity from PFC , A to PFC ,G is installed in PEM-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 1 (a). DEG is operated by constant load. In the proposed system, it corresponds to power load fluctuation with PEM-FC grid A to G.

(3) Facility Scheme Figure 2 (a) to (c) is a facility scheme installed in the building linked to CIM shown in Figure 1. Figure 2 (a) shows the facility scheme of a building of installing DEG, and Figure 2 (b) shows the facility scheme of a building of installing PEM-FC. The building in which is installed the facility of DEG shown in Figure 2 (a) is connected to any one grid, and the building in which is installed the facility shown in Figure 2 (b) is connected to all the PEMFC grids. Figure 2 (c) shows the facility scheme of a building in which DEG or PEM-FC 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 2 (a). Moreover, the generating equipment composed from a city gas reformer, PEM-FC, a boiler, a heat storage tank, a system interconnection device, etc. is installed in Figure 2 (b).

132

Shinya Obara

Figure 1. Compound Interconnection Micro-grid (CIM) model.

Fuel Cell Combined Cycle Power Generation System Installed into Micro-Grid

133

Figure 2. Equipment model installed in a building and an operation plan.

134

Shinya Obara

In the reformer, reformed gas is produced on a catalyst by making the combustion gas of city 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 the 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 2 (c).

(4) CIM Operating Method The model of the operating method of CIM is shown in Figure 2 (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.

1.2. Equipment Characteristics (1) Diesel Engine Power Generator The output characteristics result of the cogeneration system using DEG is shown in Figure 3 (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 exhaust gas heating value, and the production of electricity. The engine specifications of the cogeneration system of Figure 3 (a) are shown in Table 1 (a). Moreover, the specifications of a synchronous power generator are shown in Table 1 (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 3 (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 3 (b) is used in the analysis of this research. Moreover, the relation between the load factor and power generation efficiency shown in Figure 3 (b) should be maintained even if the capacity of DEG changes.

Fuel Cell Combined Cycle Power Generation System Installed into Micro-Grid

135

Figure 3. Power generator model.

136

Shinya Obara

(2) Solid Polymer Membrane-Type Fuel Cell The output characteristics of the fuel cell stack used in the analysis of this study are shown in Figure 3 (c). The maximum power generation efficiency of PEM-FC shown in Figure 3 (c) is 32%. The supply of city 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 PEM-FC by the calorific power adding these two city gas systems [16-19]. The electrode area of an anode and cathode of the examined PEM-FC is 1.0 m2, respectively. Moreover, the cost, calculated by dividing the calorific power of hydrogen in the reformed gas by the calorific power of the two city gas systems described above is defined as reformer efficiency.
Table 1. Specification of equipment

1.3. Analysis Method (1) Route Plan of Compound Interconnect Grid Equation (1) is an expression of total power generation efficiency total ,t in samplingtime 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 (2) is the objective function. Objective function WDay is equal to total power generation efficiency

total , Day of a representative day, and is obtained using Eq. (1). In the analysis of this study,
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.
E DEG ,t DEG ,t +

total ,t =

(E
n =1

NR

FC , n , t

FC , n ,t )

(1)

Etotal ,t

WDay = total , Day =

(
23 t =0

total , t

(2)

Fuel Cell Combined Cycle Power Generation System Installed into Micro-Grid

137

When there is the same pattern among the power-demand patterns of each house introduced into a city area model, two or more grid routes considered to be optimal appear.

(2) Analysis Flow The analysis flow that searches for the optimal solution of CIM is shown in Figure 4. First, the power-demand data of each house that composes a city area model is given to the analysis program. Next, the base load of a micro-grid 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 PEM-FC 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 PEM-FC is used to calculate the load ratio from the power demand of all the houses connected to a grid and the capacity of DEG and PEM-FC that were set up beforehand, and these are calculated by inserting them into Figures 3 (b) and (c). The power generation efficiency of DEG and each route of FC grid is given to Eq. (1), and the route of FC grid in case WDay of Eq. (2) is the
largest value, and the capacity of DEG and PEM-FC are decided to be the optimal solutions.

Figure 4. Calculation flow of generation efficiency for FC micro-grid.

138

Shinya Obara

(3) Power Demand Model Figure 5 shows the power demand model of each house in Tokyo in Japan used in the analysis, and is the mean power load of each sampling time of the representative day in January (winter), May (mid-term), and August (summer) [20-22]. 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 [23]. 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 5 is difficult, and the power demand estimate of the small offices and factories indicated in Figures 5 (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 study, 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 Figures 5) shows a peak in the morning and the afternoon. The demand of hotels (Figure 5 (e)) stabilized when midnight to early morning was excluded, and there is continuous power demand at convenience stores (Figure 5 (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 to evening with high power demand is clear in offices (Figure 5 (g)), factories (Figure 5 (h)), and hospitals (Figure 5(i)).

1.4. Case Study (1) Urban Area Model Figure 6 shows the urban area model used for analysis. The number shown in the figure is the house number, and also shows the type of each house. The urban area model is composed from 20 houses, and Table 2 shows the type of each house. The analysis investigates each case of Table 2 (a) and (b). Case A is a model (complex community) assuming an urban area that consists of various houses, and Case B is a model (residential area) assuming a residential street. In addition, each power demand pattern uses Figure 5. For example, house numbers 1 and 2 are family households (two persons), and each power demand model used for analysis is shown in Figure 5 (b). Therefore, the grid route of house numbers 1 and 2 is exchangeable. The power demand model of Figure 5 and the urban area model of Figure 6 are installed into the analysis program described in Figure 4, and the efficiency of the CIM system is verified in analysis. However, in the analysis, the power demand model of a representative day in May of Figure 5 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 study.

Fuel Cell Combined Cycle Power Generation System Installed into Micro-Grid

139

Figure 5. Power demand models.

140

Shinya Obara Table 2. Energy demand for urban model

Figure 6. Urban area model.

(2) Complex Community a. Grid Route and Generation Capacity of FC and DEG Figure 7 (a) shows the rate of the power demand of a representative day in May in the urban area model of Case A of Table 2(a). A representative day shows the greatest power demand for convenience stores (two houses), followed by hotels, factories, and small hospitals, in that order. As Figure 5 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.

Fuel Cell Combined Cycle Power Generation System Installed into Micro-Grid

141

Figure 7. Analysis results in Case A.

142

Shinya Obara Table 3. Result of power generation efficiency

b. Result of Power Generation Efficiency In order to maintain the high power generation efficiency of the whole micro-grid, it is necessary to plan the path of the FC grid containing convenience stores, hotels, factories, small hospitals, etc. with a large power demand. The grid route shown in Figure 7 (b) was obtained from the result of analyzing the power generation efficiency of each grid route. Figure 7 (b) consists of three FC grids: Grid A, Grid B, and Grid C. Figure 7 (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. PEM-FC 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. Figure 7 (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 house composition of Case A. In addition, the power supply of a representative day has more DEG than the sum total of each PEM-FC. Figure 7 (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 houses with the same power demand model in the urban area model, the grid routes shown in Figure 7 (b) differ, but there is a case where Figures 7 (c), (d), and (e) show the same results. Moreover, one set of DEG or one set of PEM-FC 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 houses (central system) is shown in Table 3 (a). The power generation efficiency of the DEG central system 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 study is larger. The result of the load distribution of the whole micro-grid of Case A is shown in Figure 8 (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 micro-grid (equal to CIM efficiency) are influenced.

Fuel Cell Combined Cycle Power Generation System Installed into Micro-Grid

143

Figure 8. Load distribution of the entire micro-grid.

(3) Residential Area a. Grid Route and Generation Capacity of FC and DEG Figure 9 (a) shows the rate of the power demand of a representative day in May in the urban area model of Case B in Table 2(b). In Case B, family households account for 18 houses and convenience stores account for two houses. 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 9 (b) was obtained from the analysis result of the power generation efficiency of each grid route. Figure 9 (b) consists of two FC grids: Grid A and Grid B. Figure 9 (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. PEM-FC 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. b. Result of Power Generation Efficiency Figure 9 (d) shows the analysis result of the electric energy to be outputted on a representative day in May by each FC grid.

144

Shinya Obara

Figure 9. Analysis results in Case B.

Fuel Cell Combined Cycle Power Generation System Installed into Micro-Grid

145

From the analysis result of Figure 9 (d), with the composition of the houses 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 PEM-FC. Figure 9 (e) shows the analysis result of the power generation 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 PEM-FC 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 houses is shown in Table 3 (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 study is larger. The result of the load distribution of the whole micro-grid of Case B is shown in Figure 8 (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 micro-grid is high.

SECTION 2
Amount of CO2 Discharged from Compound Micro-grid of Hydrogenation City-Gas Engine and PEM-FC
2.1. System Scheme (1) IMPE Model Figure 10 (a) shows a system-interconnection micro-grid. This system is interconnected with commercial power, etc. Power Pc is delivered and received between other grids, and Power Pg is supplied to a micro-grid with the generating equipment installed in the
machinery room of House 5 in the urban area model of Figure 10 (a). The power quality (frequency, voltage) of the system-interconnection micro-grid 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 10 (b) shows an independent micro-grid that does not interconnect with other grid systems. The method that supplies the power of an independent micro-grid by a one-set power-generation system is defined as a centralized system. Two sets of NEG or PEM-FC system are introduced, and how to divide into baseload operation and fluctuating-load operation, and supply power is defined as a base loadsharing system. However, the base load-sharing 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 PEM-FC 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.

146

Shinya Obara

Figure 11. Output share model of load power.

(2) Operation Method of the Micro-grid Figure 11 shows the power load pattern of the independent micro-grid shown in Figure 10 (b). The load pattern of Figure 10 (b) is separated into a base-load region and a fluctuatingload region in systems other than the centralized system. As Figure 10 (b) shows, PEM-FC system of Capacity PFC ,l is installed in House 5 linked to Grid A, and NEG of Capacity PE , m
is installed in House 19 linked to Grid B. Grid A and Grid B can deliver and receive the power by system interconnection equipment CP. Therefore, PEM-FC system of House 5 is made to correspond to the base-load region of Figure 11, and NEG of House 19 is made to correspond to a fluctuating-load region.

(3) Equipment Scheme Figure 12 shows an example of equipment schemes of the house connected to IMPE shown in Figure 10 (b). Figure 12 (a) shows the equipment scheme of House m linked to an NEG central system. The generating equipment installed with a centralized system is any one set of FC or NEG. Figure 12 (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

Fuel Cell Combined Cycle Power Generation System Installed into Micro-Grid

147

supplied through reformed gas piping. However, equipment cost can also be reduced by installing a city gas reformer in the same house, m , as NEG. NEG and PEM-FC system are installed in House l , and the equipment scheme of the IMPE system corresponding to base load or fluctuating load is shown in Figure 12 (b). The hydrogen produced by the reformer is supplied to NEG at the time of low load, and PEM-FC stack. NEG, PEM-FC stack, a city gas reformer, a boiler, a heat storage tank, an interconnection device, etc. are installed in the house shown in Figure 12 (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.

Figure 12. Energy equipment model.

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 12 (c) shows the equipment scheme of House n in which NEG or PEM-FC system is not installed. The power demand of House n is received from a micro-grid through an

148

Shinya Obara

interconnection device. 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 ).

2.2. Equipment Characteristics (1) Output Characteristics of Gas Engine Power Generator Figure 13 (a) shows the examination results of the hydrogenation rate and brake thermal efficiency of a one-cylinder city gas engine [24]. 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 13 (b) shows the relation between the mean effective pressure and brake thermal efficiency, and the hydrogenation rate [24]. The broken line shown in this figure is the hydrogenation rate indicating the maximum thermal efficiency. Figure 13 (c) shows the analysis results of the production of electricity of NEG, city gas consumption, and the amount of hydrogenation calculated from the model of Figures 13 (a) and (b). The amount of hydrogenation of Figure 13 (c) is the result when obtaining the maximum thermal efficiency.

Figure 13. Output characteristics of NEG.

Fuel Cell Combined Cycle Power Generation System Installed into Micro-Grid

149

The specifications of a city gas engine and a power generator are shown in Tables 4 (a) and consumption is zero when the production of electricity exceeds 14 kW as shown in Figure 13 (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 13 (a) describes. Figure 13 (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 13 (d) includes reformer efficiency. The reformer is as common as the PEM-FC system described below. Detail of reformer efficiency is given by following section.

(b). Hydrogen Figure 14 (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 13, and the equations (from Eq. (3) to Eq. (5)) described below. The fuel supplied to NEG has many hydrogen rates in a lowload 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. Figure 14 (b) shows the model of a load factor and CO2 emissions calculated from Figure 14 (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 latter section and Figure 15. (2) Carbon Dioxide Emissions of NEG Equation (3) is a steam-reforming reaction equation of city gas (CH4). Since Eq. (3) is an endothermic reaction, the heat for advancing the response of Eq. (3) is produced using the combustion reaction of CH4 shown in Eq. (4). Moreover, Eq. (5) is an equation that changes the carbon monoxide of Eq. (3) into carbon dioxide and hydrogen. If the hydrogen quantity supplied to NEG and PEM-FC stack is decided, the amount of city gas supplied to a reformer and the carbon dioxide to be discharged are calculable using Eqs. (3), (4), and (5). The CH4 quantity supplied to an engine is calculable using Eqs. (4), Figures 13 (a), and 13 (b).
CH 4 + H 2O CO + 3 H 2 206 [kJ/mol]

CH 4 + 2 O 2 CO 2 + 2 H 2O + 802 [kJ/mol] CO + H 2O CO 2 + H 2 + 41 [kJ/mol]

(3) (4) (5)

Equation (6) expresses the amount of carbon dioxide discharged by NEG. GE , p,t is the carbon dioxide emissions when burning CH4 with Engine p . GR , p ,t is the amount of carbon dioxide discharged 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 load-sharing system.

150
GNEG ,t =

Shinya Obara

(G
p =1

NE

E , p ,t

+ GR , p ,t + GS , p ,t )

(6)

Table 4. Specifications of NEG

Figure 14. CO2 emission characteristics of NEG.

Figure 15. Output characteristics of PEM-FC system.

(3) PEM-FC System Figure 15 shows the model of the generation efficiency of PEM-FC system and city gas reformer efficiency [25]. Moreover, generation efficiency FC of Figure 15 (a) was calculated using Eq. (7). When the sampling time is expressed with t , EFC ,t of Eq. (7), the right-hand side is the power in the inverter outlet of a PEM-FC system. QR, FC ,t expresses the calorific

Fuel Cell Combined Cycle Power Generation System Installed into Micro-Grid

151

power of CH4 for reforming, and QS , FC ,t expresses the calorific power of CH4 supplied to a heat-source burner. The maximum generation efficiency of the fuel cell shown in Figure 15 is 31%. Moreover, the reformer efficiency in Figure 15 (a) improves with the increase in a load factor. Figure 15 (b) shows the CO2 emissions of the PEM-FC system. Figure 15 (b) shows the result of calculating based on the power-generation efficiency and reformer efficiency in Figure 15 (a). At the time of the hydrogen supply to PEM-FC stack, the amount of CO2 discharged by a reforming reaction is expressed with GR, FC ,t , and the quantity discharged by a heat-source burner is expressed with GS , FC ,t . Therefore, the amount GFC ,t discharged by the generation of PEM-FC system is calculated by Eq. (8).
FC ,t =
GFC ,t EFC ,t 100 (QR, FC ,t + QS , FC ,t ) = GR , FC ,t + GS , FC ,t

of CO2

(7) (8)

As Figure 14 (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 15 (b)). From the difference in CO2 emission characteristics, NEG is advantageous in the operation of a partial load, and PEM-FC system is advantageous in steady operation at high load.

2.3. Case Study (1) Urban Area Model The urban area model analyzed in this study is shown in Figure 16. The house number is shown in this figure and the application for each house is shown in Table 5. The number of houses of an urban area model is 20. The urban area model can consider various patterns. This study examines the characteristics of the carbon dioxide emissions of the compound grid of NEG and PEM-FC system from the case of Figure 16.

Figure 16. Urban area model.

(2) Power Demand Model Figure 17 shows the power demand model of each house in Tokyo, and is the mean power load of each sampling time of a representative day in January (winter), May (midterm), and August (summer) [20-22, 26].

152

Shinya Obara

Figure 17. Power demand models.

Fuel Cell Combined Cycle Power Generation System Installed into Micro-Grid

153

Figure 18. Heat demand models in August.

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 17 (a) to 17 (d) is difficult. On the other hand, the power demand of the small offices of Figure 17 (g) and the factories of Figure 17 (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 17 (e)) have the stable demand, and convenience stores (Figure 17 (f)) have 24-hour power demand. In small offices (Figure 17 (g)), factories (Figure 17 (h)), and small hospitals (Figure 17 (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. Figure18 shows the heat demand model in August of each house described in the top [20-22, 26]. However, in a convenience store, office, and a factory, because heat pump is introduced, heat demand is not taken into consideration.

(3) Analysis Flow The analysis flow of the centralized system, base load-sharing system, and IMPE system is shown in Figure 19. First, the power demand data of each house are given to the analysis program, and the base load of the whole micro-grid is calculated. Next, the power plant capacity installed into a micro-grid 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 13 (d) or Figure 15, 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 14 or 15.

2.4. Analysis Results (1) Power Load of Micro-grid Figure 20 (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 20 (a) shows that the power plant capacity of a fluctuating load is 100 kW, and the power plant capacity of

154

Shinya Obara

a base load is 66 kW. Figure 20 (b) shows the result of the rate of a base load and a fluctuating load. The base load is 1.32 times larger. Figure 20 (c) shows the composition of the power demand added to a micro-grid. The largest load component is convenience stores (two houses), which takes 36% of the whole load. Because there is 24 hour power demand in convenience stores (Figure 17 (f)) and hotels (Figure 17 (e)), it is a component that smoothes the whole load added to the micro-grid. 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 micro-grid is made to increase.

Figure 19. Calculation flow.

Figure 20. Results of the load pattern in May reprentative day.

(2) Capacity of Power Plant The analysis results of a representative day in August are shown in Figure 21. Figure 21 (a) shows the results of the capacity of the power plant installed in each micro-grid system. to in Figure 21 expresses the power supply method described in the figure. In a centralized system ( and ), one set of 166 kW power plant is connected to a micro-grid. 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.

Fuel Cell Combined Cycle Power Generation System Installed into Micro-Grid

155

(3) Power Generation Efficiency Figure 21 (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 21 (c) shows the distribution of the power-generation efficiency of the system of to except for the centralized system. In FC base-load operation, it can operate at the maximum powergeneration efficiency shown in Figure 15 (a). The maximum power-generation efficiency of PEM-FC system is higher than the efficiency of NEG shown in Figure 13 (d). Therefore, the total power generation efficiency of the system of , , and using FC to base-load operation is high. (4) Carbon Dioxide Emissions Figure 21 (d) shows the analysis result of the amount of carbon dioxide discharged from each system. and have few CO2 emissions and these are PEM-FC system base load operations. Moreover, (NEG base load and FC load fluctuation operation) also has few CO2 emissions. When Figure 14 (b) is compared with Figure 15 (b), the change in CO2 emissions to change of a load factor has NEG larger than PEM-FC system. As Figures 14 (b) and 15 (b) showed, when the load factor of PEM-FC 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 PEM-FC system (100kW), there are few CO2 emissions than the system composed only from NEG ( and ). ' in Figure 21 (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 power-generation 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 study. 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 kg-CO2/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. 21 (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.

156

Shinya Obara

Figure 21. Analysis results.

Figure 22. Heat demand and exhaust output in August representative day.

Fuel Cell Combined Cycle Power Generation System Installed into Micro-Grid

157

(5) Heat Demand and Exhaust Heat Output Figure 22 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 16, as shown in Figure 22. When an exhaust heat network is introduced into a microgrid and exhaust heat is distributed to each house, the boiler shown in Figure 12 will become unnecessary.

SECTION 3
Dynamic Characteristics of PEM-FC / Woody Biomass Engine Hybrid Micro-grid
3.1. System Scheme (1) Hybrid Micro-grid Figure 23 shows the model of the independent micro-grid that introduces two-set PWHC (PWHC (1) in House (1), and PWHC (2) in House (5)). The micro-grid of this model consists of eight houses 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 house 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 house through hot water pipings (1) and (2). However, examination of this study is limited to the dynamic characteristics of the power for the micro-grid, and is not managed about the heat system.

Figure 23. Independent hybrid micro grid model with PEM-FC and woody biomass engine.

158

Shinya Obara

Figure 24 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 PEM-FC, 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 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 PEM-FC into an alternating current of constant frequency.

Figure 24. PWHC power supply system.

(2) Outline of Testing SEG Table 6 and Table 7 are the operating conditions and specifications of SEG and the power generator that are examined in this study. 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 25 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 7 by a belt from the power shaft of SEG. Since the test SEG is a single cylinder, its vibration is large. 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 Qh, Ex and cooling water Qh, Ey 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 ( Qh, Ez ) is measured by heat flow rate sensor q , and the heatmedium pressure is measured using sensor Pga .

Fuel Cell Combined Cycle Power Generation System Installed into Micro-Grid Table 6. SEG specification

159

Table 7. Power generator specification

Figure 25. Test woody biomass engine (SEG).

Figure 26. Examination results of SEG energy flow.

160

Shinya Obara

Figure 26 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 hightemperature exhaust gas of SEG, for example, is possible. Auxiliary machinery loss holds very few parts in the whole energy flow.

(3) Micro-grid System Operating Method Figure 27 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, PEM-FC is operated in a larger load range than the base load. When a load exceeds the base load, SEG 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.
Table 8. Transfer function of a power output

Figure 27. PWHC operation model.

Fuel Cell Combined Cycle Power Generation System Installed into Micro-Grid

161

3.2. Control Rsponse Characteristics of PEM-FC and SEG (1) Control Block Diagram Figures 28 (a) and (b) are the block diagram of the feedback control on the micro-grid by SEG and PEM-FC, respectively. Proportional-plus-integral control (PI control) is introduced into control of each system. PEM-FC 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 PEM-FC is supplied to the demand side through an inverter and a system interconnection device. The transfer functions of each equipment shown in Figures 28 (a) and (b) describes the determination method in following sections. The control block diagram in the case of one-set SEG operating corresponding to a base load, and corresponding to the load exceeding the base load by one-set PEM-FC is Figure 28 (c). SEG supplies the power to the load below the base load set up beforehand. PEM-FC is also operated when the load of the micro-grid exceeds the base load. The control block diagram in the case of one-set PEM-FC operating corresponding to the base load, and corresponding to the load exceeding the base load by one-set SEG is Figure 28 (d). The control block diagram of the PWHC micro-grid 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 28 (e). In Figure 28 (e), SEG (1) operates corresponding to the base load, and operates SEG (2), PEM-FC (1), and PEM-FC (2) according to the magnitude of a load. When supplying the power to the micro-grid from the combined cycle system, the dynamic characteristics of the micro-grid are decided with the transfer functions of each equipment. So, in this study, the transfer function and control parameters of PEM-FC, SEG, an inverter, and a system interconnection device in Figure 28 are determined by the method described to following sections. The transient response characteristics of the power output of the SEG, PEM-FC, auxiliary machine and PWHC micro-grid 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 PEM-FC Table 8 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 [5, 27]. 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 [16, 17, 28-34]. 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 PEM-FC. The transfer function and control parameters on PEM-FC of the control block diagram shown in Figure 28 introduce each value of Table 8. Figure 29 shows the result of the step response of 0.2kW, 0.4kW, 0.6kW, 0.8kW, 1.0kW of PEM-FC with a reformer [5, 27]. In the analysis of Figure 29, the control block diagram of Figure 28 (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.

162

Shinya Obara

Figure 28. Control block diagramo f power supply.

(3) Response Characteristics of SEG Figure 30 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 30, the step response of Testing SEG has large overshooting, and its settling time is long compared with PEM-FC. The heat transmission characteristics between the combustion gas of a chip and the heat exchanger of

Fuel Cell Combined Cycle Power Generation System Installed into Micro-Grid

163

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 31 shows the example as a result of a step response obtained in the operating experiment of SEG (Figure 26). The model of the transfer function that simulated this step response is shown in Figure 31. 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 32 shows the analysis result of a step response when adding PI control to the system using the transfer function in Figure 31. The control block diagram used in this analysis is Figure 28 (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 32. 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.

Figure 29. PI control step response result of PEM-FC with a reformer (P=12.0, I=1.0).

Figure 30. Step response result of SEG when not adding PI control.

164

Shinya Obara

Figure 31. Step response result of the test SEG, and response model.

Figure 32. Step response results of 2 kW SEG model.

3.3. Result of Dynamic Characteristics Analysis of PWHC Micro-grid (1) Power Response Characteristic of PWHC 1 kW PWHC micro-grid consists of 0.5 kW SEG and 0.5 kW PEM-FC. Figure 33 shows the analysis result of the step response of 0.2kW, 0.4kW, 0.6kW, 0.8kW, 1.0kW of this system. Results in case SEG corresponds to the base load and PEM-FC follows the fluctuating load exceeding the base load are Figures 33 (a) and (b). The control block diagram used in the analysis in Figures 33 (a) and (b) is Figure 28 (c). However, the value in Figure 29 ( P = 12.0 and I = 1.0 ) and Figure 32 ( P = 0.1 and I = 0.001 ) was used for the control parameter of the analysis in Figure 33 (a). The speed of response of PEM-FC shown in Figure 29 is quick compared with the speed of response of SEG shown in Figure 32. From the difference in this speed of response, as shown in the step response of 0.8kW and 1.0kW in Fig. 33 (a), the response of a quick response part and a late response part appears. Consequently, the control parameters of PEM-FC with a quick speed of response are changed, and an improvement of the response characteristics of the PWHC micro-grid is tried. Figure 33 (b) shows the response characteristics at the time of changing the control parameters of PEM-FC

Fuel Cell Combined Cycle Power Generation System Installed into Micro-Grid

165

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 33 (a), have improved.

Figure 33. Step response result of 1 kW PWHC.

Step response results in case PEM-FC corresponds to the base load and SEG follows the fluctuating load exceeding the base load are Figures 33 (c) and (d). In the analysis in Figures 33 (c) and (d), the control block diagram shown in Figure 28 (d) was used. In Figure 33 (c), the control parameters of PEM-FC 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 Fig. 33 (c) is large, the control parameters of PEM-FC is changed and an improvement is tried. Figure 33 (d) shows the response characteristics at the time of changing the control parameters of PEMFC into P = 0.95 and I = 1.1 . These control parameters were decided by trial and error. Compared with the response of Figure 33 (c), the response of Figure 33 (d) has small overshooting, and its settling time is short.

166

Shinya Obara

(2) Response Characteristics of SEG and PEM-FC Micro-grid Using Power Load Pattern for Houses a. Response Result of SEG The response characteristics in the case of supplying the power to the micro-grid from SEG or PEM-FC are investigated. However, the power load pattern added to the micro-grid 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 [35]. 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 34 shows the analysis result of a load response at the time of supplying the power to the micro-grid using 2 kW SEG. The control block diagram used in the analysis of Figure 34 is Figure 28 (a). Moreover, the control parameters of SEG are P = 0.1 and I = 0.001 as well as Figure 32. The horizontal axis of Figure 34 is the representative time of analysis. Real time is also displayed on the horizontal axis of Figure 34. Since the calculation time is enormously long, real-time analysis is performed by shortening real time to 1/180 in this study. Figure 34 (a) shows the result of a load input and the system response, and Figure 34 (b) shows the result of the error of a load input and a response. As for the broken-line part shown in Figure 34 (b), the error of the load and the response is over 5%. A large rising error occurs immediately after 0:00 in Figure 34 (b).

Figure 34. Dynamic-characteristics analysis result of the micro-grida 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.

Actually, since the system is operated continuously, this rising error does not exist. Figure 34 (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 34 (c) express settling times. The settling time when installing SEG into the micro-grid from the result of Figure 34 (c) is 10.2s at the maximum. When the micro-grid is composed from SEG, the unstable period of voltage and a frequency is 10.2s at the maximum.

Fuel Cell Combined Cycle Power Generation System Installed into Micro-Grid

167

b. Response Result of PEM-FC Figure 35 shows the analysis result of a load response at the time of installing 2 kW PEM-FC into the micro-grid. The control block diagram used in the analysis of Figure 35 is Figure 28 (b). The control parameters set up with the controller are P = 12.0 and I = 1.0 as well as Figure 29. The settling time in case PEM-FC composes the micro-grid from the result of Figure 35 (c) is 1.6 s or less. However, rising parts are excluded. The power supply due to PEM-FC has a short settling time compared with SEG. Therefore, the dynamic characteristic of the power of PEM-FC micro-grid is good compared with SEG micro-grid.

Figure 35. Dynamic-characteristics analysis result of the micro-grida at the time of installing the power demand model of two houses in Sapporo. The power is supplied to the grid from 2 kW PEM-FC of one setPEM-FC of P=12.0 and I-1.0.

(3) Response Characteristics of PWHC Micro-grid Using Power Load Pattern for Houses Figure 36 shows the analysis result of a load response of the micro-grid composed from 8 kW PWHC. Eight houses are connected to the micro-grid. Woody biomass engine generators installed into the micro-grid are 2 kW SEG (1) and 2 kW SEG (2), in addition install 2 kW PEM-FC (1) and 2 kW PEM-FC (2). Moreover, the control block diagram used in the analysis of Figure 36 is Figure 28 (e). The control parameters set up with the controller of PEM-FC are P = 0.95 and I = 1.1 as well as Figure 33 (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 micro-grid. It is because SEG (2) is followed and operated to load fluctuations. As a result, a settling time becomes long as shown in Figure 36 (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 PEM-FC. Figure 37 shows the analysis results of the load response of the micro-grid composed from one-set of 2 kW SEG and two-set of 2.5 kW PEM-FC. Eight houses are connected to the micro-grid. This system was analyzed by modifying the control block shown in Figure 28 (e). The control

168

Shinya Obara

parameters set up with the controller of PEM-FC are P = 12.0 and I = 1.0 , and SEG are P = 0.1 and I = 0.001 . The error-analysis result shown in Figures 36 (b) and 37 (b) shows the well alike profile. However, as shown in Figure 37 (c), the settling time of the micro-grid becomes very short compared with Figure 36 (c). The system of Figure 37 is the PWHC micro-grid stabilized dynamically.

Figure 36. Dynamic-characteristics analysis result of the micro-grida 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 two sets and 2.5 kW PEM-FC of two sets. PEM-FC of P=12.0 and I=1.0., and SEG of P=0.1 and I=0.001.

Figure 37. Dynamic-characteristics analysis result of the micro-grida 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 set and 2 kW PEM-FC of two sets. PEM-FC of P=0.95 and I=1.1., and SEG of P=0.1 and I=0.001.

Fuel Cell Combined Cycle Power Generation System Installed into Micro-Grid

169

ACKNOWLEDGMENTS
This work was partially supported by a Grant-in-Aid for Scientific Research (C) from JSPS.KAKENHI (17510078).

REFERENCES
[1] [2] [3] Robert, H. et al. Micro-grid: A Conceptual Solution. Proceedings of the 35th Annual IEEE Power Electronics Specialists Conference, 2004; 6: 4285-4290. Carlos, A. and Hernandez, A. Fuel Consumption Minimization of a Micro-grid. IEEE Transactions on Industry Applications, 2005; 41(3): 673- 681. Takuma, Y. and Goda, T. Micro-grid for Urban Energy. Transaction of the Society of Heating, Air-Conditioning and Sanitary Eng. of Japan, 2005; 79(7): 573- 579 (in Japanese). Obara, S. and Kudo. K. 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, 2005; 2: 202-207. Obara, S. Load Response Characteristics of a Fuel Cell Micro-grid with Control of Number of Units, International Journal of Hydrogen Energy (in press). Mohammadi Ali.. et al., Development of Highly Efficient and Clean Engine System using Natural-Gas and Hydrogen Mixture Fuel Obtained from Onboard Reforming, NEDO report ID:03B71006c, (2005). (in Japanese) Yap D., et al., Effect of Hydrogen Addition on Natural Gas HCCI Combustion, SAE Tech Pap Ser, SAE-2004-01-1972(2004). Larsen J. F. and Wallace J. S., Comparison of Emissions and Efficiency of a Turbocharged Lean-Burn Natural Gas and Hythane Fuelled Engine, ASME ICE, Vol.24(1995), pp. 31-40. Swain M. R. and Usuf M. J., The Effects of Hydrogen Addition on Natural Gas Engine Operation, SAE Tech. Pap Ser., SAE-932775(1993). A.D. Peacock and M. Newborough, "Controlling micro-CHP Systems to Modulate Electrical Load Profiles", Energy (2006), in press. A.D. Hawkes and M.A. Leach, "Cost-effective Operating Strategy for Residential Micro-Combined Heat and Power", Energy (2006), in press. J. De Ruyck et al., "Co-utilization of Biomass and Natural Gas in Combined Cycles Through Primary Steam Reforming of the Natural Gas", Energy (2006), in press. S. Bonnet, M. Alaphilippe and P. Stouffs, "Energy, Exergy and Cost Analysis of a Micro-Cogeneration System Based on an Ericsson Engine", International Journal of Thermal Sciences, 44 (2005), 11611168. H. Ozeki, "Research on a combustor for Stirling engines. 4. Effects of a combustor engine operation parameter on exhaust characteristics", Proc. of the 29th Japanese Symp. on Combustion, (1991), pp.208-210.

[4]

[5] [6]

[7] [8]

[9] [10] [11] [12] [13]

[14]

170

Shinya Obara

[15] K. Eguchi and Y. Kurosawa, "Study on high load swirl burners for Stirling engines. Reduction in NOx and heat adsorption characteristics in exhaust gas recirculation", Proc. of the 26th Japanese Symp. on Combustion, (1988), pp.131-133. [16] Takeda, Y. et al. Development of Fuel Processor for Rapid Start-up. Proc. 20th Energy System Economic and Environment Conference, 2004; Tokyo, January 29-30, ed., K. Kimura: 343-344 (in Japanese). [17] Ibe, S. et al. Development of Fuel Processor for Residential Fuel Cell Cogeneration System. Proc. 21st Annual Meeting of Japan Society of Energy and Resources, 2002; Osaka, June 12-13, ed., K. Abe: 493-496 (in Japanese) [18] Mikkola, M., Experimental Studies on Polymer Electrolyte Membrane Fuel Cell Stacks, Master's thesis submitted in partial fulfillment of the requirements for the degree of Master of Science in Technology, Helsinki University of Technology, (2001), pp.58-79. [19] Ibaraki Prefecture Government Office of Education, Modeling of hydrogen energy system, High school active science project research report, Ibaraki, Japan, 2002 (in Japanese) [20] Architectural Institute of Japan, Nationwide research study concerning energy consumption in the house in fiscal year 2001, 2002; 3: 3-6 (in Japanese) [21] Nagase, O. et al. Quantification of the energy consumed in a house. Proc. 19th Energy System Economic and Environment Conference, 2003; 461-466 (in Japanese) [22] Hatano, Y. et al. Investigation on the energy use characteristics in the apartment house of hot water supply and heating of an area, Proceedings of the Meeting of the Society of Heating, Air-Conditioning and Sanitary Engineering, 2003; 1745-1748 (in Japanese) [23] National Astronomical Observatory, Rika Nenpyo. Chronological Scientific Tables CD-ROM, Maruzen Co., Ltd., 2003. [24] Mohammadi Ali.. et al., Development of Highly Efficient and Clean Engine System using Natural-Gas and Hydrogen Mixture Fuel Obtained from Onboard Reforming, NEDO report ID:03B71006c, (2005). (in Japanese) [25] Yasuda I., Development of Hydrogen Production Technology for Fuel Cell, Energy Synthesis Engineering, Vol28 No.2 (2005) (in Japanese) [26] Yoshinaga, M. et al., Investigation on the energy consumption of a house, and a resident's consciousness, Proceedings of the Meeting of the Society of Heating, AirConditioning and Sanitary Engineering, (2003), pp.1729-1732. (in Japanese). [27] Shin'ya Obara and K. Kudo, "Installation Planning of Small-scale Fuel Cell Cogeneration Considering Transient Response Characteristics (Load Response Characteristics of Electric Power Output)", Journal of Environment and Engineering, (2006), in press. [28] Y. Zhang, M. Ouyang, Q. Lu, J. Luo and X. Li, A Model Predicting Performance of Proton Exchange Membrane Fuel Cell Stack Thermal Systems", Applied Thermal Engineering, 24(2004), pp. 501-513. [29] K. Sedghisigarchi, 2004, Dynamic and Transient Analysis of Power Distribution System with Fuel Cells-Part 1: Fuel-Cell Dynamic Model, IEEE Tran. Energy Conversion, Vol. 19, No. 2(2004), pp. 423-428. [30] J. Hamelin, K. Agbossou, A. Laperriere, F. Laurencelle and T. K. Bose, Dynamic Behavior of a PEM Fuel Cell Stack for Stationary Applications", Int. J. Hydrogen Energy, Vol. 26(2001), pp. 625-629.

Fuel Cell Combined Cycle Power Generation System Installed into Micro-Grid

171

[31] S. Yerramalla, A. Davari, A. Feliachi and T. Biswas, Modeling and Simulation of the Dynamic Behavior of a Polymer Electrolyte Membrane Fuel Cell", J. Power Sources, Vol. 124(2003), pp.104-113. [32] K. Oda, S. Sakamoto, M. Ueda, A. Fuji and T. Ouki, A Small-Scale Reformer for Fuel Cell Application", Sanyo Technical Review, Vol. 31, No. 2(1999), pp.99-106, Sanyo Electric Co., Ltd., Tokyo, Japan.(in Japanese). [33] S. Koike, H. Inaka, T. Ohmura, K. Hirai, and K. Kishi, Demonstration Program of 1kW-class PEMFC System for Residential Use by the Japan Gas Association", Proc. 3rd Int. Fuel Cell Conference, (1999), pp.497-498. [34] Kyoto Denkiki Co., Ltd. A System Connection Inverter Catalog and an examination Data Sheet. 2001". [35] K. Narita, "The Research on Unused Energy of the Cold Region City and Utilization for the District Heat and Cooling, Ph.D. thesis, (1996), Hokkaido University. (in Japanese) [36] Shin'ya Obara, "Fundamental Characteristics of PEMFC-Stirling Engine Combined Cycle for Apartment House", Transactions of the ASME, Journal of Energy Resources Technology, under review.

In: Leading-Edge Electric Power Research Editor: C. M. OSullivan, pp. 173-204

ISBN: 978-1-60021-981-8 2008 Nova Science Publishers, Inc.

Chapter 3

ELECTRICITY FROM RENEWABLE ENERGY SOURCES: A MULTI-CRITERIA EVALUATION FRAMEWORK OF TECHNOLOGIES


Fausto Cavallaro
Department S.E.G.e S, University of Molise, Via De Sanctis, 86100 Campobasso, Italy

The energy policy of many Western governments aims to diversify supply and reduce dependence on foreign sources and thus to maximise benefits from internal resources. Undoubtedly, the main strategy underlying this is one that seeks to optimise the use of renewable energy sources (RES). The development of these sources, as well as their market penetration, depends however not only on political will but also on sound management of energy demand in order to rationalise and stabilise energy consumption. In addition to fortifying the guaranteed energy supply, RES represent a potential that cannot be overlooked. This lies in their ability to reduce greenhouse gas emissions and thus to stem the growing trend of global warming, one which has accelerated particularly in recent years and which is due mainly to the use of fossil fuels for producing electricity. The use of RES for the production of electric power brings huge benefits both in terms of environmental protection as well as savings in non-renewable resources. Nevertheless, the very nature of RES raise technical and economic problems that create a considerable gap between their potential capacity and ways to feasibly exploit them. Their many different forms and the ways in which they may be used have to be carefully examined in order to evaluate the costs and other technical and environmental factors involved. The planning and appraisal of sustainable energy projects involve rather complex tasks. This is due to the fact that the decision making process is the closing link in the process of analysing and handling different types of information: environmental, technical, economic and social. Such information can play a strategic role in steering the decision maker towards one choice instead of another. Some of these variables (technical and economic) can be handled fairly easily by numerical models whilst others, particularly ones relating to

Tel. +39-0874-4041 fax. +39-0874-311124. E-mail: cavallaro@unimol.it

174

Fausto Cavallaro

environmental impacts, may only be adjudicated qualitatively. In many cases therefore, traditional evaluation methods and the chief economic and financial indicators are unable to deal with all the components involved in an environmentally valid energy project. Multicriteria methods provide a flexible tool that is able to handle and bring together a wide range of variables appraised in different ways and thus offer valid assistance to the decision maker in mapping out the problem. This manuscripts sets out the application of a multi-criteria method to make a preliminary assessment of technologies. As this work demonstrates, multi-criteria analysis can provide a technical-scientific decision making support tool that is able to justify its choices clearly and consistently, especially in the renewable energy sector.

1. THE ELECTRICITY MARKET AND RENEWABLE ENERGY SOURCES: INTRODUCTORY CONSIDERATIONS


Energy and the supply of energy sources have played a central role in the development of modern society. The technological revolution of the last century would not have been achieved without the invention and the rapid expansion of systems for electricity distribution. Up until the energy crises of the 1970s and early 1980s, satisfying energy demand was basically a question of the availability of resources and the best technology on hand. The last 20-30 years, however, have seen a change in the way of interpreting the idea of availability and of energy supply. The main factor that triggered this change is tied to the sharp rise in the price of energy caused by the first oil crisis in the seventies. In the west, that era heralded the collapse of the myth of cheap, plentiful and easily available energy and raised in its place concerns about the imminent exhaustion of natural resources. At the same time, the worries linked to the environmental consequences deriving from an increasingly greater use of hydrocarbons led to the search for energy technologies that were environmentally compatible. The 1990s represented the phase in which electricity markets and national energy sectors came to be decentralised and liberalised and which has now led to significant changes in the ownership of energy systems. Energy markets were liberalised on the basis of the principle that market mechanisms would have provided an opportunity to balance social and economic benefits and would have increased the efficiency of operations relating to energy supply. Energy grids, of which electricity and gas supply constitute at least 90% of the total, have been State regulated in all countries, albeit to a greater or lesser extent and in a variety of ways. The State intervenes by means of price controls and also by imposing rules controlling the behaviour of service providers as well as by erecting entry barriers to restrict access to this sector. Regulation and control of the sector has always been justified by citing public interest and this is the reason it has affected public utilities. The strategic importance these sectors occupy in the overall functioning of the economy involve above all basic services to which access must be guaranteed to all consumers on equal terms. Over recent years, the traditional theory of a natural monopoly has been attacked and thus revised. Several authors sustain that in the electricity sector in particular a natural monopoly exists in the stages of transmission and distribution but not the phase regarding power generation where there is nothing to restrict healthy and free competition between competing

Electricity from Renewable Energy Sources

175

producers The goal of free competition has been sought by deregulating specific stages of the production chain. Recently, the analysis and appraisal of electricity power markets of some countries have shown that the targets set to improve the efficiency of the market have not been fully met. The lack of clear and adequate regulations may have hindered the optimal development of energy systems and instead favoured a lowering of the safety standards applied to energy supply . There are in fact technical limits to the scope of liberalisation that in practice make it impracticable to indiscriminately introduce competition. Electrical power, the core-product of energy markets, is a product with unique characteristics that set it apart from other types of commodities, in particular: it cannot be accumulated, electricity is still a good that must be consumed as it is provided; imposing the need to ensure a balance between demand and supply at all times. This imposes the need to have an ancillary service that is capable of acting as a safety valve. Such a service is guaranteed in real time by a System Operator that acquires from generating companies a quota of production as a reserve in order to adjust (increasing or reducing) the amount of power supplied; demand is subject to variations that are temporal and stochastic; the energy produced in the short term can be carried only by the existing distribution grid, a great deal of time and money are required to construct new lines.

Independently of the type of market operating (monopoly or competitive), the technical limitations makes the presence of a System Operator imperative in order to guarantee the stability of the entire system and to resolve congestion on the grid. The current energy system comprises colossal power stations, huge storage depots and refineries and a large network of power lines. Targeting renewable sources entails making profound changes to the current setup of the energy industry; to move towards a system that is increasingly more geographically scattered, technologically advanced and able to handle power generation and demand spread over a wide geographical area. The desired system would be one that: reduces the energy production chain, creates electricity and power directly from the sun and wind, and would gradually allow small users to become increasingly self-sufficient and thus become less dependent on large installations generating and distributing energy. The challenge lies in getting environmental and energy objectives to converge and the overall success of future energy policy will be to demonstrate that economic growth, an assured energy supply and environmental protection are compatible goals. Although some technologies exploiting renewable energy sources (RES) have reached a certain maturity, there are numerous hurdles to their market penetration. It is fundamental to kick-start the launch of RES in order to accelerate and increase their market share. This strategy would favour the creation of economies of scale and consequently reduce costs. Now, the intense attention directed towards the environment has prioritised those RES that would have a minimal impact not only on the environment, but also on health and the quality of life. Therefore, this growing awareness of environmental issues has partially modified the traditional decision making structure in the energy field. Indeed, the need to

176

Fausto Cavallaro

incorporate into energy planning considerations that are strictly environmentally related has resulted in the adoption of multi-criteria decision models. In order to face the very real threat of climate change, the European Commission has set a strategic target for energy policy: by 2020 to reduce greenhouse gas emissions by at least 20 % compared to 1990 levels, in a way that is compatible with competitiveness objectives. To promote safety and sustainability the European energy system must take action on four main fronts [1]: the conversion and efficient use of energy in all sectors of the economy associated with a fall in energy intensity; diversification of the energy mix towards renewable energy sources and technologies for energy conversion with low carbon emissions for electricity, heating and cooling; decarbonisation of transportation by shifting to alternative fuels; the complete liberalisation and interconnection of energy systems using smart information and communication technologies to provide a flexible and interactive (customers/operators) service network..

Technology will play a central role in achieving the goals of the new energy policy for Europe. The Commission forecasts an annual investment of around 1 billion Euros over the period 2007 - 2013 in research and technological innovation that would allow costs of renewable energy sources to be reduced, energy efficiency to be increased and would ensure that European industry took the lead the world over. In 2007, under the European energy policy framework, the Commission intends to devise the first European strategic plan for energy technologies with the underlying objective of speeding up innovation in the energy technology field and thus to motivate European industry to transform the risks coming from climate change and the assurance of supply into opportunities to increase competitiveness. The considerable attention now being paid to the environment has therefore given pride of place to the renewable energy sources that are capable of minimising impacts on the environment, health and quality of life. The need, therefore, to include considerations of a purely environmental nature in energy planning has to some extent modified the traditional decision making framework widely adopted by the energy sector and has instead promoted the development and adoption of Decision support systems (DSS) that use multicriteria algorithms. The use of decision making tools, referred to in the literature as Decision Support Systems (DSS), for resolving environmental problems is wide-ranging. DSS based on multicriteria algorithms do not replace decision makers, rather they assist them in all the phases of the decision making process by supplying useful information to reach decisions that are transparent with a clearly documented trail. Various studies have been developed to illustrate the potential applications of this approach: for the evaluation of energy options when compared to a set of criteria and in order to make the choices clearer [2] [3] [4] [5] [6] [7], for the assessment of geothermal energy projects [8]; for concentrated solar power [9], for the siting of power plants [10]; for the evaluation of energy strategies for small islands [11] [12], for a review of application of multicriteria decision making to energy planning [13] [14], [15] and [16].

Electricity from Renewable Energy Sources

177

2. DECISION MAKING IN ENERGY AND ENVIRONMENTAL SECTORS


2.1. The Nature of a Decision
Decisions are taken in all spheres of human activity. These decisions can thus be analysed using different approaches and disciplines from the study of the behaviour of the individual to the economy and to management. The three main paradigms in decision theory are as follows [17]: Decisions under conditions of certainty (perfect information). This requires the existence of someone (decision maker) who must take decisions that are wholly consistent with one another, who has a clear and precise idea of the objectives set beforehand, and who believes that there is one action of all the possible solutions that is better than all the rest. The task of decision theory is thus to discover this solution by means of specific mathematical models. Decisions under conditions of uncertainty. In this case the decision maker is not able to produce all the information necessary to make a strictly rational choice. In such circumstances it is said that the decision maker works under conditions of bounded rationality [18] and the outcome of the decision will therefore depend on the circumstances of which knowledge is imperfect. The comparison of the preferability of the various options is based on the probability of random or unknown circumstances occuring. Multicriteria decisions. Multicriteria analysis has evolved from the previous approaches. It in fact enables choices to be made between a range of solutions where the choice cannot be based on a single criterion.

In many cases the decision may correspond to the verdict, simple or complex, deriving from a phase involving study, analysis and negotiating the preferences of those involved (the players) who work in different ways and to differing degrees to determine the outcome (i.e. the decision). Broadly speaking, the decision is generated by a dynamic and interactive process involving the various players. Nevertheless, the leading role in the decision making process is generally assigned to the decision maker who evaluates the various alternatives and ranks them. Decision-making activities include all the methods the policy maker can use in the phase of selecting projects of a public nature that are affected by issues not related solely to the actions that are closely connected with carrying out the project itself and the direct economic consequences thereof, but also that apply to the entire socio-economic context and the environment. A set of methods belong to this category which are well-established within decision support systems. The most commonly used of these tried and tested methods is costbenefit analysis. Over the years a vast body of literature has flourished in the area of multi criteria analysis about the limitations of this method and the criticisms aimed at it. The ideological base of MCDA contains the notion that all data, all the consequences and all the prospects that a certain behaviour or action will meet and fulful the criteria laid down are made available systematically and accurately.

178

Fausto Cavallaro

2.2. Planning Energy Systems and the Decision Support System (DSS)
The traditional definition of an energy system is of an integrated unit for energy production, transmission and distribution. A broader interpretation describes energy systems as interconnected infrastructures that combine energy sources, the means to convert these into other forms of energy, the apparatus and procedures for its distribution, the community and the way in which energy is used and the natural and economic setting [19]. This more wide-ranging definition includes both the technical and economic aspects of energy infrastructure. On this basis, the planning of an energy system corresponds to a process of choosing the sources and the technologies needed to produce, transmit and distribute energy so as to meet the demand for it [20]. In many cases, planning is regarded as a decisionmaking process; the only difference lying in the fact that, in general, the output of the decision making procedure is a choice, whilst the output of planning is a plan i.e. the description of what must be done, when and by whom. Furthermore, while a decision in principle may have a temporal and spatial dimension and can be identified as the choice of a specific option, the activity of planning is a dynamic and iterative process [20] [21]. Another consideration is that in each country, specific rules, laws and institutional frameworks, decision making mechanisms and the status of different institutions and centres of power can play a strategic role in energy planning. In general, two types of players can be singled out: decision-makers who are generally in charge of the planning process and the stakeholders who, despite not having any direct power to take planning decisions, may take part in negotiations in order to put pressure on the decision maker to take their claims into account. In actual fact, stakeholders can heavily influence the final outcome [20]. Over the last few decades, a number of studies have been carried out to identify useful practices and tools to aid policy makers in setting out energy strategies. Much of this research has involved drawing up possible future scenarios and energy balances together with studies to appraise the potential of renewable sources and to look at the interactions between energy production, the economy and the environment. A knowledge base was thereby built up and refined which was required to organise energy policy strategically, and it was from this that the first Decision Support Systems (DSS) emerged to aid decision makers in their planning. DSS help decision makers not only in operational decisions but also in strategic ones that are longer term and wider in scope. DSS are well suited to dealing with highly structured problems or semi-structured ones (for which it is not possible to produce an ideal solution). DSS help the user in all phases of the decision making process by using relevant information in order to make choices that can be documented and are transparent. These tools, via interactive procedures, provide the decision maker with the information necessary to understand the problem, to explore the data from a variety of perspectives that are based on the needs of the user as well as to evaluate scenarios arising from choices made and to justify what in theory is the best solution. A DSS must also offer the opportunity to create new models or to modify the existing ones but not replace the decision maker: the decision is reached by subjecting the information processed by the system to human assessment. For it not to remain purely a tool for simulation or as an academic exercise it must not supply preestablished truths, rather it must be inserted into the decision making pathways and must be able to accept and handle a system of shared Community values and achieve a consensus that is consistent with an approach based on social and political debate. DSS differ from traditional systems for handling information because they require intense interaction between

Electricity from Renewable Energy Sources

179

the user, or decision maker, and the system and they are also able to use multicriteria methods. These methods analyse a decision problem by comparing a number of alternative actions on the basis of various evaluation criteria (the criteria are the rules and principles used to judge courses of action) and allow rankings of the alternatives to be generated by assigning a score to each of them that is a measure of their utility. These methods are used not so much to identify the optimal solution but to generate the information needed for the decision to be taken, highlighting amongst other things the conflicts between the different groups and individuals involved. The use of this tool allows the items for appraisal to be broadened to include values that by their nature cannot easily be expressed in strictly financial terms.

2.3. Problems and Limitations Linked to the Use of Cost-Benefit Analysis in Environmental Applications
Cost-benefit analysis (CBA) is one of the most well-known methods for choosing between different projects. It is widely used as a financial management tool to justify an investment in economic terms rather than as a planning support tool. CBA is a technique with theoretical roots in the economics of welfare and allows the decision making process to be carried out for projects of social importance. This type of analysis was created in the context of the classic economy and classified as a tool of the neoclassical economy of the 20th century. The rigour of the underlying paradigm has led it to become the main accepted tool for project appraisal. Nowadays, it is used by practically all institutions in the West, with the World Bank in first place, others include the European Ministries of Economy and Transport, the European Investment Bank, the Italian Ministries for Infrastructure and the Environment (it is compulsory to include it in Studies of Environmental Impact Assessment). This technique seeks to enable decision makers to evaluate the advantages and disadvantages of a certain action and thus aims to systematically identify the benefits and the costs produced by a specific sum of expenditure in order to identify which out of a number of options offers the greatest net social advantage. The underlying concept of the theory is to measure the opportunity costs that carrying out a project holds for society (economic CBA) or for a private party (financial CBA). The greater is the surplus generated by carrying out the project compared with not doing so the greater the increase in welfare that society will enjoy and the project is therefore desirable (if there are the resources to do so). Identifying and estimating the gains and losses to society and the decision rules fall to the impartial role of the decision maker. The goal of the latter is to maximise the net social profit (Us) so that the rule of choice is to select projects where Social Benefits (Bj) outweigh Social Costs (Cj) (discounted to present value as necessary) the most:

B j C j = max U s
j =1 j =1

(1)

This surplus is measured by assigning a monetary value to all the types of costs/benefits involved in the project, not merely those requiring paying out sums of money but also indirect

180

Fausto Cavallaro

costs (such as pollution, time, etc.) that in effect represent the consumption of a scarce resource rather than its price. CBA applied to the public sector (energy choices have a social worth) takes into account the entire economic system and the influence that it can exert on the pricing system often becoming a highly complex activity that can even affect the economy as a whole [22]. It accounts for everything to which a monetary value can be assigned. Such an operation will inevitably lead to approximations and arbitrary valuations and may not be easy to perform for all the items considered in the analysis. Considerable difficulties arise when it is desirable to include in the analysis environmental impacts, such as the various forms of pollution or the social impacts on the geographical area affected by the intervention or the project. Generally, these impacts are not easy to translate into monetary terms and, it may be considered unethical to set a price to measure certain values such as biodiversity, human health, quality of life and social factors. Another critical element lies in deciding the discount rate to apply, in fact, the higher the rate the lower the present value of long term benefits. For the decision maker the task of CBA is to act as a guide in the decision making procedure and to provide a meaningful scheme of whether or not to carry out a specific project. When the necessary information is not available or is not reliable enough, the project appraisal is carried out under uncertain conditions. The most direct way to evaluate costs and benefits is to refer to market values, in other words to the demand and supply curves of the abovementioned good (here the environment). However, because of distortions present on the markets, (monopoly and oligopoly) it is possible that some prices do not truly reflect the value (social) attached to some goods. In this case, market prices should be replaced, in theory, by a set of prices called shadow prices or account prices, i.e. the values found in a market free of imbalances and imperfections [23]. Unfortunately, it is extremely difficult to calculate shadow prices. Market imperfections imply deviations from situations of Pareto optimality and thus, in this case,decisions based on costs and benefits valued at market prices may be erroneous. The difficulties attached to evaluation tend to increase when moving to consider incalculable and intangible benefits i.e. ones that do not have a market price. Benefits and costs with no market prices are normally valued on the basis of estimate of individuals WTP (willingness to pay) or by WTA (willingness to accept); however, these techniques are often flawed which discourages their use. A criticism often launched at CBA is that it does not take into account certain benefits as they are so difficult to measure. The analysis seems highly imbalanced towards factors that are purely quantitative/monetary and is therefore probably little suited to project appraisal of energy and environment projects. The unrealistic notion that everything has a monetary value leads inevitably to errors in project assessment. The adoption of a single unit of measure, money, does not manage to be truly an aid to decision making in contexts in which it is necessary to take qualitative factors into consideration.

Electricity from Renewable Energy Sources

181

3. MULTI-CRITERIA ASSESSMENT AIDS TO ENVIRONMENTAL AND ENERGY DECISIONS


In the decision making process, decision makers often make great efforts to find the optimal solution. Unfortunately, in reality, an ideal solution exists if only a single criterion is considered, thus this is a totally inadequate approach in the majority of situations requiring decisions to be taken. The multi-criteria approach differs substantially from CBA in that the overall merit of the project is evaluated by considering it from differing viewpoints or by applying heterogeneous criteria. The impacts produced by the proposals under review are estimated in respect of each criteria and, unlike CBA, these need not necessarily be expressed in monetary terms but may be either quantitative or qualitative values measured using a range of different scales. The choice is made by assessing the contributions made by the various project options and comparing them with the overall objective considered from diverse, and often conflicting, standpoints. From the above arises the need to develop a planning and management tool that can assist the decision maker in assessing a set of alternatives from different viewpoints and to choose the option of compromise, namely the one held to be most acceptable by all the criteria considered altogether. The activity linked to the search for a best compromise solution requires a suitable assessment method and the various multi-criteria methods available seem best suited to such a purpose. The multi-criteria decision process is shown in Figure 1.

Fomulation of options Selection of decision process Performance evaluation

Selection of criteria

Decide decisions parameters

Application of the method

Evaluation of results

Decision

Figure 1. Multicriteria decision procedure: Source [13].

182

Fausto Cavallaro

Buchanan et al. (1998) [24] and [25] have argued that good decisions will typically come from a good decision process and suggest that where possible the subjective and objective parts of the decision process should be separated. A decision problem can be conceived as comprising two components; a set of objectively defined alternatives and a set of subjectively defined criteria. The relationship between the alternatives and the criteria is described using attributes which describe, as objectively as possible, the features of alternatives that are relevant to the decision problem. Each criterion attempts to reflect a decision makers preference with respect to a certain feature of the decision problem. These preferences, being specific to a decision maker, are subjective. As stated earlier, in MCDA the decision making process consists normally in making a choice between different elements examined by the decision maker and evaluated using a set of criteria. The elements are called actions and make up part of a set of actions or alternatives. The final solution according to Roy is a creation rather than a discovery [26] and [27]. Thus the main objective of a Multiple Criteria Decision Aid (MCDA) is to build or create a support tool for decision makers that conforms to their objectives and priorities (a constructive or creative approach) [27]. The ideal solution, the option that performs best for all the criteria selected, is difficult to achieve. Therefore it is necessary instead to find a compromise from among the different hypothetical solutions. It is for this reason that a choice resulting from MCDA is justified and not optimum. The multicriterion aggregation procedure (MCAP) is the heart of the MCDA. A flowchart of the procedure is illustrated in Figure 2.

Figure 2. Scheme of a MCDA procedure: source [28].

The points in favour of a decision making model built on a multi-criteria algorithm are summarised below: it can handle the large amounts of, often conflicting, information, data, relations and objectives that are generally encountered when facing a specific decision problem; it does not unveil the solution to the decision maker as a revealed truth, instead it sustains the entire decision making process providing the means to deal with the information to hand; the approach is based on systematic observation and on the verification of factors influencing the decision, thus it is not a black box type of decision model but a transparent tool;

Electricity from Renewable Energy Sources

183

it provides the instruments to construct the problems clearly in order to make them more understandable; it enables the decision making process to be monitored and checked as it evolves.

3.1. The Promethee Method


The above highlights why there is a need to develop a planning and management tool that can assist the decision maker. The methodology adopted for the purpose of this case study is based on the method of outranking called PROMETHEE (Preference Ranking Organization Method of Enrichment Evaluation) devised by Brans J.P.et al. [29] [30] [31] [32]. This technique, besides possessing all the advantages of B. Roys outranking methods, is also easy to use and its level of complexity is low. It is based on ranking and is well-suited to problems in which there are a finite number of actions to be assessed on the basis of a range of conflicting criteria. The following procedure is recommended to implement the method: Identification of alternatives Under MCDA the decision procedure is normally carried out by choosing between different elements that the decision maker has to examine and to assess them using a set of criteria. These elements are called actions and they make up part of a global set labelled actions or alternatives A = {a1 K, ai ,K, a m }; Defining a set of criteria. The criteria represent the tools which enable alternatives to be compared from a specific point of view. It must be remembered that the selection of criteria is of prime importance in the resolution of a given problem, meaning that it is vital to identify a coherent family of criteria C = c1 ,K, c j ,K, ck . The alternatives are compared pairwise

under each criterion and the decision maker, faced with the two actions ai and am, can express: an outright preference (aiPam); a weak preference, if it is less marked, (aiQam); indifference (aiIam); or incomparability (aiRam) if none of the former apply. For each criterion the decision maker can choose from a set of six different types of preference functions to model the decision makers preferences (see fig. 3); Evaluation matrix. Once the set of criteria and the alternatives have been selected then the payoff matrix is built. This matrix tabulates, for each criterionalternative pair, the quantitative and qualitative measures of the effect produced by that alternative with respect to that criterion. The matrix may contain data measured on a cardinal or an ordinal scale. Each alternative Ai = ai ,1 ,K ai , j ,K, ai ,m is composed of a group of valuations aij representing

the evaluation given to the alternative i with respect to the criteria j; Determining the multi-criteria preference index The degree of preference of an alternative ai in comparison to am is expressed by a number between 0 and 1 (from 0 indicating no preference or indifference up to 1 for an outright preference). When the pairs of alternatives ai and am are compared the outcome of the comparison must be expressed in terms of preference in the following way [30]: Pk(d) = 0 means there is indifference between ai and am or no preference; Pk(d) 0 expresses a weak preference for ai over am; Pk(d) 1 strong preference for ai over am; Pk(d) = 1 outright preference for ai over am.

184

Fausto Cavallaro

In practice this preference function Pk(d) represents the difference between the evaluation of the two alternatives, thus it can be expressed as follows [32]:

Pk (c(ai ) ck (am )) = Pk (d ) [0,1]

Pk (ai , am ) = Pk [d (ai , am )]

(2) (3)

Figure 3. Criteria parameters.

Once the decision maker has described the preference function Pk (k= 1,2,3,.n represent the criteria) then a vector containing the weights of each criterion must be defined

W T = [w1 ,K, wk ] . The weights represent the relative importance of the criteria used for

the assessment, if all criteria are equally important then the value assigned to each of them will be identical. A variety of techniques exist to determine weights, the simplest but also the most arbitrary is direct assignment where weights are set by the decision maker, other

Electricity from Renewable Energy Sources

185

techniques require that the decision maker and an analyst work together to obtain a vector of weights that conforms as closely as possible to the decision makers preferences. In addition to weighting the method involves setting thresholds that delineate the decision makers preferences for each criterion and the critical thresholds are thus: the indifference threshold qi and the preference threshold pi (a more exhaustive description of the procedure can be found in the literature). The index of preference is calculated for each pair of actions ai and am as the weighted average of preferences calculated for each criterion. The index is therefore defined as follows [30]:

(a , a
i

)=

w
k =1

Pk (ck (ai ) ck (am ))

W
k =1

(4)

(ai, am) represents the strength of the decision makers preference for action ai over action am considering all criteria simultaneously and (am, ai) how much am is preferred over ai. Its value falls between 0 and 1 whereby: (ai, am) 0 indicates a weak preference for ai over am for all criteria; (ai, am) 1 indicates a strong preference for ai over am for all criteria.

Ranking the Alternatives The traditionally non-compensatory and methodologically important models include ones in which preferences are aggregated by means of outranking relations. Outranking is a binary relation S defined in A such that ai S am if, given the information relating to the decision makers preferences there are enough arguments to decide that ai is at least as good as am while there is no reason to refute this statement, i.e. aiSjam implies amSjai. Let us consider how each alternative ai A is evaluated and therefore define the two following outranking
flows:

+ (ai ) =

1 (ai , am ) n 1 xA

(5)

This indicates a preference for action ai above all others and shows how good action ai is (positive outranking flow).

(ai ) =

1 (am , ai ) n 1 xA

(6)

This indicates a preference for all the other actions compared with ai and shows how weak action ai is (negative outranking flow).

186

Fausto Cavallaro

According to PROMETHEE I ai is superior to am if the leaving flow of ai is greater than the leaving flow of am and the entering flow of ai is smaller than the entering flow of am. The PROMETHEE I partial preorder (PI, II, RI) is obtained by considering the intersection of these two preorders [30]:
+ (ai ) > + (am ) and (ai ) < (am ) I + + if ai P am (a i outrank a m ) (ai ) = (am ) and (ai ) < (am ) + + (ai ) > (am ) and (ai ) = (am ) I + + (ai ) = (am ) and (ai ) = (am ) ai I am (a i is indifferent to a m ) if I ai R am (a i and a m are incomparable) otherwise

(7)

where PI, II, and RI stand for preference, indifference and incomparability. Finally ai outranks am if:

+ (ai ) + (am ) and (ai ) (am )

(8)

Equality in + and - indicates indifference between the two compared alternatives. Under the Promethee I method some actions remain incomparable, in the case that a complete preorder is required that eliminates any incomparable items, then Promethee II can give a complete ranking as follows [31]:

net (ai ) = + (ai ) (ai )


The net flow is the difference between the out-flow and the in-flow.

(9)

4. EVALUATION OF ALTERNATIVE ENERGY PROJECTS USING PROMETHEE


4.1 The Proposed Energy Options
4.1.1. Photovoltaic Photovoltaic conversion technology (PV) was initially developed in the late 1950s as part of the space programme which required a reliable and inexhaustible source of energy. PV has been known since the end of the last century but the first commercial application was achieved in 1954; by Bell laboratory researchers in the United States when they perfected the first photovoltaic cell using monocrystalline silicon. Nowadays, its use is spreading very rapidly, in part due to advances in technology, even to terrestrial applications such as fuel for isolated users or for installations in buildings linked up to a pre-existing electricity grid.

Electricity from Renewable Energy Sources

187

The growth in the PV market certainly represents one of the long term strategic objectives for future worldwide energy policy and poses a research challenge in the field of RES. In general, commercial competitiveness is the element that most heavily affects the extent of the spread of installations exploiting RES as this is heavily affected by the cost of building the installations and their operating and maintenance costs. The production costs of technologies that exploit solar energy are unfortunately still extremely high but these could be reduced in the future if production volumes improve. From a theoretical point of view, cost reductions can come about when a combination of improvements is seen: in technological performance of production together with the optimisation of production cycles to enhance levels of the amounts produced. The crucial parameter for shifting the production cost curve is represented by capacity to innovate that is generated by research activity. Thus cost reductions can be achieved by steady market growth together with research efforts and spinoffs from other high-tech sectors of industry, such as micro-electronics, nanotechnology, the car industry and space sector. As has happened in other technological sectors, new products will come onto the market thereby allowing further cost reductions. The world PV market has seen considerable development over recent years, with a recorded annual growth rate of more than 30% and reached a total installed capacity of over 5000 MW in 2005 [33]. According to data from the report Marketbuzz, 2006 [34], despite the high price of silicon, new installations worldwide in 2005 amounted to 1,460 MW; a growth of 34% compared to 2004. 57% of the installations were in Germany alone leading it to exceed 1,500 MW of total power [33]. Market conditions differ substantially from country to country and this is due to the different energy policies implemented by different governments, whether there are support programmes for renewable sources and the differing degree of liberalisation of the electricity market. It is interesting to note that the specific legal provisions and regulations adopted by each country affect the effectiveness of any measures adopted to develop this sector. In some states where the rules do not include a system of charges to cover expenditure then the market impact is fairly marginal. In other countries, where prices are sufficiently high, effectiveness can be limited by allowing incentive tariffs for too short a period of validity or by bureaucracy and administration that is overly complicated and labyrinthine. In Europe, the markets in Germany and Spain have been highly dynamic and production of cells and modules has grown more rapidly than in other member states. Undoubtedly, this is attributable not only to a model of incentives that is clear, transparent and streamlined, but also to stable political and socio economic conditions that have favoured the creation of a secure and trustworthy climate for investors. Such a scenario has not only persuaded private and commercial investors to install PV plants but has, above all, boosted investments in R&D and has created the conditions to expand production capacity of cells and modules of industries in this sector[35].

4.1.2. Wind Power Nowadays, wind energy is without a doubt the most mature and commercially competitive of the new renewable sources and represents the segment of the market enjoying the highest growth rate in the entire energy sector. The expansion and the commercialisation of wind turbines has always coincided with their technological development. Now the market and technologies are specialising in order to maximise production under all conditions. Thus, there are now wind turbines specifically designed for offshore applications, to handle variable windspeeds and direction, turbines for turbulent winds and for small scale applications (from

188

Fausto Cavallaro

farms on the American flatlands to villages in developing countries). Modern wind turbines have modular features, they are reliable and can be installed on site in a very short space of time, they are built to operate continuously, are low maintenance and require few personnel over their lifespan lasting more than 20 years [36]. Areas for research and development (R&D) are numerous and concern mainly the use of innovative materials in order to build increasingly larger systems at contained prices and to increase the efficiency and reliability of the system. The geographical area that is most widely equipped with wind power installations is Europe, accounting for around 74% of the total, with 50% of this being the German Wind Park. Spain, Denmark and Italy are the other countries in Europe with significant wind energy installations whilst in the rest of the world the USA takes the lead with 14% of the total, India accounts for 6%, while China and Japan the figure is around1.5%. Whilst Germany is the undisputed world leader in the sector, the USA has shown an upward but inconstant trend in wind power development and India has also shown marked development in recent years [36]. Electricity production from wind power in the EU amounted to 69 TWh in 2005 with Germany (27.2 TWh) and Spain (21 TWh) accounting for more than 50% of this. Denmark came in third place (6.6 TWh), followed by Italy, Holland and the UK with production of around 2 TWh [33]. The industry that produces wind generators is highly concentrated, with the four leading firms accounting for around 70% of total power in Europe in 2005: the Danish company Vestas with 35%, the German Enercon with 14.4%, Gamesa from Spain with 13.4%, and the US business GEWind with 12.4% [33]. The costs of generating electricity from wind sources have come down steadily and visibly over the last 15 years as a result of the increased efficiency of wind turbines as well as their lower cost due not only to economies of scale that have been seen in the sector but also as a result of research and new technologies available, especially in the processes involved in manufacturing the various component parts of the aerogenerators.

4.1.3. Solar Chimney The thermal solar chimney is a recently developed technology patented by Schlaich Bergermann und Partner which uses a large cylindrical tower that is able to exploit energy from the sun to produce electricity. The system comprises a glass collector, a chimney and wind turbines. It works on the basis of the following principle: a large mass of cold air enters freely underneath a large glass roof (glass collector) that is open around its periphery. Solar radiation heats this air until it reaches a temperature in excess of 35C, thus creating an artificial greenhouse. The hot air tends naturally to move towards the centre of the collector where a cylindrical tower made of cement is located. This mass of hot air rises (hot air being less dense and thus lighter than cold) up the chimney tower thereby supplying a natural convective force (see Figure 4) [37] [38]. This flow of air, which rises at a speed of 14-16 m/s, is captured by a set of wind turbines located at the chimney base which convert kinetic energy from the wind induced by solar heat into mechanical energy and then into electricity. 4.1.4. Parabolic Solar Trough The technology relating to linear parabolic troughs is at a more advanced stage than other types of concentrated solar power technologies (parabolic dish and solar tower). Systems

Electricity from Renewable Energy Sources

189

using linear parabolic mirrors called SEGS (Solar Electric Generating Systems) are to date the most developed commercially speaking, typically ranging in size from 30 to 80 MWatt.

Figure 4. Solar chimney principle [37].

Figure 5. Diagram of a parabolic trough power plant with two-tank molten salt storage[41].

These devices first appeared in 1984 when the LUZ Company installed a 14 MW solar energy power plant (SEGS I) in southern California, using parabolic trough solar collectors and supplementing it with natural gas as fuel to allow the system to work during periods

190

Fausto Cavallaro

when there is little sun or while closed down for maintenance. A parabolic trough power plant with heat storage is made up of three basic parts [39] [40] [41]: 1) the solar field fitted with a circuit for heat transfer; 2) a system for storing heat; 3) a power block comprising a turbine, a generator and a cooling system (see Figure 5). This technology uses a curved mirror system to concentrate solar radiation onto a high thermal performance absorbent pipe laid along the focal line of the concentrators inside of which flows a fluid for heat transfer. In such an installation, the solar field has a modular structure composed of linear parabolic collectors linked in series laid out in parallel rows up to several hundred metres in length. The fluid that carries the heat absorbed from the sun is generally a mineral oil and is pumped through receiver pipes to a power plant. Here a heat exchanger converts the heat into steam which is then sent to a turbine to produce electricity.

4.1.5. Solar Tower The technology relating to Solar power towers is commercially at a less advanced stage than linear parabolic trough collectors. Despite this, a number of experimental stations have been tested on the field in a variety of sites scattered all over the world in the last 15 years. These have demonstrated their project feasibility and the economic potential of this technology. The plant consists of a set of mirrors, called heliostats, that track the movement of the sun on a double axis and which reflect solar energy onto a receiver (heat exchanger) mounted on the top of a tower positioned at the centre of the array of mirrors. A fluid transfers the heat from the receiver to a steam generator that drives a turbine. The heat transfer medium can be: steam/water, molten salts, liquid sodium or air. If gas or compressed air is used then extremely high temperatures can be reached thus achieving a very high level of efficiency [42]. As for other renewable energy sources, one of the limitations solar energy has to overcome is the inconstancy of the energy available which means that storage systems are a highly important feature that affect the advancement and spread of the technologies developed. A storage system has to guarantee energy supply even when no solar energy is available (at night or when the sky is overcast or cloudy). The energy produced by thermal solar installations may not necessarily be limited only to hours of sunshine and by cloud movements. The central receiver can heat the fluid, for example molten salts, which also serve as energy storage. The hypothetical installation consists of a type of solar power tower like the ones known as PS10 with a power 11 MWe which are to be built in Spain. The PS10 solar tower according to designers should be capable of producing 23 million kWh of electricity annually, that is enough energy to satisfy the demand of 10.000 families. The project envisages the use of saturated steam as Heat Transfer Fluid (HTF). The system combines a field of 624 heliostats (each with a surface area of 121 m2) produced by the Spanish firm Solucar, for a total reflective surface area of 75,216 m2, a 100 metre high tower, a receiver that is able to produce saturated steam at 40 bar and 250 C and heat storage system for steam with a capacity of 15 MWh able to supply 50 minutes of plant operation time at 50% load [43].

Electricity from Renewable Energy Sources

191

Figure 6. Flow diagram of the PS10 solar tower power plant [44].

4.1.6. Dish - Stirling Dish Stirling systems have been tested mainly in the United States and in Europe since the mid 1980s and the results have been highly encouraging. In these systems the conversion of solar energy into electricity is particularly efficient with a net average annual yield rate ranging between 18 and 23%, higher than any other solar energy system, and have attained a record rate of 29% for a brief time. The dish concentrator reflects solar rays onto a concave receiver positioned at the focal point of the concentrator. Solar radiation is absorbed by the receiver which heats a gas (helium or hydrogen) in the Stirling engine up to a temperature of around 650C [44]. The heat from the sun is converted into mechanical energy by the Stirling engine and this mechanical energy is subsequently converted into electricity by a generator directly connected to the engine. Optimal functioning requires that the concentrator is perfectly orientated towards the sun, therefore it is mounted on a two-axis tracking system that allows the concentrator to be aligned vertically and horizontally (see Figure 7).

Figure 7. Scheme of the Dish/Stirling system [44].

192

Fausto Cavallaro

The alignment towards the sun is controlled by a tracking sensor of the sun or by a special software that instantaneously and continuously calculates the position of the sun [45] [46]. The choice of installing a Stirling engine is dictated by the fact that the Stirling cycle is the most efficient thermodynamic cycle for transforming heat into mechanical energy and electricity, thus its extraordinary properties make it suitable for this type of application. The size of individual installations can vary between 5 and 50 kWatt. The most important feature of these systems is their modularity allowing installations of any size or power to be built. The beauty of this technology is that its size can be adjusted to fit user needs: from a few kW, for use in remote sites or islands, up to hundreds of kW for distributed generation uses connected to the electricity grid. Unfortunately, the high unit costs reflect the fact that these systems have not reached a high level of technological maturity. The technology is still at the prototype stage of development and the industrialisation that would allow it to be commercially exploited on a large scale is not foreseen in the short term.

4.1.7. Biomass The case studied envisaged the construction of a 10 MWe installation using a steam boiler fuelled by energy crops, in particular by Miscanthus which is capable of producing 38,6 MWth with a 25% yield. A fluid bed combustion system was chosen over a oven grid combustion system despite its higher cost because it offers greater assurances in terms of environmental performance. Environmental performance levels are important because this system can reduce SOx by as much as 90% to stable gas residues and NOx as a result of the lower temperature for combustion. The phase to remove pollutants concludes with the extraction of particulate from smoke by the use of filters.

4.2. Sets of Criteria: Identification and Selection


The criteria are the tools that enable alternatives to be compared from a specific viewpoint. Undoubtedly, selecting criteria is the most delicate part in formulating the problem before the decision maker, and thus it is requires the utmost care and attention. The number of criteria is heavily dependent on the availability of both quantitative and qualitative information and data. Here 14 criteria were selected; 9 of these technical-economic and 5 socio-environmental. Quantitative measures apply to 8 of the criteria while the remaining 6, being qualitative in nature, were scored by applying impact scales from either 1-4 or 1-5.

Economic and Technical Criteria These criteria refer to the costs that must be borne in order to realize the various projects included in each strategy and to guarantee the supply of energy. These factors are of special interest to State authorities.
Investment costs. This includes all costs relating to the purchase of mechanical equipment, technological installations, to construction of roads and connections to the national grid, to engineering services, drilling and other incidental construction work. This criterion is measured in Euros;

Electricity from Renewable Energy Sources

193

Operating and maintenance costs. This includes all the costs relating to plants, employees wages, materials and installations, transport and hire charges, and any ground rentals payable. This criterion is measured in Euros; Levelized electricity cost (LEC). This measures the production cost per kWh of the electricity generated by the plant expressed as Euro cents. This parameter is important and useful for assessing how commercially competitive the system is compared with conventional energy production technologies; Financial risk. This identifies the degree of financial risk attached to the technological options proposed and is measured using the following scale of values: Low risk=1 Medium=2 High=3 Very high=4 Primary energy saving. This refers to the amount of fossil fuel currently used by power plants to produce electricity that could be saved. It is measured in ton/per annum; Maturity of technology. Measures the degree of reliability of the technology adopted as well as how widespread the technology is at both national and European level. This is appraised using a qualitative judgment transformed into the following fourpoint scale [6]: Technologies at theoretical level=1 Technologies tested in laboratory= 2 Technologies only performed in pilot plants and/or under construction =3 Technologies requiring further improvements to increase their efficiency levels=4; Commercially mature technologies on the market=5; Continuity of power supply: This criterion indicates whether the energy supply is subject to interruptions (e.g. PV does not work at night, wind generators cannot function when there is no wind, etc.) and thereby affects the stability of the electricity grid. This case is also evaluated qualitatively and expressed via the following fourpoint scale: Highly discontinuous activity =1 Moderately discontinuous activity =2 Slightly discontinuous activity =3; Stable and continued activity (except when the plant undergoes maintenance)=4; Storage capacity : this criterion indicates whether any storage systems are fitted that are able to ensure continuity of electricity supply (e.g. in the absence of solar radiation or other factors ) and is calculated on the basis of the number of hours of autonomy provided; Realization time. This measures the time to realize and put into operation the plants designed. It is expressed in number of months.

Environmental and Social Criteria These criteria refer to protection of the environment and to the principle of sustainability:

194

Fausto Cavallaro Sustainability of Climate Change: This refers to the amount of CO2 emissions avoided as a result of the production of the proposed plants. It is measured in tons/per annum. Sustainability of other impacts: This criterion takes into account other impacts: the visual nuisance that may be created by the development of a project in a specific area or any noise disturbance and odours arising from productive activity of plants, the potential risk to eco-systems caused by the production operations of the various projects included in the strategies. This is also measured qualitatively and translated into the following five-point scale [6]: Extremely high impact=1 High impact=2 Moderate impact=3 Slight impact=4 No impact =5. Social acceptability. Expresses the index of acceptance by the local population regarding the hypothesized realization of the projects under review. The following four-point qualitative scale was applied: The majority of inhabitants are against the installation of any plant whatsoever regardless of where it is =1 The opinion of the population regarding the installations is split =2 The majority accepts the installations provided they are located far from residential areas =3 The majority of inhabitants are favourably disposed towards the installations =4 Contribution to local development. This criterion estimates the global social and economic effects that may be felt in the areas affected by the initiatives. The potential effects are: the creation of new jobs, new supply chain businesses, emerging energy sector businesses, industrial districts etc. The following rating scale was applied: Impact on local economy rated weak =1 Impact on local economy rated moderate (some permanent jobs)=2 Impact on local economy rated medium-high (jobs + supply chain businesses)=3 Impact on local economy rated high (strong impetus to local development, creation of small industrial districts)=4 Land use. This criterion quantifies the area occupied by the plants and not available for possible alternative uses (i.e. agriculture or other economic activities). It is measured in m2 of land used.

4.3. The Evaluation Matrix


Table 1 shows the matrix containing the alternatives and how these perform with respect to the evaluation criteria selected. The options proposed are the following: wind power, a medium sized installation made up of 15 turbines of 600 kW each giving a total power of 9 MW (W.1) and a larger 30 MW installation using 15 turbines of 2000 kW each (W.2). The two photovoltaic options considered are a 5MW installation (PV.1) and one of 15 MW (PV.2).

Table 1. Evaluation matrix


Criteria Alternatives
A.1 Investment costs Euro (000) W.1 W.2 PV.1 Wind 600kW Wind 2000kW Photovoltaic 5MW Photovoltaic 15MW Solar Chimney 30MW Solar Chimney 15MW Parabolic solar trough 50MW Parabolic solar trough 50MW Solar Tower 11MW Solar DishStirling Biomass 20MW 10,800 45,000 30,000 A.2 A.3 A.4 Financial risk qualitative low low middle B.1 Primary energy saving ton/y 10,180 35,650 3,431 B.2 Maturity of technology qualitative commercial commercial further impr B.3 Continuity of power supply qualitative slight disc. slight disc. high disc. B.4 Storage capacity hours 0 0 0 B.5 Realization time months 18 36 24 C.1 Sustainabilty of climate change ton/y 17,720 62.000 5,900 C.2 Other environmental impacts qualitative slight high slight C.3 Social acceptability qualitative accept majority no favorable C.4 Contribution to local development qualitative moderate moderate moderate C.5 Land use Km2 0.02 0.065 0.09

O&M cost

LEC

Euro (000) 324 1,350 450

c/Euro 0.07 0.065 0.6

PV.2

75,000

1,125

0.5

high

10,290

further impr

high disc. middle disc. middle disc. middle disc. high disc.

36

17,910

moderate

split

medium-high

0.27

SCh.1

153,000

11,800

0.17

very high

34,000

theory

10

36

59,160

high

majority no

medium-high

6.831

SCh.2

99,000

6,800

0.15

very high

21,000

theory

10

36

36,540

high

majority no

medium-high

4.153

PST.1

117,000

2,879

0.106

middle

70,000

pilot

36

121,800

moderate

split

weak

1.459

PST.2

91,000

2,670

0.12

middle

55,900

pilot

24

97,200

moderate

split

weak

0.855

STo

36,000

2,175

0.22

middle

12,000

pilot

high disc.

0.5

36

20,880

slight

split

moderate

0.372

DS

180,000

8,035

0.38

very high

26,300

lab

high disc.

48

45,700

moderate

accept

moderate

0.84

Biom

40.000

1,025

0.9

low

6,690

commercial

slight disc.

24

19,400

high

majority no

medium-high

0.06

Table 2. Thresholds

A.1 Min/Max Weight Preference function Indifference threshold Preference threshold Threshold unit Minimize 0.8 V-Shape

A.2 Minimize 0.8 V-Shape

A.3 Minimize 0.9 Linear

A.4 Minimize 0.7 Linear

B.1 Maximize 0.9 Linear

B.2 Maximize 0.4 Linear

B.3 Maximize 0.7 Linear

B.4 Maximize 0.4 V-Shape

B.5 Minimize 0.4 V-Shape

C.1 Maximize 0.9 Linear

C.2 Maximize 0.9 Linear

C.3 Maximize 0.5 Linear

C.4 Maximize 0.5 Linear

C.5 Minimize 0.8 Linear

5%

0.5

5%

0.5

5%

0.5

0.5

0.5

5%

10%

10%

15%

10%

1.5

10

10%

10%

Percent

Percent

Percent

Absolute

Percent

Absolute

Absolute

Absolute

Absolute

Percent

Absolute

Absolute

Absolute

Percent

Table 3. Preference flows


Actions W.1 W.2 PV.1 PV.2 SCh.1 SCh.2 PST.1 PST.2 STo DS Biom + 0.6531 0.5524 0.4385 0.3218 0.2752 0.2582 0.449 0.4503 0.407 0.276 0.4035 0.181 0.238 0.404 0.484 0.511 0.555 0.371 0.363 0.407 0.559 0.412 Net flow (a) = + (a) (a) 0.473 0.315 0.034 -0.16 -0.24 -0.3 0.078 0.088 0 -0.28 -0.01 Ranking 1 2 5 8 9 11 4 3 6 10 7

Table 4. Weight stability interval


Absolute values Min 0.621 0.384 0.799 0 0.766 0.03 0 0.027 0.302 0.69 0.846 0.399 0 0.489 Relative values (%) Min 6.59% 4.18% 8.41% 0.00% 8.09% 0.33% 0.00% 0.30% 3.18% 7.35% 8.86% 4.20% 0.00% 5.27%

Criteria A.1 A.2 A.3 A.4 B.1 B.2 B.3 B.4 B.5 C.1 C.2 C.3 C.4 C.5

Weight 0.8 0.8 0.9 0.7 0.9 0.4 0.7 0.4 0.4 0.9 0.9 0.5 0.5 0.8

Max 1.067 1.039 1.476 0.815 1.316 0.458 0.801 0.501 0.485 1.342 1.471 0.826 0.59 0.912

Weight 8.33% 8.33% 9.38% 7.29% 9.38% 4.17% 7.29% 4.17% 4.17% 9.38% 9.38% 5.21% 5.21% 8.33%

Max 10.81% 10.56% 14.50% 8.39% 13.14% 4.74% 8.25% 5.16% 5.01% 13.36% 14.46% 8.32% 6.08% 9.39%

Figure 8. Partial ranking.

Figure 9. Final ranking.

Figure 10. Partial ranking (equal weight for criteria).

Figure 11. Final ranking (equal weight for criteria).

Electricity from Renewable Energy Sources

199

The options using solar chimney technology relate to one installation with a power capacity of 30 MW (SCh.1) installed on a site where solar radiation amounts to 1,800 kWh/m2/y (typical of the southern Mediterranean), and the other of 15MW (SCh.2) which is designed to be installed in areas with solar radiation of 2,300 kWh/m2/yThe proposed options for linear parabolic trough collectors are a 50MW installation (PST.1), fitted with an energy storage system and the other of the same capacity but without any storage system (PST.2), The remaining options proposed are one using solar power tower technology with a power capacity of 10MW (STo), an installation containing parabolic dish-stirling concentrators (DS) and a steam boiler of 20 MW fuelled by biomass (energy crops) (Biom).

Figure 12. GAIA plane.

The performance data relating to items measured quantitatively were extrapolated from published findings in the literature. The data evaluated qualitatively are the outcome of assessments and estimates. Before looking at the results in detail it is important to clarify a number of points regarding the data reported in the matrix. The costs relating to investment and maintenance, the industrial cost of production per kWh and the level of energy production, have been calculated based on data extracted from the following publications:

200

Fausto Cavallaro

[41] [43] [44] [45] [46] [47] [48] [49]. The criterion relating to financial risk based on market estimates shows a very high risk profile for options SCh.1, SCh.2 and DS reflecting the fact that the technologies are still in the preliminary study phase or in the experimental phase in the laboratory, thus their economic and financial profile is highly uncertain. The continuity of energy supply is an extremely important criterion to ensure that the grid is stable and not liable to black-outs. Nearly all the technologies analysed, with the exception of those equipped with a storage system, are subject to a fair degree of discontinuity, either because of differences between daytime and night or varying meteorological/climatic conditions. Environmental impact is probably the most delicate item of the whole assessment. The most significant impact is found for the option using biomass due to emissions of some pollutants (not harmful), and for the solar chimney options as result of the large area of land required to build the installation on, as well as the great visual impact created by the tower. It is also reasonable to assume that the wind energy installation will also be visually intrusive because the wind turbines are so large in size. These also constitute a potential collision hazard to birds in flight. The alternatives relating to parabolic trough and solar tower technologies apart from occupying a certain amount of land do not give rise to any other significant changes to the environment therefore their environmental impact are judged to be moderate. The greatest uncertainties attach to the evaluation of social acceptability and their contribution to local development as these could be strongly affected by prejudices that bias the results. Generally speaking, the population views installations using renewable sources favourably, but only if they are built far from human settlements and do not cause any nuisance to the population. The level of acceptance will therefore depend greatly on the exact locations of the installations. Estimates relating to these two criteria come from a survey carried out on a sample of the population that did not show react positively to the proposed projects.

4.4. Main Results


Two distinct rankings of alternatives are computed and displayed. The first is PROMETHEE I which gives a partial ranking. It is based on strongly established preferences so some actions remain incomparable under this method. Figure 8 graphically illustrates the positions of each alternative in the partial ranking and it is immediately apparent that the best performers are W.1, W.2 and PST.2. The first two of these also come out top in the final ranking and thus attest to the credibility of both the method and to the findings. Clearly, wind power is at a more commercial and competitive stage of development compared to other new energy technologies and the results obtained are comforting in that they are consistent with a well-consolidated fact. Next in the ranking come the options PST.1, PV.1, STo, Biom, and PV.2. The lowest ranked are the options SCh.1 and SCh.2 which are shown to be incomparable with the DS option. These technologies are still heavily penalised by high costs of investment and maintenance. The highly innovative status of these projects means that technologically they are still very immature and this greatly affects the economic-financial risks attached to these initiatives. Figure 9 shows the results from PROMETHEE 2 which gives a complete ranking: all actions are ranked from the best to worst leaving no incomparable pair of actions. This

Electricity from Renewable Energy Sources

201

information is easier to use than partial ranking but does reflect less reliable preferences. Table 4 gives the weights assigned to the criteria together with the weight stability intervals that give for each criterion the limits within which its weight can be modified without changing the PROMETHEE II complete ranking. The results of multi-criteria analysis hinge on the weightings allocated and thresholds set. As stated earlier, the weights express the importance of each criterion and obviously may deeply influence the final outcome of the entire calculation procedure. For some authors, the problem of how to determine the weights to assign is still unresolved since the different outranking methods do not lay down any standard procedure or guidelines for determining them. In order to examine any changes in the final outcome after correcting the weights previously assigned to the various criteria a sensitivity analysis was performed by assigning an equal weight to all the criteria selected. This calculation revealed that the position in the ranking did change for some options although the overall structure of the ranking from the previous analysis was more or less unchanged. The options PV.1 and Biom moved up from 5th and 7th place to 3rd and 4th and emerged as incomparable with each other. The options PST.1 and PST.2 slid from 2nd and 3rd place down to 5th and 6th place and were also found to be incomparable (see Figure 10 and 11). The GAIA (geometrical analysis for interactive aid) plane provides the decision maker with a comprehensive graphical image of the decision problem and it is complementary to the multicriteria analysis. This tool provides clear graphical information regarding the conflicting characters of the criteria and about the weights on the final decision. Figure 12 illustrates this analysis: actions are represented as triangles and criteria as lines. The direction of the PI axis (the line that joins the two dots) identifies the compromise solution, in our case the axis is moving towards alternative W.1 and W.2. By examining a GAIA graph it is possible to see where the action lies in relation to the criteria, measure how intensely the criteria affect each action and thereby identify the criteria that are in line with or conflict with the various alternatives.

CONCLUSION
Figure 8 graphically illustrates the positions of each alternative in the partial ranking and it is immediately apparent that the best performers are W.1, W.2 and PST.2. These are followed by the options PST.1, PV.1, STo, Biom, and PV.2. The bottom ranking positions are occupied by the options SCh.1 and SCh.2, which are shown to be incomparable with the DS option. A sensitivity analysis was performed by equally weighting all the criteria and comparing the ranking obtained with the original. The two rankings were found to be similar although the position of some options shifted slightly; with the options PV.1 and Biom moving up from 5th and 7th place to 3rd and 4th and emerged as incomparable with each other whilst PST.1 and PST.2 moved down from 2nd and 3rd place to 5th and 6th and were also found to be incomparable with each other. This work has therefore attempted to test the soundness and strength of multicriteria analysis as a means to serve energy planners as an unambiguous tool for decision making. DSS are well-suited to dealing with highly structured problems or semi-structured ones (for which it is not possible to produce an ideal solution). DSS help the user in all phases of the

202

Fausto Cavallaro

decision making process by using relevant information in order to make choices that can be documented and are transparent. Assessment procedures and energy planning may appear complex because of the number and diversity of the items to evaluate, the uncertainty of data and conflicts between interested parties. The decision making process of an energy project is the closing link in the process of analysing and handling different types of information: environmental, technical, economic and social. As this work demonstrates, multicriteria analysis can provide a technical-scientific decision making support tool that is able to justify its choices clearly and consistently.

ACKNOWLEDGMENT
I wish to thank Dr. Maria Cristiana Laur and Ms Susan H. Parker B. A., A.C.A. for their precious help and cooperation.

REFERENCES
Communication from the Commission to the Council, the European Parliament, the European Economic and Social Committee and the Committee of the Regions. Towards a European Strategic Energy Technology Plan. SEC (2007) 12, COM (2006) 847 10/01/2007. [2] Siskos J., Hubert P.H. (1983). Multi-criteria analysis of the impacts of energy alternatives: a survey and a new comparative approach. European Journal of Operational Research, 13, 278299. [3] Roy B., Bouyssou D. (1986). Comparison of two decision-aid models applied to a nuclear plant siting example. European Journal of Operational Research, 25, 200-215. [4] Georgopoulou E., Lalas D., Papagiannakis L. (1997). A multicriteria decision aid approach for energy planning problems: the case of renewable energy option. European Journal of Operational Research, 3, 38-54. [5] Georgopoulou E., Sarafidis Y., Diakoulaki D. (1998). Design and implementation of a group DSS for sustaining renewable energies exploitation. European Journal of Operational Research, 109, 483500. [6] Beccali M., Cellura M., Mistretta M. (2003). Decision-making in energy planning. Application of the Electre method at regional level for the diffusion of renewable energy technology. Renewable Energy, 28, 2063-2087. [7] Haralambopoulos D.A., Polatidis H. (2003). Renewable energy projects: structuring a multi-criteria group decision-making framework. Renewable Energy, 28, 961-73. [8] Goumas M., Lygerou V., Papayannakis L. (1999). Computational methods for planning and evaluating geothermal energy projects. Energy Policy, 27, 147-154. [9] Cavallaro F. (2006). A decision-making support to assess concentrated solar thermal technologies. Proceedings of 13th International Symposium on Concentrated Solar Power and Chemical Energy Technologies, SolarPACES2006 A4-S3, Seville Spain. [10] Barda O.H., Dupuis J., Lencioni P. (1990). Multicriteria location of thermal power plants. European Journal of Operational Research, 45, 332-346. [1]

Electricity from Renewable Energy Sources

203

[11] Cavallaro F. (1999). Unanalisi multicriteriale per la gestione di strategie energetiche alternative nelle isole minori in Sicilia. Economia delle fonti di energia e dellambiente (IEFE-Bocconi), 3, 59-90. [12] Cavallaro F., Ciraolo L. (2005). A multicriteria approach to evaluate wind energy plants on an Italian island. Energy Policy, 33, 235-244. [13] Pohekar S.D., Ramachandran M. (2004 Application of multi-criteria decision making to sustainable energy planning A review. Renewable and Sustainable Energy Reviews, 8, 365-381. [14] Loken E. (2007). Use of multicriteria decision analysis methods for energy planning problems, Renewable amd Sustainable Energy Reviews, 11, 1584-1595. [15] Catrinu M., Bakken B.H., Holen A. (2004). Modelling local energy systems from a multicriteria perspective. Presented at the 17th International conference on efficiency, optimization, simulation and environmental impact of energy and process system. Guanajuato - Mexico. [16] Zhou P., Ang B.W., Poh K.L. (2006). Decision analysis in energy and environmental modeling: An update. Energy, 31, 26042622. [17] Paruccini M. (1991). Decision Support System for Environmental Management. European Commission, DG JRC, Report EUR 14087. [18] Simon H.D. (1957). Models of Man. New York USA: John Wiley. [19] Groscurth H.M., Schweiker A. (1995). Contribution of Computer Models to Solving the Energy Problem. Energy Sources, 17 (2), 161-177. [20] Catrinu M.D. (2006). Decision Aid for Planning Local Energy Systems: Application of Multi- Criteria Decision Analysis, Doctoral Theses 2006: 62, Trondheim: Norwegian University of Science and Technology, Faculty of Information Technology, Mathematics and Electrical Engineering, Department of Electrical Power Engineering. [21] Bogetoft P., Pruzan P. (1997). Planning with Multiple Criteria - Investigation, Communication and Choice, Handelshjskolens forlag, Copenhagen Business School. [22] Mishan E.J., (1967) The cost of economic growth. Staples Press, London. [23] Gerelli E., Panella G., Cellerino R. (1984). Valutazione dimpatto ambientale e calcolo economico. Milano - Italy: Franco Angeli.. [24] Buchanan J.T., Erez J., Henig M.I. (1998). Objectivity and subjectivity in the decision making process. Annals of Operations Research (Issue on Preference Modelling), 80, 333-345. [25] Henig M.I., Buchanan J.T. (1996). Solving MCDM problems: Process concepts. Journal of Multi Criteria Decision Analysis, 5(1), 3-12. [26] Roy B. (1985) Mthodologie multicritere daide la decision. Paris - France: Economica. [27] Roy B. (1990). The outranking approach and the foundations of ELECTRE methods. In Bana e Costa C.A. (editor), Readings in Multiple Criteria Decision Aid (pp. 1-660). Berlin: Springer-Verlag. [28] Guitoni A., Martel J.M. (1998). Tentative guidelines to help choosing an appropriate MCDA method. European Journal of Operational Research, 109, 501-521. [29] Brans J.P.,Vincke Ph. (1985). PROMETHEE. A new family of outranking methods in MCDM. Management Science, 31, 6 647-656. [30] Brans J.P., Vincke Ph., Mareschal B. (1986). How to select and how to rank projects: The Promethee method. European Journal of Operational Research, 24, 228-238.

204

Fausto Cavallaro

[31] Brans J.P., Mareschal B. (1994). The Promcalc and Gaia decision support system for multicriteria decision aid. Decision Support System, 12, 297-310. [32] Brans J.P., Mareschal B. (1998). Multicriteria decision aid the Promethee-Gaia solution. Working paper, Vrije Universiteit Brussel, STOOTW/288. [33] Nomisma Energia (2007). Le nuove fonti rinnovabili per lenergia elettrica in Europa. [34] www.marketbuzz.com. [35] Arnulf Jger-Waldau, PV Status Report 2005 - Research, Solar Cell Production and Market Implementation of Photovoltaics. European Commission, DG JRC, Institute for Environment and Sustainability, Renewable Energies Unit. EUR 21836 EN. [36] ENEA (2005). Rapporto Energia e Ambiente. [37] Schlaich J., Bergermann R., Schiel W., Weinrebe G., (2005). Design of Commercial Solar Updraft Tower Systems Utilization of Solar Induced Convective Flows for Power Generation. Journal of Solar Energy Engineering, 127, 117-124. [38] Von Backstrom T.W., Gannon A.J. (2004). Solar chimney turbine characteristics. Solar Energy, 76, 235241. [39] www.solarmillenium.de. [40] Price H., Kearney D. (1999). Parabolic-Trough Technology Roadmap: A Pathway for Sustained Commercial Development and Deployment of Parabolic- Trough Technology. NREL Report. [41] Herrmann U., Kelly B., Price H. (2004). Two-tank molten salt storage for parabolic trough solar power plants. Energy, 29, 883893. [42] Reilly H.E., Kolb G.J., (2001). An evaluation of molten salt power towers including results of the solar two project. Sandia National Laboratories - USA. [43] Osuna R., Olavarra R., Morillo R., Snchez M., Cantero F., Fernndez-Quero V., Robles P., Lpez del Cerro T., Esteban A., Cern F., Talegn J., Romero M., Tllez F., Marcos M.J., Martnez D., Valverde A., Monterreal R., Pitz-Paal R., Brakmann G., Ruiz V., Silva M., Menna P. (2006). PS10- Construction of a 11MW solar thermal tower plant in Seville - Spain. Proceedings of 13th International Symposium on Concentrated Solar Power and Chemical Energy Technologies SolarPACES2006 A4S3, Seville Spain. [44] Pitz-Paal R., Dersch J., Milow B. (editors). European Concentrated Solar Thermal Road-Mapping Roadmap Document (SES-CT-2003-502578) ECOSTAR. [45] Marketaki K., Gekas V. (1999). Use of the thermodymamic cycle Stirling for electricity production. Proceedings of the 6th Panhellenic Symposium of Soft Energy Sources. [46] Tsoutsos T., Gekas V., Marketaki K. (2003). Technical and economical evaluation of solar thermal power generation. Renewable Energy, 28, 873886. [47] Trieb F., Langnib O., Klaib H. (1997). Solar electricity generation a comparative view of technologies, costs and environmental impact Solar energy, 59, 89-99. [48] Weinrebe G., Bergermann R., Schlaich J., Schiel W., Hornidge D. (2006). Commercial Aspects of Solar updraft towers. Proceedings of 13th International Symposium on Concentrated Solar Power and Chemical Energy Technologies, SolarPACES2006 A4S3, Seville Spain.. [49] Kearney D., Kelly B., Herrmann U., Cable R., Pacheco J., Mahoney R., Price H., Blake D., Nava P., Potrovitza N. (2004). Engineering aspects of a molten salt heat transfer fluid in a trough solar field. Energy, 29, 861870.

In: Leading-Edge Electric Power Research Editor: C. M. OSullivan, pp. 205-238

ISBN: 978-1-60021-981-8 2008 Nova Science Publishers, Inc.

Chapter 4

GAS TURBINES AND ELECTRIC DISTRIBUTION SYSTEM


Francisco Jurado
University of Jan, Department of Electrical Engineering 23700 Alfonso X, n 28, EPS Linares (Jan), Spain

ABSTRACT
Lately, the use of gas turbines following the deregulation of the electricity supply industry has become greater quickly. The motivation for modeling the gas turbines and their controllers is determinant to the interpreting of their impacts on distribution systems. The model predictive control (MPC) is used to damp the oscillation when the power distribution system is subjected to a disturbance. MPC is selected because it can explicitly handle the nonlinearities, and constraints of many variables in a single control formulation. The IEEE 13 node power distribution system is employed to demonstrate the effectiveness of MPC to damp the oscillations of gas turbines. Among fossil fuels, gas is the most quickest, with a growth rate nearly double that of coal and oil. The electricity generation field is the leading market for gas. The natural gas business has a great interaction with the electricity market in terms of fuel consumption and energy conversion. On the other hand, the transmission and distribution activities are very similar with the natural gas transportation through pipelines. The power losses in gas and electric systems are compared. It is also demonstrated that the electricity system results more convenient for longer distances of gas wells from electricity consumption area.

Keywords: Distribution networks, gas turbines, modeling, power loss, predictive control.

E-mail: fjurado@ujaen.es Telephone: +34-953-648518 Fax: +34-953-648586.

206

Francisco Jurado

NOMENCLATURE
Ao a, b, c cpa cpg cps compressor exit flow area valve parameters specific heat of air at constant pressure (J/(kg K)) specific heat of combustion gases (J/(kg K)) specific heat of steam (J/(kg K)) internal diameter of pipe between nodes (m) valve position vector of mass flow rates through branches fuel demand signal Fanning friction coefficient flow rate through compressor (m3/s) pipeline flow rate (m3/s) gas specific gravity horsepower required to pump gas down PID parameter PID parameter fuel system gain constant factor which depends on LHV lower heating value (MJ/kg) pipeline length between nodes (m) polytropic index rotation speed of the turbine (rad/s) branches in the system compressor power consumption (W) air pressure at compressor inlet (Pa) air pressure at compressor outlet (Pa) real power required at the ith bus real power generated at the ith bus real power loss mechanical power delivered by turbine (W) total mechanical power delivered by turbine (W) pressure of combustion gases at turbine inlet (Pa) pressure of combustion gases at turbine outlet (Pa) pressure ratio (outlet/inlet) time (s) mechanical torque delivered by turbine (Nm) outlet air temperature (K) temperature of injected steam (K) average gas temperature (K)

Dk
e1 f Fd

Fk fk f kij
G

Hk
KI KP kf kLHV LHV

Lk
ma N

NP
Pc pcin pcout

PDi PGi
PL Pm PT pTin pTout rc t T Tcout Tis

Tka

Gas Turbines and Electric Distribution System

207

Tki T0
TTin U(t) Vi YT , UT w wa wf wg wis

compressor suction temperature (K) standard temperature (K) turbine inlet gas temperature (K) control signal voltage magnitude at the ith bus finite time Fourier transforms vector of gas injections at each node air mass flow into the compressor (kg/s) fuel mass flow (kg/s) turbine gas mass flow (kg/s) injection steam mass flow (kg/s) vector of gas demands vector of gas supplies magnitude of the i-jth element of the bus admittance matrix average gas compressibility factor gas compressibility factor at compressor inlet

wL wS
Yij

Za Z ki

specific heat ratio

cp cv

ij h25 hIC hIT N i

angle of the i-jth element of the bus admittance matrix specific enthalpy of reaction at reference temperature of 25C (J/kg) isentropic enthalpy change for a compression from pcin to pcout (J/kg) isentropic enthalpy change for a gas expansion from pTin to pTout (J/kg) rotation speed deviation of the turbine (rad/s) phase angle of the voltage pipeline efficiency overall compressor efficiency compressor efficiency overall turbine efficiency transmission efficiency from turbine to compressor compressor polytropic efficiency pressure at node i (Pa) pressure at node j (Pa) compressor suction pressure (Pa) compressor discharge pressure (Pa) standard pressure (Pa) inlet air density fuel system time constant (s)

T trans c

i
j

ic
jc

0
i f

208

Francisco Jurado

1. INTRODUCTION
Gas turbines can offer solutions to todays energy situation as a supplement or support function to the conventional central generation and power system (Willis and Scott, 2000). Complimentary answers are needed to meet projected growth in new load and peak demand while providing power system stability, security and end-user power quality solutions. Distributed generation (DG) integration into the existing power grid can enhance asset utilization without demanding major capital investment in new large generation or energy delivery facilities. Synchronous machine stability surveys have been a discipline of interest for many years. Much of the work produced has been based on steam- or hydro-turbine generating units. Power system stability is normally associated only with large utility systems. However, with DG operating in parallel with the utility, stability has become an issue that is crucial to preserve critical functions (Jurado and Carpio, 2005). Compared to the transmission system, the distribution system has several important characteristics. The power of DG is relatively small compared to the capacity of the substation. The substation is stiff enough to keep the frequency constant, thus can be conceived as an infinite bus. Model predictive control (MPC) is a control strategy that uses a model of the system to predict the response over a future interval, called the costing or prediction horizon (Maciejowski, 2002; Richalet, 1993; Qin and Badgwell, 2000). The application of MPC to control the gas turbine was introduced in (van Essen and de Lange, 2001; Vroemen et al., 1999). Model based control schemes are highly related to the accuracy of the process model. Evans concentrated on testing the gas turbine using small amplitude multisine signals and frequency domain techniques to identify linear models of high accuracy at a range of different operating points (Evans et al., 2000). The implementation of an efficient method for computing low order linear system models of gas turbines from time domain simulations is presented in (Jurado and Cano, 2004). This method is the Box-Jenkins algorithm for calculating the transfer function of a linear system from samples of its input and output. The fact that the dynamics of these models change with operating points evidenced that the gas turbine is nonlinear, so the need was evident for a more accurate nonlinear modeling of the gas turbine. The work was formulated further by Chiras who used Nonlinear AutoRegressive Moving Average with eXogenous inputs (NARMAX), to represent the global dynamics of the gas turbine. It was showed that both models were suitable for representing engine dynamics throughout its operating range (Chiras et al., 2002). The Hammerstein model is a special kind of nonlinear systems which has applications in many engineering problems (Narendra and Gallman, 1996). A frequency domain identification approach for Hammerstein models is proposed in (Jurado, 2005). By exploring the fundamental frequency, the linear part and the nonlinear part can be identified. The power system stability mostly depends on the excess kinetic energy stored in the generators during the fault duration period. The efficient control of the extra energy can be the most direct method to reach the system stabilization. There are various types of energy storage devices which can be applied to control the surplus energy. However, all of the energy storage devices have the common shortcomings that they are very costly and require highly advanced control strategy for the operations.

Gas Turbines and Electric Distribution System

209

It is attractive if the excess kinetic energy could be controlled through the adjustment of the fuel control valve. The MPC is designed for this purpose. In this work, the MPC is used to prevent the transient instability in power distribution system with DG. The unbundling of the electricity sector and the force of competition have developed new technologies of the generation and the delivery of electricity which mean less pollutant, higher efficiency, and less costly means of supplying the load. These technologies often apply to conventional (coal, oil, gas, hydro, nuclear) and unconventional (solar, wind, fuel cells, microturbines) sources of energy. The possible alternative of replacing coal and oil burning plants with natural gas plants could greatly meliorate the sustainability of forests, waters, and farmlands, which are negatively affected by acid deposition. Natural gas is produced primarily at remote sites and transmission pipelines, distribution pipelines, vast underground storage facilities, and compressors are fabricated to deliver the natural gas from wellheads to power generating sites and end users. Despite the strong worldwide growth in demand for natural gas, the reserves continue to rise, thanks to the improvements in gas exploration technologies. The exploitable proven reserves are of around 150 tera cubic meters (Tm3), with a residual life, at present consumption rate, of around 65 years; considering another 200 Tm3 of potential discoveries, gas has a future potential nearly twice that of oil. The electricity generation field is the leading market for gas. This extraordinary growth in the electricity sector is driven by various factors (DOE, 2001 a; DOE, 2001 b): (1) Of all forms of energy consumption, electricity is becoming the most important for final consumers, with an ever increasing penetration rate: today around 37% total. (2) The new technologies for combined cycle gas turbines plants allow a very high efficiency (close to 60%), low emissions and environmental impact, very attractive investment costs and short completion times. (3) The ever increasing environmental concerns for power plant emissions and the explosion of Independent Power Producers in a deregulated market that is quickly changing the rules of the game. A great number of pipeline systems are under study and for the medium/long term very long and high capacity pipelines are being considered (e.g. Yamal - Europe and Turkmenistan- Europe), analyzing also new technology solutions (high pressure pipelines) to reduce the gas transmission costs. This paper compares power losses between gas and electricity distribution systems, of interest when natural gas is needed for electricity generation. The article is organized as follows. In Section 2, general principles of gas turbines are explained. In Section 3, the Hammestein model is introduced. The MPC is formulated in Section 4. Section 5 presents a review of the components of gas system. Some basic concepts of the gas steady-state equations are presented in Section 6. Section 7 describes electric power losses. Section 8 depicts some simulation results and discussion. Finally, conclusions are presented in Section 9.

210

Francisco Jurado

2. GAS TURBINE MODEL


The plant model is a physics based component level model (CLM) of this gas turbine configuration. This model is very detailed, high-fidelity, and models each component. A thorough introduction to the gas turbine theory is provided in (Cohen et al., 1998). There exist a large number of publications on the modeling of gas turbines. The model complexity varies according to the intended application. A detailed first principle modeling based upon fundamental mass, momentum and energy balances is reported in (Fawke et al., 1972). These models describe the spatially distributed nature of the gas flow dynamics by dividing the gas turbine into a number of sections. Throughout each section, the thermodynamic state is assumed to be constant with respect to location, but varying with respect to time. Mathematically, the full partial differential equation model is reduced to a set of ordinary differential equations, which are facilitated easily within a computer simulation program. For a detailed model, a section might consist of a single compressor or turbine stage. Much simpler models result if the gas turbine is decomposed into just three sections corresponding to the main turbine components, i.e. compressor, combustor and turbine, as in (Hussain and Seifi, 1992). Instead of applying the fundamental conservation equations, as described above, another modeling approach is to characterize the gas turbine performance by utilizing the real steady state engine performance data, as in (Hung, 1991). It is assumed that transient thermodynamic and flow processes are characterized by a continuous progression along the steady state performance curves, which is known as the quasi-static assumption. The dynamics of the gas turbine, e.g. combustion delay, motor inertia, fuel pump lag etc. are then represented as lumped quantities separate from the steady-state performance curves. Very simple models result if it is further assumed that the gas turbine is operated at all times close to the rated speed (Rowen, 1983). This model was utilized in a simulation of an island grid (Sharma, 1998). System simulation based on dynamic coupling equations is widely assumed and introduced (Schobeiri et al., 1994; and Garrard, 1996) with respect to aero engines and (Botros et al., 1991; Botros, 1994) with regard to compressor stations. For control purposes, fast simulation is demanded and the model configuration is kept as simple as possible. Air at the atmospheric pressure enters the gas turbine at the compressor inlet. After compression of the air to achieve the most favorable conditions for combustion, the fuel gas is mixed with the air in the combustion chamber. Then, the combustion takes place and the hot exhaust gases are expanded through the turbine to produce the mechanical power. In terms of energy conversion, the chemical energy present in the combustion reactants is transferred to the gas stream during combustion. This energy - measured in terms of gas enthalpy- is then converted into the mechanical work by expanding the gas through the turbine. Thus the excess mechanical power available for application elsewhere, after accounting for the power required to drive the compressor, is derived ultimately from the combustion process. Without combustion, assuming 100 percent efficient compressor and turbine operation, the power developed by the turbine would be exactly matched by the power required to drive the compressor. The main modeling assumptions are as follows:

Gas Turbines and Electric Distribution System

211

(1) air and combustion products are treated as perfect gases (2) specific heats are assumed constant for combustion products, air and injected steam (3) flow through nozzles is described by a one dimensional adiabatic uniform polytropic process (4) energy storage and transport delay in the compressor, turbine and combustion chamber are relatively small, thus steady state equations are applied. (5) inlet kinetic energy of gas flows into the compressor and turbine are treated as negligible (6) air mass flow through the compressor is controllable via inlet guide vanes. The compressor is described by one dimensional steady flow nozzle equation for a uniform polytropic compression. This can be expressed as (Jurado and Cano, 2004):
1/2 ma +1 2ma ma 2/ ma wa = Ao r rc p i cin c c ( ma 1)

(1)

Compressor power consumption equation is given by

Pc =

ctrans

wa hIC

(2)

Combustion energy equation is expressed as

+ wis c ps (298 Tis ) = 0

w g c pg (TTin 298 ) + w f h25 + wa c pa (298 Tcout ) +

(3)

Power delivery equation is written as

PT = T wg hIT Pm = PT Pc

(4) (5)

Figure 1 shows the block diagram of the gas turbine. The concept of the gas turbine control system, which is applied in this paper, is based on the Speedtronic Mark 4 description as presented in (Rowen, 1988). The fuel flow out from the fuel systems results from the inertia of the fuel system actuator and of the valve positioner. The fuel system actuator equation is:

wf =

kf e f s +1 1

(6)

212 The valve positioner equation is:

Francisco Jurado

e1 =

a Fd bs + c

(7)

The turbine torque function is given by:

T = k LHV ( w f 0.23 ) + 0.5( )

(8)

where the input variable to the fuel system is Fd. The output variable from the fuel system model is wf . A single gas turbine does not require the digital setpoint feature. The kLHV factor depends on the LHV. The kLHV and 0.23 factors cater for the typical turbine power/fuel rate characteristic, which rises linearly from zero power at 23 % fuel rate to rated output at 100 % fuel rate.

Figure 1. Block diagram of gas turbine control model.

Equation (8) allows the turbine torque to be calculated algebraically. This torque is used in the equations which model the mechanical system:

Pm = TN

(9)

In this paper, input variable to the turbine is wf and output variable from the turbine is N.

Gas Turbines and Electric Distribution System

213

3. HAMMERSTEIN MODEL
Since MPC is a model-based control an internal model is needed to predict the future responses of the plant to control inputs. As the CLM is a very large and complicated model, a new model is developed to be used in the MPC. Following the model structure proposed in (Jurado, 2005), a Hammerstein model of a gas turbine is applied that meets the above specifications. The model is designed to replicate both transient and steady state performance. Consider the Hammerstein model shown in Figure 2, where u(t), v(t), y(t) and yf(t), are the system input, noise, output and filtered output, respectively. x(t) denotes the unavailable internal signal. These are continuous time signals. u(iTs) and yf(iTs) denote the sampled input and sampled filtered output signals respectively with the sampling interval Ts. The filter is a lowpass filter at the designers disposal.

Figure 2. Hammerstein model.

The goal of the frequency domain identification is to apply inputs of the form,

u ( t ) = A cos (k t ) ,

k 0,

t [ 0, T ]

(10)

( s ) based on the finite sampled and then, to determine a pair of the estimates f (.) and G inputs and filtered outputs u(iTs) and yf (iTs) so that

( .) f ( .) , f

(s) G (s) G

(11)

( s ) , not its discretized model, is our in some sense. Note that the continuous time model G

interest.

214

Francisco Jurado

4. MODEL PREDICTIVE CONTROL


4.1. Theoretical Background
Model Predictive Control (MPC) refers to a class of control algorithms in which a dynamic process model is used to predict and optimize system performance. MPC is rather a methodology than a single technique. The methodology of controllers belonging to the MPC family is characterized by the following strategy illustrated in Figure 3.

Figure 3. Strategy of model predictive controller.

As shown in Figure 3, in MPC, the future outputs (gas turbine speed) for a determined prediction horizon Hp are predicted at each instant k using a prediction model. These predicted outputs y ( k + j ) , j = 1,..., H p depend on the state of the model at the current time k (given, for instance, by the past inputs and outputs) and on the future control signals

u (k + j) .

The control signal (mass flow) change only inside the control horizon, Hc, remaining constant afterwards,

u ( k + j ) = u ( k + H c 1) , j = H c ,..., H p

(12)

The set of control signals is calculated by optimizing a cost function in order to keep the process as close as possible to the reference trajectory (gas turbine speed reference),

Gas Turbines and Electric Distribution System

215

( k + j ) , j = 1,..., Hp. This criterion usually requires a quadratic function of the errors
between the predicted output signal and the reference trajectory. The control effort is included in the objective function in most cases. An explicit solution can be obtained if the criterion is quadratic, the model is linear and there are no constraints. Otherwise an iterative optimization method has to be used. In practice all systems are subject to restrictions. The actuators have a limited field of action, as in the case of valves. Constructive reasons, safety or environmental ones can cause limits in the system variables such as fuel flow or maximum temperatures and pressures. All of them lead to the introduction of constraints in the MPC problem. Usually, input constraints like

umin u ( k + j ) umax , j = 1,..., Hc


umin u ( k + j ) umax , j = 1,..., Hc 1

(13) (14)

are hard constraints in the sense that they must be satisfied. Conversely, output constraints can be viewed as soft constraints because their violation may be necessary to obtain a feasible optimization problem:

ymin y ( k + j ) ymax , j = j1 ,..., H p


where j1 represents the lower limit for output constraint enforcement.

(15)

4.2. Linear Model Based Predictive Control


The basic idea is to use the linear model to predict the future system behavior. This model is used throughout the entire prediction horizon. Even if this model is very accurate at the linearization point, its accuracy decreases over the prediction horizon. As a consequence, there may be a significant prediction error at k + H p .

4.3. Hammerstein Model Based Predictive Control


Due to the relatively simple block-oriented structure, the application of Hammerstein models in MPC is more straightforward than the application of the general Nonlinear AutoRegressive Exogenous (NARX) or NARMAX models. In this section, the Hammerstein model is implemented in MPC by inverting the static nonlinear model element f(u), as indicated in Figure 2. As the remaining part of the prediction model is the linear dynamic part of the Hammerstein model, the MPC optimization can be solved by quadratic programming. The combination of the inverse static nonlinear model element and the nonlinear system results in a transformed dynamical system. This system is linear if the system is of the

216

Francisco Jurado

Hammerstein type. As the inversion of the single-input single-output and multiple-input single-output static nonlinear model element is a straightforward analytical procedure, the computational demand of the controller is quite comparable to the linear generalized predictive control (GPC). This is a significant advantage compared to other nonlinear models which require the use of nonlinear programming or linearization techniques. In order to cope with the model-plant mismatch and also with disturbances (load changes), the internal model control (IMC) scheme (Garcia and Morari, 1982) is used. The resulting scheme is depicted in Figure 4.

Figure 4. Hammerstein model based predictive control scheme.

4.4. Optimization
In general, the GPC algorithm computes the control sequence u ( k + j ) , j = 1,..., Hc , such that the following quadratic cost function is minimized:

J ( H p1 , H p 2 , H c , ) =

j = H p1

( k + j )) ( ( k + j ) y

H p2

+ u 2 ( k + j 1)
j =1

Hc

(16)

( k + j ) denotes the predicted system output, ( k + j ) the modified setpoint that Here, y
is assumed to be known in advance, H p1 is the minimum costing horizon, H p 2 is the maximum costing or prediction horizon, H c is the control horizon, and suppression coefficient.

is the move

5. COMPONENTS OF GAS SYSTEM


Delivering the natural gas from a gas wellhead to end customers is comprised by a tremendous segment of the gas industry, which implicates gas wells, transmission and

Gas Turbines and Electric Distribution System

217

distribution pipelines, underground storages, compressors, and valves (Tobin, 2001; California Energy Comm., 2001). Gas Well: A gas well is usually located at sites which are far from load centers. Gas wells can be sorted into offshore and onshore. Transmission Pipelines: Transmission pipelines tackle the duty of transporting natural gas from wellheads or producers to local distribution companies or directly to large commercial and industrial users. Distribution Pipelines: Distribution pipelines generally allow the final link in the natural gas delivery chain. Distribution pipelines, which constitute the largest section in the natural gas system, deliver natural gas from city gate stations, underground storage facilities, and other gas supply sources to local industrial, as well as commercial and residential, customers. These pipelines work at a lower pressure level than transmission pipelines and offer different pressure services for different customers by adjusting the associated pressure regulators. for example, pipelines connected to gas-burning power plants require high-pressure services. Nevertheless, residential customers would need low-pressure gas for appliances. Underground Storage: Unlike electric power systems, which must uninterrupted monitor the entire system and adapt to changes instantaneously as electricity demand fluctuates, the gas industry can inject gas into certain underground storage facilities during off-peak periods for mitigating the high demand during peak hours and maintaining a steady flow through other pipelines when contingencies occur. Compressor: A compressor operates similar to step-up transformers in electric power systems. As gas is carried through a pipeline, its pressure would drop. Thus, the compressor must be an essential component in natural gas systems to maintain the desired pressure level in the transmission and distribution pipelines. Other compressors can be installed along pipelines (ordinarily at 50100-mi intervals). The optimized location of compressors in pipeline planning could diminish the operation cost dramatically, improve the market competition, and assure a reliable gas supply to customers. Valve: A valve is a protective device which serves similar to breakers, fuses, and switches in electric power systems. It can insulate faulted sections and maintain the operation of other components in natural gas systems by holding a desired pressure level.

6. GAS STEADY-STATE EQUATIONS


The steady-state flow of gas in a pipeline may be represented by equations that vary according to the gas working pressure and friction. These factors influence the gas flows that can vary from small values, in low-pressure distribution systems, to vary 1arge values, in high-pressure transportation systems. The effects of friction are difficult to measure and are the main reason for variations in the flow equations. The friction factor is not a constant for a given section of a pipeline and it is dependent on the roughness of the internal pipe surface, gas velocity, gas density, gas viscosity and the internal diameter of the pipe. After defining the gas operating conditions, the problem of static simulation is to estimate the values of pressure at the nodes and the flows in the individual pipes for known values of sources pressures and of gas consumption in the nodes. The pressures at the nodes and the

218

Francisco Jurado

flow in the pipes first satisfy the flow equation, and together with the values of loads and values of sources must accomplish the similar Kirchhoffs laws for the electrical systems. The above general flow equation implies assumptions which are: 1. 2. 3. 4. Isothermal flow due to insignificant temperature changes. Negligible kinetic energy change and constant compressibility across the pipe. Validity of the Darcy friction loss relationship across the pipe. Constant friction coefficient along the pipe length.

Under normal conditions, either an isothermal or an adiabatic approach is assumed. For the case of slow transients caused by fluctuations in demand, it is adopted that the gas in the pipe has sufficient time to reach thermal equilibrium with its constant-temperature environment. Likewise, when rapid transients were under consideration, it was assumed that the pressure changes occurred without any delay, allowing no time for heat transfer to take place between the gas in the pipe and the surroundings. Sometimes, this supposition of a process having a constant temperature or is adiabatic is not valid. The DarcyWeisbach equation for the friction loss in pipes yields better precision than other equations such as HazenWilliams because the friction factor it involves is determined as a function of both the relative pipe wall roughness and the Reynolds number. The others, like HazenWilliams, Manning, and Scobey assume that the flow is in the rough pipe zone and neglect the effect of Reynolds number. Many methods of meshed gas flow simulation may be used, such as, the Newton-nodal method, Hard-Cross nodal method, Newton-loop method and Hard-Cross loop method. The Newton-loop method has a respectable convergence compared with the other ones (Gay, 1971; Gay and Preece, 1975; Cochran, 1996). Three basic types of entities are considered for the modeling of natural gas transmission network: pipelines, compressor stations, both of which are symbolized by branches, and interconnection points, represented by nodes (Osiadacz, 1989; Wong and Larson, 1968; Olorunntwo, 1981; Wu et al., 2000; Carter, 2001).

6.1. Flow Equation


For isothermal gas flow in a long horizontal pipeline, say number k, which begins at node i and ends at node j, the general steady-state flow rate is often expressed by the following formula (Osiadacz, 1989) derived from energy balance:

f k = f kij = Sij 6.18* 10


where

T0 0

Sij

2 i

5 - 2 j Dk

(17)

Fk GLk Tka Z a

Gas Turbines and Electric Distribution System

219

Sij

+1 if i - j > 0 -1 if i - j < 0

In equation (1), the friction factor Fk depends on the flow region (laminar flow, mixed or transition flow, or fully turbulent flow), For fully turbulent flow (Reynolds number 4000) region in a high-pressure network, Weymouth suggested that the friction factor Fk varies as a function of the diameter Dk ( Weymouth, 1942)

Fk =

0.1089 Dk 3
1

(18)

In terms of field units, equation (17) becomes

f k = f kij = 8.41* 10 7 Sij M k Sij i2 - 2 j


where

(19)

Mk =

1.22* 10 10 T0 Dk 3 0 GLk Tka Z a

As suggested in equation (19), the gas flow can be found once i and j are known for given conditions. Equation (19), known as Weymouth flow equation, is most acceptable for large diameter ( 0.254 m) lines with high pressures.

6.2. Compressor Modeling


During transportation of gas in pipelines, the gas flow loses a part of its initial energy due to frictional resistance which results in a loss of pressure. To compensate the loss of energy and to move the gas, compressor stations are established in the network. The key factor to establish the representation of the centrifugal compressor is the horsepower consumption, which is a function of the amount of gas that flows through the compressor and the pressure ratio between the suction and the discharge. After empirical adjustment to describe for deviation from ideal gas behavior, the actual adiabatic compressor horsepower equation (Olorunntwo, 1981) at T0 = 15.5 C (= 288.65 K) and 0 = 101.00819 kPa becomes

220

Francisco Jurado

H k = H kij

jc = 0.0155 Bk f k ic

1 Z ki

(20)

where

Bk =

1972.47Tki k 1

6.3. Conservation of Flow


The mass-flow balance equation at each node can be expressed in a matrix form as

( A+U ) f
where

+ w T = 0

(21)

if branch k enters node i, +1, Aik = -1, if branch k leaves node i, 0, if branch k is not connected to node i. +1, if the kth compressor has its outlet at node i, U ik = -1, if the kth compressor has its inlet at node i, 0, otherwise.
+1, if the kth turbine gets gas from node i, Tik = 0, otherwise.

The matrix A, known as the branch-nodal incidence matrix (Osiadacz, 1989), corresponds the interconnection of pipelines and nodes. In addition, it is defined the matrix U, which describes the connection of compressors and nodes. The vector of gas injections w is found by

w = wS wL

(22)

Thus, a negative gas injection means that gas is taken out of the network. The matrix T and the vector represent where gas is withdrawn to power a gas turbine to operate the compressor. Therefore if a gas compressor, say k, between nodes i and j, is driven by a gasfired turbine, and the gas is tapped from the suction pipeline i, the following representation is obtained:

Tik = +1,

T jk = 0,

and k = amount tapped

Gas Turbines and Electric Distribution System Conversely, if the gas were tapped at the compressor outlet, it would have
Tik = 0, T jk = +1, and k = amount tapped

221

Analytically, k can be approximated as


2 k = Tk + Tk H kij + Tk H kij

(23)

where H k = H kij is the horsepower demanded for the gas compressor k in equation (20).

6.4. Power Losses


During transportation of gas in pipelines, the gas stream loses a part of its initial energy due to frictional resistance which results in a loss of pressure. The losses of gas distribution system can be indicated as

where

NP k=1

f k i - j

(24)

N P branches in the system.

7. ELECTRIC POWER LOSSES


Differences between natural gas and electricity systems are established as follows. Electricity displaces at the speed of light, while natural gas travels 4060 mi/h. Electricity is not a storable article of commerce. So the contingency-constrained network flow operation could forbid transmission systems from employing their maximum capacity. Consequently, the value of a transmission line may not necessarily be reflected in its current flow. The possibility to store gas in tanks and in pipelines palliates this problem for gas. Natural gas utilities typically trust on the natural gas storage to increase supplies flowing through the pipeline system and to meet the total natural gas demand. Economies of scale are very great in electric power transmission projects. It is much cheaper to install the required capacity of a transmission line initially than to retrofit the line later. However, gas pipelines are normally operated at a lower pressure and the pressure is elevated later to obtain additional capacity. Natural gas pipeline flows can be operated independent of the gas network constituents.

222

Francisco Jurado

Electric power systems may suffer significant losses. The losses depend on the line resistance and currents and are usually referred to as thermal losses. While the line resistances are fixed, the currents are a complex function of the system topology and the location of generation and load. Consider the well known power flow equations, with complex power Si = P i + jQi , injected at bus i as (Grainger and Stevenson, 1994)

Pi = Vi j =1 YijV j cos ( i j ij )
n

(25)

Qi = Vi j =1 YijV j sin ( i j ij )
n

(26)

In this article, only the real power injections as they relate to electric losses are of concern. The system losses can be expressed as

PL = i =1 PGi i =1 PDi
n n

(27)

8. RESULTS
8.1. Identification of the Gas Turbine Model
A second-order term is sufficient to model the static nonlinear behavior of the engine and the linear part is a second order transfer function. The noise v(t) is a random signal uniformly distributed and the input is

u ( t ) = A cos (i t ) ,

A = 1,

i = 1, 2,3

with 1= 0.6, 2=1.2, 3=6 and Ti= 100((2)/i). For the input frequency i, the sampling interval is set to be /(50i). No lowpass filter is used in simulations, i.e., y(t) = yf(t). Thus, ( s ) are given by (.) and G the estimates of f

f ( u ) = 0.002u2 + 0.816u + 3.458

G (s) =

0.066 ( s + 0.428 ) ( s + 0.932 )( s + 0.429 )

which are very close to the true but unknown f ( u ) and G ( s ) . The true (solid line) and the estimated (circle) nonlinearities are shown in Figure 5. They are basically indistinguishable.

Gas Turbines and Electric Distribution System

223

Figure 5. True (solid) and the estimated (circle) nonlinearities.

8.2. Model Based Predictive Control


The gas turbine-generator is natural gas operated, with the rating of 1 MW. The systems are modeled dynamically (MATLAB, 2007). The nominal operating conditions of the gas turbine and generator considered in this paper are given in Table 1 (Brown Energy Systems, 2007). Permanent magnet generator provides rapid response to varying loads and constant excitation under all conditions. The gearbox is created to ensure optimal performance. The control system displayed in Fig. 4 is applied to enable the control of speed of the gas turbine. A future control trajectory is generated as a possible solution by the optimizer based on the Hammerstein model using proposed method. At each sampling instant, only the first predicted input signal from the obtained control trajectory is applied to control the gas turbine. The load is the major disturbance affecting the gas turbine. The amount of mass flow can be controlled according to the load. Hammerstein MPC is employed to illustrate the performance of the gas turbine. The MATLAB implementation of quadratic programming is used (Coleman and Branch, 2004). The MPC parameters are selected according to the tuning rules given in (Soeterboek, 1992). Minimum prediction horizon H p1 is always set to the model time-delay d. There is no reason for choosing it smaller because the d-1 first predictions depend on past control inputs only and cannot be affected by the first action u(t). From another point of view, it is not recommended to select it bigger because this can lead to unpredictable results. For the gas turbine, it is set to l (sampling period) and not tuned. A rule of thumb is that the prediction horizon H p 2 should be taken close to the rise time of system (Clarke et al., 1987). Nevertheless, often it is not possible to choose it this long since the calculation time required by MPC is too demanding. Commonly it is tuned through empirical observation.

224

Francisco Jurado Table 1. Operating point data


GAS TURBINE 1000 kW 25 % Natural gas 0.07 kg/s 100 r/s 950 F Synchronous, brushless 1500 kVA, 3 phase 1200 kW, PF 0.8 30 r/s 480/460 V 1806/1884 A 100 r/s -30 r/s

Power output Efficiency (LHV) Fuel Mass flow Turbine speed Exhaust temperature Permanent magnet generator Type Rated Capacity (kVA) Rated Capacity (kW) Speed Voltage Current Gearbox

From repeated tests on the gas turbine, H p 2 is set around 30 (sampling periods) for the best control performance for both small and large random step changes. The sample time is limited to 0.1 s. Soeterboek advises H c is equal to the number of output lag terms (Soeterboek, 1992). If

H c is made longer, the control performance is slightly ameliorated and the calculating time is
also increased. Based to the simulation results, it is set to 2 (sampling periods), which is the same as the number of output lag terms. The purpose of the move suppression coefficient is to punish large changes in the process input and reduce actuator wear. It is usual to set as a constant in the range [0, 1]. For the gas turbine, to achieve the best control performance, it is set to 0.05. With the MPC variables set to H p1 =1, H p 2 =30, H c =2 and =0.05, the controller results in the system responses are shown in Figures 6 and 7. The responses are for large random step changes. The results depict that a fast rise rime is attained, with almost no overshoot, evidencing proposed method offers a near optimal performance for both small and large random step changes. This result illustrates the capabilities of the MPC controller to track a reference trajectory.

8.3. Simulation Results


The IEEE 13 node test feeder (Kersting, 2001) is a 4.16 kV short and highly loaded feeder. Also, its overhead and underground lines, shunt capacitor banks and spot and distributed loads provide a useful distribution system model.

Gas Turbines and Electric Distribution System

225

Figure 6. Performance of MPC on several large random set point changes. Comparison between rotational speed and speed reference.

Figure 7. Performance of MPC on several large random set point changes. Mass flow for set point changes.

226

Francisco Jurado

This node test feeder is used as the test system to investigate the dynamic characteristics of the distribution system with two gas turbines and the effectiveness of the MPC on the stability of distribution system. Figure 8 shows this test system.

Figure 8. One line diagram of IEEE 13 node feeder with gas turbines.

Figure 9. Rotor speed deviation of GT1.

Gas Turbines and Electric Distribution System

227

The two gas turbines are connected at nodes 652 and 634, the initial active power of gas turbine GT1 is 1 p.u., and initial active power of gas turbine GT2 is 0.5 p.u. To investigate the dynamic behavior of the gas turbines under large disturbance conditions, a three-phase fault is applied on node 671 for 400 ms. At t = 0 s, the fault occurs at node 671 and the fault is cleared at t= 0.4 s. Figures 9 and 10 show the responses of rotor speed of the gas turbines to the three-phase fault when the MPC is applied in the governor-gas turbine system. GT1 has a greater oscillation than that of the GT2. This is because the initial power of GT1 is higher than that of GT2. The simulation results demonstrate that MPC can effectively damp the oscillation of both gas turbines, thus the MPC can ameliorate the dynamic characteristic of the whole distribution system.

Figure 10. Rotor speed deviation of GT2.

8.4. Distribution Systems


A series of comparisons between gas and electricity systems has been performed. It is of no use to consider a generic possible substitution of gas pipelines with electric lines: pipelines are clearly the basic solution for multipurpose gas utilization (electricity generation, industrial and domestic uses, petrochemicals, etc.), particularly when very large gas flow rates are involved. The same reliability/availability assumptions for the delivered electricity have been taken into account for both gas and electricity systems. This hypothesis is secure on the basis of the actual performance of both gas and electricity systems.

228

Francisco Jurado

The IEEE 37-bus test system shown in Figure 11, which can be considered as a distribution system, is applied to verify the method presented. A gas turbine is integrated into electric distribution network. While a gas well is incorporates into the gas distribution system. The electric power losses are obtained from the results of power flow studies using MATLAB. Figure 12 shows the electric power losses. In Figure 13 the power losses are reported as a function of the node (bus) and pipeline in the gas distribution system. The simulations have been developed by NEPLAN. The total power loss of the electricity distribution system reaches 34 kW, while the total power loss of the gas distribution system is only 2.131 kW. Lengths of electric lines and equivalent gas pipelines are reported in Table 2.

Figure 11. IEEE 37- bus test system.


IEEE-37 Bus
16000 14000 12000 10000 8000 6000 4000 2000 0
L1-2 L2-3 L3-4 L4-5 L5-6 L6-7 L7-8 L8-9 L9-10 L10-11 L11-12 L12-13 L12-26 L9-25 L25-35 L25-36 L7-24 L6-23 L6-22 L3-14 L14-15 L15-16 L16-33 L33-34 L16-30 L30-32 L15-28 L28-29 L30-31 L3-17 L17-18 L17-27 L4-19 L19-20 L20-21 L20-37

Power loss (W)

Line

Figure 12. Power losses of the IEEE 37-bus test system. Electricity distribution system.

Gas Turbines and Electric Distribution System


IEEE-37 Bus
300 Power loss (W) 250 200 150 100 50 0 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37

229

Location-Bus N
IEEE- 37 Bus 800 600 400 200 0

Power loss (W)

Pipeline

Figure 13. Power losses of the IEEE 37-bus test system. Gas distribution system.

Table 2. Lengths of electric lines and gas pipelines. IEEE 37-bus test system
Pipeline L-1-2 L-2-3 L-3-4 L-4-5 L-5-6 L-6-7 L-7-8 L-8-9 L-9-10 L-10-11 L-11-12 L-12-13 L-12-26 L-9-25 L-25-35 L-25-36 L-7-24 L-6-23 L-6-22 L-3-14 Length (m) 643,8 334 459 208 69,6 111,3 111,3 195 222,7 139 139 139,2 69,6 181 69,6 445 111,3 0,001 208 125

230

Francisco Jurado Table 2. (Continued)

Pipeline L-14-15 L-15-16 L-16-33 L-33-34 L-16-30 L-30-32 L-15-28 L-28-29 L-30-31 L-3-17 L-17-18 L-17-27 L-4-19 L-19-20 L-20-21 L-20-37

Length (m) 181 27,8 208,8 97,4 320,1 41,7 27,8 180,9 264,5 139 111,3 83,5 83,5 97,4 97,4 69,6

Figure 14 shows the next system studied. It represents a six-bus 25 kV distribution network with lines ranging in length from 16 to 32 km. Node 1 is the reference bus. It is assumed that nodes 1 and 2 are connected to gas turbines. They are voltage controlled buses.

Figure 14. IEEE 6- bus test system.

Gas Turbines and Electric Distribution System

231

Figure 15 displays the electric power loss and Figure 16 the power loss in the gas distribution system. The total power loss of the electricity system extends to 495 kW, whereas the total power loss of the gas system reaches 126.6 kW. Lengths of electric lines and equivalent gas pipelines are described in Table 3. The computation of losses is illustrated by means of Tables 4 and 5.
IEEE-6 Bus
200 Power loss (kW) 150 100 50 0 L2-3 L2-1 L5-6 L4-5 L3-4 L1-5 L1-6
4

Line

Figure 15. Power losses of the IEEE 6-bus test system. Electricity distribution system.

IEEE- 6 Bus
100 Power-Loss (kW) 80 60 40 20 0 3 2 1 6 5 Location-Bus N

IEEE-6 Bus
50 Power loss (kW) 40 30 20 10 0 L-2-3 L-1-2 L-5-6 L-4-5 L-3-4 L-1-5 L-1-6

Pipeline

Figure 16. Power losses of the IEEE 6-bus test system. Gas distribution system.

232

Francisco Jurado Table 3. Lengths of electric lines and gas pipelines. IEEE 6-bus test system
Pipeline L-1-2 L-2-3 L-3-4 L-4-5 L-5-6 L-1-6 L-1-5 Length (km) 16 16 16 16 16 17,6 32

Table 4. Lengths of electric lines and gas pipelines. IEEE 30-bus test system
Pipeline L1-2 L1-3 L2-4 L2-6 L2-5 L4-6 L3-4 L5-7 L8-28 L6-28 L6-8 L7-6 L12-14 L12-15 L12-16 L14-15 L16-17 L15-18 L18-19 L19-20 L10-20 L10-17 L10-21 L10-22 L21-22 L15-23 L23-24 L22-24 L24-25 L25-27 L27-29 L27-30 L29-30 Length (km) 71,5 91,9 132,8 204,3 204,3 102,2 51,2 26,6 204,3 204,3 106,2 30,6 61,3 34,7 28,6 34,7 53,1 32,7 36,8 16,3 53,1 38,8 46,9 40,8 18,4 40,9 51,1 47,0 20,4 30,6 30,6 61,3 57,2

Gas Turbines and Electric Distribution System Table 4. (Continued)


Pipeline L25-26 L6-9 L6-10 L4-12 L27-28 L9-10 Length (km) 51,1 0 0 0 0 0

233

Table 5. Computation of losses. Electricity distribution system


Line flow (kW) 15 8.140 5.620 -15 7.913 -7.760 3.754 -3.718 -3.518 3.550 -2.988 -5.524 3.009 -7.983 495

L1-2 L1-6 L1-5 L2-1 L2-3 L3-2 L3-4 L4-3 L4-5 L5-4 L5-6 L5-1 L6-5 L6-1 Losses (kW)

The power losses depend on the ambient temperature as formulated in Sections 6 and 7. Figure 17 displays the power losses for different temperatures in this gas distribution system.

Figure 17. Power losses for different temperatures. Gas distribution system.

234

Francisco Jurado

8.5. Subtransmission System


The proposed method is tested on the IEEE 30-bus test system shown in Figure 18, which can be considered as a meshed subtransmission system. The system has 30 buses (mainly 132- and 33-kV buses) and 41 lines. Six gas turbines are considered to be connected to the system.

Figure 18. IEEE 30- bus test system.

Gas compressors stations constitute a major part of the operational plant on each transmission system. Their purpose is to restore the gas pressure reduction induced by frictional pressure losses. The compressors are driven mostly by gas turbines which use natural gas as fuel, taken directly from the transmission pipelines. The compressor unit comprises three main components, a gas generator, a power turbine and a centrifugal gas compressor. The maximum shaft powers of the units range from 5.5 MW to more than 20 MW. At each compressor station, there are installed between two and three centrifugal compressors, driven by gas turbines. In this paper, the compressors are installed at 60 mi intervals. Figure 19 displays the electric power loss and Figure 20 the power loss in the gas distribution system. The total power loss of the electricity system corresponds to 17.86 MW,

Gas Turbines and Electric Distribution System

235

however the total power loss of the gas system represents 202.34 MW. Lengths of electric lines and equivalent gas pipelines are depicted in Table 6.
IEEE-30 Bus
6 Power loss (MW) 5 4 3 2 1 0
L1-2 L5-7 L7-6 L6-8 L27-28 L27-29 L29-30 L27-30 L27-25 L25-26 L1-3 L3-4 L4-6 L4-12 L14-15 L12-14 L16-17 L12-16 L24-22 L23-24 L15-23 L21-22 L10-21 L22-10 L20-10 L19-20 L6-10 L2-4 L12-13 L9-10 L9-11 L6-9 L18-19 L15-18 L12-15 L24-25 L17-10 L2-6 L2-5 L8-28 L6-28

Line

Figure 19. Power losses of the IEEE 30-bus test system. Electricity distribution system.

IEEE-30 Bus
100 Power loss (MW) 0 1 50

11

13

15

17

19

21

23

25

27

29

Location-Bus N
IEEE-30 Bus

Pip eline

Figure 20. Power losses of the IEEE 30-bus test system. Gas distribution system.

It is quite clear that in distribution systems, due to better efficiency of gas system, the losses are larger in case of electricity system. For the shortest lengths here considered (IEEE 37-bus test system) the losses are the lowest ones. For the longest lengths (IEEE 30-bus test system), the losses are larger in case of gas system. Also for the shorter distances the gas pipeline feeding a local power plant in the consumption area is an interesting and attractive option to be considered with respect to a distribution of electricity.

236

Francisco Jurado Table 6. Computation of losses. Gas distribution system


fk (m3/s) 0 0 0.345 0.625 0.431 0.431 (mbar) 0 0 653.62 2255.12 724.23 3.7705 Losses (kW) 0 0 13.529934 84.6421707 18.7454865 9.75931083 126.676902

N-1 N-2 N-3 N-4 N-5 N-6

CONCLUSIONS
In this article, a Hammerstein model of a gas turbine plant and its MPC has been presented. The model is suitable for use in power system stability studies. A MPC is designed for the gas turbine in order to improve system dynamic performance. The proposed model was tested on a simple distribution system. The simulation results with and without MPC are compared. It was observed that the proposed model with MPC improves the dynamic performance of the system. Due to the continuous developments of gas turbines, combined cycle power plants allow a very high efficiency, low emissions and very attractive investments cost. This and environmental concerns are enhancing gas consumption for electric power generation. The performed comparison between gas and electricity systems, of interest when natural gas is needed for electricity generation, highlights that in the investigated cases the power losses are larger in case of electricity transmission over short distances than in case of gas transport and electricity generation close to final users. However for longer distances the use of electric transmission systems is an attractive option to a gas pipeline feeding a power plant located into the consumption area.

REFERENCES
Botros K.K., Campbell P.J., Mah D.B., 1991. Dynamic simulation of compressor station operation including centrifugal compressor and gas turbine. Journal of Engineering for Gas Turbines and Power-Transactions of the ASME, Vol. 113, No. 2, pp. 300311. Botros K.K., 1994. Transient phenomena in compressor stations during surge. Journal of Engineering for Gas Turbines and Power-Transactions of the ASME, Vol. 116, No. 1, pp. 133142. Brown Energy Systems, 2007. New 1 MW Multi-Fuel Gas Turbine Generator. Available: http://www.brownmarine.com/tg01.htm. California Energy Comm., 2001. Natural gas infrastructure issues, Sacramento, CA. Carter R., Goodreau M., Rachford H., 2001. Optimizing pipeline operations through mathematical advances. Pipeline and Gas Journal, Vol. 228, No. 10, pp. 51-53.

Gas Turbines and Electric Distribution System

237

Chiras N., Evans C., Rees D., 2002. Global nonlinear modeling of gas turbine dynamics using NARMAX structures. ASME Journal of Engineering and Power, Vol. 124, No. 4, pp. 817-826. Clarke D.W., Mothadi C., Tuffs P.S., 1987. Generalized predictive control. Part I. The basic algorithm, Automatica, Vol. 23, No. 2, pp. 137-148. Cochran T.W., 1996. Calculate pipeline flow of compressible fluids. Chemical Engineering, Vol. 103, No. 2, pp. 115-122. Cohen H., Rogers G.F.C., Saravanamuttoo, H.I.H., 1998. Gas turbine theory. 4th ed. Longman, England. Coleman T., Branch M.A., Grace A., 2004. Optimization Toolbox, Version 2.2. The MathWorks, Inc., Natick, MA. DOE, 2001 a. Annual energy outlook 2002 with projections to 2020. Energy Inf. Admin. (EIA). DOE, 2001 b. U.S. natural gas markets: Recent trends and prospects for the future. Energy Inf. Admin. (EIA). Evans C., Rees D., Borrell A., 2000. Identification of aircraft gas turbine dynamics using frequency-domain techniques. Control Engineering Practice, Vol. 8, No. 4, pp. 457-467. Fawke A.J., Saravanamuttoo H.I.H., Holmes M., 1972. Experimental verification of a digital computer simulation method for predicting gas turbine dynamic behaviour. Institution of Mechanical Engineers Proc. Vol. 186, No. 27, pp. 323329. Garcia C.E., Morari M., 1982. Internal model control: 1. A unifying review and some new results. Ind. Eng. Chem. Process Design and Development, Vol. 21, No. 2, pp. 308-323. Garrard D., 1996. ATEC: The aerodynamic turbine engine code for the analysis of transient and dynamic gas turbine engine system operations, part 1: Model development. ASME paper 96-GT-193. Gay B., 1971. Middleton P. Solution of gas network problems. Chemical Engineering Science, Vol. 26, No. 1, pp. 109-123. Gay B., Preece P.E., 1975. Matrix methods for the solution of fluid network problems. Trans. of the Institution of Chemical Engineers, Vol. 53, No.1, pp. 12-15. Grainger J.J., Stevenson Jr. W.D., 1994. Power system ana1ysis. New York: McGraw-Hill, Inc. Hung W.W., 1991. Dynamic simulation of gas-turbine generating unit. IEE Proc.-C Generation Transmission and Distribution, Vol. 138, No. 4, pp. 342-350. Hussain A., Seifi H., 1992. Dynamic modeling of a single shaft gas turbine. Proc. of the IFAC Symposium on Control of Power Plants and Power Systems, Munich, Germany, Pergamon Press, pp. 43-48. Jurado F., Cano A., 2004. Use of ARX algorithms for modelling micro-turbines on the distribution feeder. IEE Proceedings Generation Transmission and Distribution, Vol. 151, No. 2, pp. 232-238. Jurado F., Carpio J., 2005. Enhancing the distribution networks stability using distributed generation. The International Journal for Computation and Mathematics in Electrical and Electronic Engineering (COMPEL), Vol. 24, No. 1, pp. 107-126. Jurado F., 2005. Modelling micro-turbines using Hammerstein models. International Journal of Energy Research, Vol. 29, No. 9, pp. 841-855. Kersting W.H., 2001. Radial distribution test feeders. Proc. IEEE/PES Summer Meeting, Vol. 2, pp. 908 912.

238

Francisco Jurado

Maciejowski J.M., 2002. Predictive Control with Constraints. Prentice Hall, London. MATLAB, 2007. Version 7. The Mathworks Inc., Natick, MA. Narendra K., Gallman P., 1996. An iterative method for the identification of nonlinear systems using a Hammerstein model. IEEE Trans. Automatic Control, Vol. 11, No. 3, pp. 546550. NEPLAN, 2005. Power Systems Engineering, Erlenbach, Switzerland. Olorunntwo F.O., 1981. Natural gas transmission system optimization. PhD thesis, The University of Texas at Austin. Osiadacz A.J., 1989. Simulation and Analysis of Gas Network. Houston: Gulf Publishing Company. Qin S.J., Badgwell T.A., 2000. An overview of nonlinear predictive control applications, in Allgwer, F., and Zheng, A. (Eds.), Nonlinear Model Predictive Control. ser. Progress in Systems and Control Theory, MA: Birkhuser, Vol. 26, pp. 369392. Richalet J., 1993. Industrial applications of model based predictive control. Automatica, Vol. 29, No. 5, pp. 12511274. Rowen W.I., 1983. Simplified Mathematical Representations of Heavy-Duty Gas Turbines. ASME Journal of Engineering for Power, Vol. 105, No. 4, pp. 865-869. Rowen W.J., 1988. Speedtronic Mark IV control system. Alsthom Gas Turbine Reference Library, AGTR 880. Schobeiri M.T., Attia M., Lippke C., 1994. GETRAN: A generic, modularly structured computer code for simulation of dynamic behavior of aero- and power generation gas turbine engines. Journal of Engineering for Gas Turbines and Power-Transactions of the ASME, Vol. 116, No. 3, pp. 483494. Sharma C., 1998. Modeling of an Island Grid. IEEE Trans. Power Systems, Vol. 13, No. 3, pp. 971-978. Soeterboek A.R.M., 1992. Predictive Control; A Unified Approach. Prentice-Hall, Upper Saddle River, NJ. Tobin J., 2001. Natural gas transportation-infrastructure issues and operational trends. Energy Inf. Admin. (EIA)/Natural Gas Div. van Essen H.A., de Lange H.C., 2001. Nonlinear model predictive control experiments on a laboratory gas turbine installation. Journal of Engineering for Gas Turbines and PowerTransactions of the ASME, Vol. 123, No. 2, pp. 347-352. Vroemen B.G., van Essen H.A., van Steenhoven A.A., Kok J.J., 1999. Nonlinear Model Predictive Control of a Laboratory Gas Turbine Installation. Journal of Engineering for Gas Turbines and Power-Transactions of the ASME, Vol. 121, No. 4, pp. 629634. Weymouth T.R., 1942, Problems in natural gas engineering. ASME Trans., Vol. 34, pp. 185234. Willis H.L., Scott W.G., 2000. Distributed Power Generation: Planning and Evaluation. Marcel Dekker, New York. Wong P.J., Larson R.E., 1968. Optimization of natural-gas pipeline systems via dynamic programming. IEEE Trans. Automatic Control, Vol. 13, No. 5, pp. 475-481. Wu S., Rios-Mercado R.Z., Boyd E.A., Scott L.R., 2000. Model relaxations for the fuel cost minimization of steady-state gas pipeline networks. Mathematical and Computer Modelling, Vol. 31, No. 2-3, pp. 197-220.

In: Leading-Edge Electric Power Research Editor: C. M. OSullivan, pp. 239-263

ISBN: 978-1-60021-981-8 2008 Nova Science Publishers, Inc.

Chapter 5

MICRO CCHP: FUTURE RESIDENTIAL ENERGY CENTER


R. Z. Wang and D. W. Wu
Institute of Refrigeration and Cryogenics, Shanghai Jiao Tong University, China

ABSTRACT
Combined cooling, heating and power (CCHP) system, as a distributed energy system, can work all the year and provide cooling/hot-water/power in summer, heating/hot-water/power in winter and hot-water/power in other seasons. In CCHP systems, the total energy efficiency increases to over 85%, while the average energy efficiency of conventional fossil fuel fired electricity generation systems is around 40%. The energy efficiency promotion of CCHP systems results in emission reduction compared to the conventional methods of generating heat and electricity separately. And as a distributed energy resource, CCHP systems also increase in the reliability of the energy supply. With the overall development of CCHP systems and related technologies, the utilization of micro CCHP systems in the residential sector is emerged as a growing potential. The article focuses on the micro CCHP systems for single-family applications (around 10 kW) and multi-family or residential district applications (under 200 kW). The status quo of micro CCHP systems is briefly presented and diverse combinations of technologies existing in applications or experimental units are listed through comprehensive literature review. Various technologies available or under development are introduced, such as reciprocating internal combustion engine, micro-turbine, fuel cell, Stirling engine, absorption chiller, adsorption chiller and so on. Afterward, the tendency and issues of micro CCHP systems are discussed. The review shows that micro-CCHP applications are entering into average families as a next-generation residential energy supply center.

240

R. Z. Wang and D. W. Wu

INTRODUCTION
The conventional way to cover electricity, heating and cooling demands is to purchase electricity from the local grid, generate heat by burning fuel in a boiler and obtain space cooling power from diverse air-conditions. But in a CCHP system, byproduct heat that can be up to eighty percent of total primary energy in combustion-based electricity generation is recycled for different uses. CCHP, generally, is defined as combined production of electrical and useful thermal energy from the same primary energy source [1]. In some literatures, CCHP systems are also named as Tri-generation and BCHP (Building Cooling Heating and Power) systems.

Figure 1. Energy Flow of Traditional mode.

Figure 2. Energy Flow of CCHP System.

Micro CCHP: Future Residential Energy Center

241

Recent development of CCHP systems, to large extent, is related to the emergence of DER (Distributed/Decentralized Energy Resources) - a novel technical concept in the region of energy supply. DER is defined as an electricity generation system located in or near users, providing electrical and thermal energy synchronously to meet the demands of local users in priority. The CCHP systems discussed in this article is in the category of DER, and they are different from traditional CHP systems, which, mostly, are large-scale centralized power plants without cooling power generated. Distributed CCHP systems share some advantages [2,3,4,5] compared to traditional energy supplies, along with their developing tendency and promising prospect. First of all, overall energy efficiency is dramatically improved, ranging from 70% to more than 90% compared to up to 40-45% of typical centralized power plants. With the primary energy saving, vast cost reductions including fuel costs, transmission and distribution savings can be achieved. Secondly, environment benefits derive from emission reduction. This benefit can be viewed from two aspects sorted by different kinds of prime movers. Some prime movers with new technologies like fuel cells, micro-turbines do expel much less emissions including NOx, CO2 than the traditional technologies taken by centralized power plants do. However, other prime movers equipped in CCHP systems with smaller capacity as their same kinds of large counterpoints in centralized power plants, emit a bit more amount of NOx and CO2 per kW electricity generated. Nevertheless, energy efficiency promotion of CCHP systems should be taken into account at this time. Burning much less fuel to meet same demands results significant emission reduction, which surely exceeds the impact of emission augment caused by slight decrease of small-scale prime mover convert efficiency. Last but not the least, CCHP systems increase in the reliability of the energy supply. Obviously, generation/distribution system malfunctions, terrible weather and terrorism are fatal threatens leading to disruptions of centralized power plants. A smaller, more flexible and dispersed system as distributed CCHP unit is possible to avoid these threatens being realities, and then limited influences and fast recovery could be achieved if these situations unluckily happened. A study following the 11th September attacks suggested that a system based more on distributed generation plants may be five times less sensitive to systematic attack than a centralized power system [6]. Distributed CCHP systems are suitable for various industrial, institutional, commercial and residential applications, and the capacities of these systems range widely from less than 1 kW in domestic dwellings to more than 1 MW in some industries or university campuses. Since last decade, many CCHP applications have been set up in hospitals, university campuses, commercial complexes, hotels, clubs, leisure centers, office buildings and residential districts. With the progress of some critical technologies such as fuel cells and micro-turbines, more and more literatures focus on small, even micro level CCHP systems recently, which are especially for single- family (1-5kW) or multi-family residential buildings and small residential districts, ranging from 5kW up to 200kW.

MICRO RESIDENTIAL CCHP SYSTEMS


Different prime movers with heat recovery equipments and further connecting with different cooling or dehumidification options can result in various kinds of CCHP systems,

242

R. Z. Wang and D. W. Wu

but only several modes of combination are adopted in recent commercial market, other promising possibilities still remain in laboratory to overcome their disadvantages in technology or economy. Reciprocating engines, micro-turbines, electrical chillers and absorption chillers are currently predominant for the maturity and stability of these technologies. Reciprocating internal combustion engines plus absorption or electrical (engine driven) chillers are popular for small scale utilizations. Jacket cooling fluids, lubricating oil systems, and engine exhaust are three heat recovery options which can produce hot water using exchangers, for heat demands and other cooling and dehumidification usages, seen in Figure 3. This kind of configuration has existed for quite a few years, and there are many applications ranging form 100kW up to 1MW in industrial, institutional, commercial sectors. But distributed CCHP applications with engines below 100kW for residential dwellings can be seldom found until recent years. In 2001, M.A. Smith [8] analyzes a micro CCHP system with an engine and a heat pump in his articles. The rated capacity of the engine is as low as 1.5kW and with the help of a heat pump the heat generated is around 4.5kW. Miguez [9, 10] also illustrates design and performance of a CCHP system with engine (9.6kW) and heat pump equipment. In 2004, a micro CCHP system at Shanghai Jiao Tong University with novel adsorption chiller generating cooling power, which uses heat recovered from a gas engine of 12kW rated capacity, was experimented [11]. Commercial compact micro CCHP products also can be found in US and Europe now. The ecopower micro CCHP unit of Marathon Engine Co. fueled by natural gas or propane gas, generates 2-5kW electricity and up to 13.8kW thermal power at the max. temperature of 75 . Typical applications of these units are single and multi-family homes [12].

Figure 3. Schematic of reciprocating engine heat recovery [6, 7].

Recently, a household size CCHP unit based on a small-scale diesel engine generator set (9.5kW) is reported [13]. An absorption refrigerator driven by the exhaust heat from the

Micro CCHP: Future Residential Energy Center

243

engine is as the cooling option of this system. It is point out in the paper that the CO2 emission per unit (kWh) of useful energy output from this micro CCHP system is dramatically reduced compared to that of conventional generations.

Figure 4. Schematic diagram of micro-turbine [6, 7].

Another popular prime mover, micro-turbines are classified into two categories: combustion turbine and Rankine cycle turbine according to literatures. Capstone Turbine Co. is the leading competitor in the field of micro combustion turbines. Current production microturbines range in net power output from 30 to 250kW. Their low maintenance and clean exhaust make them a reliable choice for base load CCHP applications. Integrating hot water heat recovery into the micro-turbine package has proven cost effective, and a growing number of commercial installations are saving money using this technology [14]. Absorption chillers and desiccant dehumidifiers driven by recovery heat of micro combustion turbines are employed to meet cooling demands of users. This configuration of CCHP systems is applied in many locations, especially in the US, where turbine-based units have become serious competitors with engine-based units in the CCHP market. But micro turbine unit is not applied separately for single dwellings. In most cases, several modularized minimal units of 30kW can be combined together to fit users electricity profiles, while they still share flexibility in operation. The capacity of most applications with micro combustion turbines, especially in hotels, hospitals and university campus buildings, are beyond the micro level. One obvious drawback of this technology is prominent high initial investment cost, which prevents its popularities in residential sector. Average residential building owners maybe cannot afford to this advanced technology. In the other category, some micro Rankine cycle turbines combined with solar collection are reported. Both W. Yagoub [15] and S.B. Riffat [16] introduced a solar energy-gas driven micro-CHP system. Solar energy collector of 25kW thermal capacity, supplemented by a condensing gas boiler, was used to drive a 1.5 kW Rankine cycle micro-turbine generator

244

R. Z. Wang and D. W. Wu

unit. The literatures of micro CCHP units with Rankine cycle micro-turbine are limited. This type of turbine is not a popular choice in the residential building field till now.

Figure 5. STM 4-120 power unit packaged DG system [17].

Among the newly emerged prime mover technologies, the Stirling engine is viewed as a promising prime mover in small commercial and residential applications for their low emissions, fewer moving parts, low noise, small-scale availability and relatively low byproduct heat. The Stirling engine CCHP systems are also suitable for modular installation as micro combustion turbines. And due to temperature limitations of the engine components, heat recovery from coolant systems account for almost 50% of the heat input. This results in a significant amount of heat suitable for space heating, cooking, potable hot water, and low temperature processes. The possible cooling and dehumidification options for Stirling engines are absorption chillers, dehumidifiers and adsorption chillers. There has also been research on the feasibility of CCHP driven by Stirling engines [18]. Currently, only a few commercial Stirling engine units can be found in the CCHP market. The most active company developing Stirling engine distributed generation technology is STM Power, Inc. STM has conducted field tests of this application with their 25kW model, STM 4-120 [17], which is the first commercialized Stirling engine in the world. The new prototype, STMs 55kW engine, can produce 92kW, while a 3kW engine will produce 6kW. Other companies in this industry reported systems under development that range from 55 watts to 3,000 watts. Sigma co. is developing a 3 kW electrical output and 9 kW thermal output engine for a single-family dwelling. The electrical efficiency of the unit is reported to be 25% [19]. And SOLO, a German Company has developed Stirling engine CHP unit fueled by natural gas. The unit generates electrical power of 29 kW and thermal power of 824 kW and has an overall efficiency of 9296% [20]. In Canada, a project was initiated to integrate a prototype micro CCHP unit into a residential house that would provide electricity and heat to the house, and supply surplus electricity back to the grid [21]. This Stirling engine based micro CCHP unit, fuelled by natural gas, had an electrical output of 736W and a thermal output of 6.5kW. In

Micro CCHP: Future Residential Energy Center

245

general, Stirling engine is an ideal technology for residential micro CCHP systems, but it also has the disadvantage as micro turbine has, its high capital cost. Though still on the brink of market entry, fuel cells are the focus of interest as the prime mover technology for micro CCHP systems to serve a variety of residential buildings in the future. Fuel cells are electrochemical devices that convert the energy of a chemical reaction directly into electricity and heat. They are similar in principle to primary batteries except that the fuel and oxidant are supplied to the cell, rather than stored internally. They are silent in operation, modular without moving parts, which are perfect merits for residential sector, but there is limited experience to validate potential applications. Moreover, fuel cells micro CCHP systems carry high capital costs and higher project risk due to unproven durability and reliability. At the beginning of this century, several first prototype systems were reported by utility companies, heating appliance manufacturers and RandD centers in Europe, Japan, and the US. After that, further steps towards industrialization and product development were reported. Yasuhiro Hamada et al. tested the performance of a 1kW polymer electrolyte fuel cell as a residential energy system, of which the electrical efficiency and heat recovery efficiency are 42.5% and 49.2% respectively [22]. Afterward, Tokyo Gas Co., Ltd. marketed the first domestic polymer electrolyte fuel cell with 1kW capacity and 31% generation efficiency in 2005 [23]. G. Gigliucci et al. [24] introduced a PEM fuel cell CHP system supplied by HPower in Italy. The system converts natural gas into electricity and heat, at nominal conditions, 4kW of electric power and 6.8kW of thermal power: the former is delivered to local loads using electric load following capability; the latter is delivered to the experimental area hydraulic refrigeration circuit. S. Giddey et al. [25] describes the design and assembly of a 1kW PEM stack tested, analysis of the results and problems encountered during operation. The electrical efficiency of the stack varies from 39 to 41%. The recoverable combined heat and power efficiency of the stack is 65% without external thermal insulation and 80% with external thermal insulation. A PEMFC system is studied by Charles-Emile Hubert [26]. Five units were installed from November 2002 to May 2003 and have been operated in real life conditions. They deliver up to 4kW of AC power and about 6kW of heat. P. Koenig et al. [27] analyzes a prototype PEMFC CHP system for decentralized energy supply in domestic applications. The complete system supplies 2kW electricity and approximately 4 kW heating power at 60 for domestic hot water and space heating. The tests include steady state measurements under different electrical and thermal loads as well as an analysis of the dynamic behavior of the system during load changes. While polymer electrolyte fuel cell keeps developing, solid oxide fuel cell remains one of the most promising options for distributed CCHP applications, with the prospect for incredibly high electrical generation efficiency. Several companies are keen on developing SOFC technology for the residential CCHP market since 1995. Fuel Cells Bulletin [28] reports that a Home Energy Centre, provided by the Baxi Group, supplied all the heat and electricity for a new four-bedroom house in Scotland. The Home Energy Centre is based around a PEM fuel cell combined with a natural gas reformer. It provides 1.5kW of electricity and 18kW of heat, sufficient for all of an average domestic propertys heat requirements and up to 75% of its electrical power needs, all year round. At the same time, some other researchers try to make SOFC applications more suitable for modern family dwellings. A.D. Hawkes et al. [29] point out that SOFC-based micro CCHP applications have a low heat-topower ratio and may benefit from avoidance of thermal cycling.

246

R. Z. Wang and D. W. Wu

Figure 6. A solid polymer fuel cell system [29].

They also find that these units suit to slow space heating demands, where the heating system is on constantly during virtually all of the winter period. R.J. Braun et al. [30] evaluate five different SOFC system designs in terms of their energetic performance and suitability for meeting residential thermal-to-electric ratios. Effective system concepts and key performance parameters are identified. The results indicate that maximum efficiency is achieved when cathode and anode gas recirculation is used along with internal reforming of methane. System electric efficiencies of 45% and combined heat and power efficiencies of 88% are described. In the near future, fuel cell based residential micro-CCHP systems will compete with traditional energy supplies. Some literatures assess energy utilization, emission aspect or economic feasibility of this type of micro CCHP in the residential sector. V. Dorer [31] establishes a methodology for assessing the performance of SOFC and PEMFC systems in terms of primary energy demand and the CO2 emissions by transient computer simulations. Adam Hawkes [32] explores the performance of a hypothetical SOFC system under UK market conditions at that time. He indicates that the optimized result of a household SOFC micro CHP system depends on system size, energy import prices, electricity export price, stack capital costs or an improvement in stack life time. Another financial analysis of SOFC units is carried out by Kari Alanne et al. [33], to evaluate the sensitivity of the maximum allowable capital cost with respect to system sizing, acceptable payback period, energy price and the electricity buyback strategy of an energy utility. Based on this financial analysis, micro (12kW) SOFC systems seem to be feasible in the considered case.

Micro CCHP: Future Residential Energy Center

247

STATUS AND DEVELOPMENT OF MICRO CCHP TECHNOLOGIES


A typical CCHP system consists of five basic elements: prime mover; electricity generator; heat recovery system; thermally activated equipment (or other cooling options) and management and control system. Among them, prime movers obviously play a critical role; they are the keystones of CCHP systems and, to some extent, they determine possibilities and availability of other related technologies. (Seen in Table 2). Although steam turbine and combustion turbine are two crucial technologies for centralized CHP plants, there is almost no small capacity product in the residential level. For micro CCHP systems, the options of prime movers can be reciprocating internal combustion engine, micro-turbine, stirling engine and fuel cell. All the options can be selected by users to meet dissimilar demands and limitations from site to site: especially local heat and electricity profiles, regional emissions and noises regulations and installation place restrictions. Thermally activated equipment is another essential part of a CCHP system to provide cooling or dehumidification. Commercialized thermally activated technologies include absorption chillers and desiccant dehumidifiers. Moreover, novel adsorption chillers approaching commercial stage can be another choice for micro CCHP systems. Some existed systems also apply electric chillers or engine-driven chillers integrated with prime movers to fulfill cooling demands, which combined with thermally activated technologies to be called cooling options of CCHP system in some literatures.

Prime Movers
Reciprocating Internal Combustion Engines [1,6,7,17,34-36] Two types of internal combustion engines are currently in use; spark ignition engines, which are operated mainly with natural gas; and compression ignition engines, which can use diesel fuel, as well as other petroleum products, such as heavy fuel oil or biodiesel. In addition to fast startup capability and good operating reliability, relatively high efficiency at partial load operation gives users a flexible power source. Reciprocating engines are by far the most commonly used power generation equipment under 1 MW. Although they are a mature technology, reciprocating engines have obvious drawbacks. Relatively high vibrations require shock absorption and shielding measures to reduce acoustic noise. A large number of moving parts with frequent maintenance intervals, increase maintenance costs and strongly offset fuel efficiency advantages. Moreover, high emissions, -particularly nitrogen oxides-- are the underlying aspect of this technology and need to be improved. Major manufacturers around the world continuously develop new engines with lower emissions; at the same time, emissions control options, such as selective catalytic reduction (SCR), have been utilized to reduce emissions. Micro-Turbines [1,7,17, 34,35,37] Micro-turbines extend combustion turbine technology to smaller scales. They are primarily fuelled with natural gas, but they can also operate with diesel, gasoline or other similar high-energy fuels. Research on biogas is ongoing. Micro-turbines have only one

248

R. Z. Wang and D. W. Wu

moving part; they use air bearings and do not need lubricating oil, although they have extremely high rotational speed, up to 120000 rpm. A striking characteristic is their flexibility that small-scale individual units can be combined readily into large systems of multiple units. Additionally, there are environmental advantages, such as low combustion temperatures assuring low NOx emissions levels and less noise than an engine of comparable size. This technology has been commercialized for only a few years and is offered by a small number of suppliers. The main disadvantage at this stage is its high costs compared with engine. Other flaws include relatively low electrical efficiency and sensitivity of efficiency to changes in ambient conditions.

Stirling Engines [1,17,34,35] Compared to conventional internal combustion engines, Stirling engines have an external combustion device in which the cycle medium--generally helium or hydrogen--is not exchanged during each cycle, but remains within the cycle while the energy driving the cycle is applied externally. Stirling engines can operate on almost any fuel (gasoline, alcohol, natural gas or butane), with the external combustion that facilitates the control of the combustion process and results in low air emissions, low noise and more efficient process. In addition, fewer moving parts compared to conventional engines limit wear on components and reduce vibration levels. Stirling engine is still in its development. High cost also prevents popularization of this technology. Nevertheless, the promising prospects of stirling engines stimulate further research, especially for CCHP applications. Small size and quiet operation mean that they will integrate well into residential or portable applications. Some literature indicates the possibility of using a solar dish to heat the Stirling engine, thus eliminating the need for combustion of a fuel. Fuel Cells [1,6,7,34,35,37,38] Fuel cells are quiet, compact power generators without moving parts, which use hydrogen and oxygen to make electricity; at the same time, provide heat for a wide range of applications. In general, fuel cells show high electrical efficiencies under varying load, and which results in low emissions. Besides transportation sector, power generation is another promising market. Five major fuel cell technologies listed below have the most attractive prospects. A detail comparison of the characteristics of these fuel cells appears in Table 1. Proton Exchange Membrane Fuel Cell (PEMFC) Also known as Polymer Electrolyte Membrane Fuel Cell, PEM fuel cells are quite simple and can be made very small to adjust to variable power demands. They are easier to start up and they apply solid electrolyte that reduces corrosion. At the same time, the low operating temperature requires the use of an expensive platinum catalyst, which limits the cogeneration potential. As for the fuel sources, this fuel cell technology is highly sensitive to fuel impurities and hydrogen storage; delivery and reforming technology has yet to evolve. With relatively low quality heat, PEM fuel cell is unlikely to be widely used for high voltage stationary power generation; but smallscale domestic CCHP applications --the simplest thermal load of which is hot water-- would be considerable.

Table 1. Characteristics of fuel cells [17,34,35]


PEMFC Charge Carrier Type of Electrolyte H+ ions Polymeric membrane AFC OH- ions Aqueous potassium hydroxide soaked in a matrix Plastic, metal Platinum Purified Air or O2 Clean hydrogen or hydrazine 60-80C 10-200kW 32-70% CO, CO2, and Sulfur PAFC H+ ions Phosphoric acid solutions Carbon, porous ceramics Platinum Air or Oxygen- Enriched Air Hydrocarbons or alcohols 100-200C 100-200kW 40-55% CO>1%, Sulfur MCFC CO3= ions Phosphoric acid (Immobilized liquid) High temp metals, porous ceramic Nickel Air Clean hydrogen, nature gas, propane, diesel 600-700C 250kW-5MW 55-57% Sulfur SOFC O= ions Stabilized zirconia ceramic matrix with free oxide ions Ceramic, high temp metals Parasites Air Natural gas or propane 600-1000C 100kW-10MW 50-60% Sulfur

Typical Construction Catalyst Oxidant Fuel Operational Temperature Size Range Electrical Efficiency Primary Contaminants

Plastic, metal or carbon Platinum Air or O2 Hydrocarbons or methanol 50-100C 3-250kW 30-50% CO, Sulfur, and NH3

Table 2. Characteristics and parameters of prime movers in CCHP systems [1,17,34,35,37,47,48]


Diesel engines Capacity range Fuel used Efficiency electrical (%) Efficiency overall (%) Power to Heat Ratio Output heat temperature () Noise CO2 emissions (kg/ MWh) NOx emissions (kg/ MWh) Availability (%) Part load performance Life cycle (year) Average cost investment ($/kW) Operating and maintenances costs ($/kWh)
*

5kW-20MW Gas, Propane, Distillate Oils, Biogas 35-45 65-90 0.8-2.4 * Loud 650 10 95 Good 20 340-1000 0.0075-0.015

Spark ignition engines 3kW-6MW Gas, Biogas, Liquid Fuels, Propane 25-43 70-92 0.5-0.7 * Loud 500-620 0.2-1.0 95 Good 20 800-1600 0.0075-0.015

Micro-turbines 15-300kW Gas, Propane, Distillate Oils, Biogas 15-30 60-85 1.2-1.7 200-350** Fair 720 0.1 98 Fair 10 900-1500 0.01-0.02

String Engines 1kW-1.5MW Any (Gas, Alcohol, Butane, Biogas) ~ 40 65-85 1.2-1.7 60-200 Fair 672*** 0.23**** N/A Good 10 1300-2000 N/A

Fuel Cells 5kW-2MW Hydrogen and fuels containing hydrocarbons 37-60 85-90 0.8-1.1 260-370 Quiet 430-490 0.005-0.01 90-95 Good 10-20 2500-3500 0.007-0.05

Up to a third of the fuel energy is available in the exhaust at temperatures from 370-540C; other rejected heat is low temperature, often too low for most processes. (Jacket cooling water at 80 to 95C, lube oil cooling at 70C and intercooler heat rejection at 60C, all difficult to use in CHP). ** 650C without recuperator. *** Stirling Engine Emission Characteristics / STM 4260. Gas-Fired Distributed Energy Resource Technology Characterizations.

Micro CCHP: Future Residential Energy Center

251

Alkaline Fuel Cell (AFC) Alkaline fuel cells are the first fuel cells used on spacecrafts and space shuttles. The technology shares obvious merits, such as low operating temperature, rapid startup time, readily available non-precious metal electrodes, and high efficiency, up to 70%. However, the primary disadvantage is the tendency to absorb carbon dioxide, converting the alkaline electrolyte to an aqueous carbonate electrolyte that is less conductive. Thus, the fuel input should be restricted to pure hydrogen, which limits applications to those in which pure hydrogen are available. If the CO2 is removed from fuel and oxygen streams, the operating costs are much greater. Although the attractiveness of AFC has declined substantially with the interest and improvements in PEMFC technology, recent developers still believe that it can be used for many applications, such as stationary power generation, but also mobile applications including both marine and road vehicles. Phosphoric Acid Fuel Cell (PAFC) Phosphoric acid fuel cells are the most mature of the technologies in commercial production, although its costs remain uncompetitive with other non-fuel cell technologies. Hydrogen is still the ultimate fuel for the reaction in the phosphoric acid fuel cell, but various fuels, including natural gas, LPG and methanol, can be used as raw input converted by a reformer. Other advantages are its resistance to fuel impurities, and the ability to use a less expensive catalyst. The drawbacks of this fuel cell include a lower efficiency than other fuel cell technologies and corrosive liquid electrolyte. In the near future, with lower operating temperatures, PAFC would be ideal for small and midsize power plants, replacing large electrical generators and other types of CCHP utilities in hospitals, hotels and airports. Molten Carbonate Fuel Cells (MCFC) A molten carbonate fuel cell uses a molten carbonate salt mixture as its electrolyte. The composition of the electrolyte varies, but usually consists of lithium carbonate and potassium carbonate, which is chemically aggressive and puts strain on the stability and wear of the cell components. As a result, MCFC is more expensive than either SOFC or PEMFC in terms of capital cost. Fuel reforming of MCFC occurs inside the stack and tolerates impurities; therefore, this technology may use a variety of fuels. In addition, the high operating temperature allows for combined heat and power generation and high fueltoelectricity efficiency. Nevertheless, the long startup time to reach operating temperatures, and poorer flexibility in output, make MCFC ideally suited to base load power generation where continuous operation is necessary, such as heavy industries and national electrical grid networks. Solid Oxide Fuel Cell (SOFC) Due to allsolidstate ceramic construction, solid oxide fuel cells share important characteristics, such as stability and reliability. A variety of hydrocarbon fuels can be used, like gasoline, methanol and natural gas. As another asset, the high operating temperature makes internal reforming possible and removes the need for a catalyst, which also produces high grade waste heat suited well to CCHP applications. But the high temperature also creates some difficulties: expensive alloys for components are required, quit a long time is needed for the electrolyte to heat. Start-up time is less of an issue for stationary and continuous

252

R. Z. Wang and D. W. Wu

applications. They generally achieve around 60% efficiency in a 5MW plant, compared to around 30% for a traditional gas turbine. The last critical problem that prevents its commercialization is the comparatively high costs of SOFC.

ThermallyActivated Technologies
An important difference between CCHP systems and conventional cogenerations is that CCHP systems including some cooling or dehumidification components provide not only electricity and heating but also cooling capacity for space or process. These cooling or dehumidification options can employ advanced thermallyactivated technologies as well as traditional technologies. But recent research indicates that thermallyactivated technologies are favored, as the overall efficiency of CCHP systems is promoted with applications of these thermallyactivated technologies. In addition to high efficiency, other benefits such as low emissions and cost reduction are also achieved with thermallyactivated technologies. Major thermallyactivated technologies include absorption chillers, adsorption chillers and desiccant dehumidifiers. These cooling and dehumidification systems can be driven by steam, hot water or hot exhaust gas derived from prime movers. However, waste heat from various prime movers falls into different temperature ranges; at the same time, cooling and dehumidification systems have their own suitable working temperature. As a result, optimal matching of recoverable energy streams with thermally driven technologies is shown in Table 3. Table 3. Recoverable energy qualities with matching technologies [39]
Power Source Solid Oxide Fuel Cell Micro-turbine Phosphoric Acid Fuel Cell Stirling Engine IC Engine PEM Fuel Cell Temp. ~ 480C ~ 320C ~ 120C ~ 90C ~ 80C ~ 60C Matching Technology Triple-effect/ Double-effect absorption Triple-effect/ Double-effect absorption Double-effect/ Single-effect absorption Single-effect absorption, adsorption or dehumidification Single-effect absorption, adsorption or dehumidification Single-effect absorption, adsorption or dehumidification

Absorption Chillers [7,39,40,41] Absorption chillers are one of the commercialized thermallyactivated technologies widely applied in CCHP systems; they are similar to vapor compression chillers, with a few key differences. The basic difference is that a vapor compression chiller uses a rotating device to raise the pressure of refrigerant vapors, while an absorption chiller uses heat to compress the refrigerant vapors to a highpressure. Therefore, this thermal compressor has no moving parts. Depending on how many times the heat supply is utilized; absorption chillers can be divided into singleeffect, doubleeffect and tripleeffect. The parameters and traits of different absorption chillers can be viewed in Table 4.

Table 4. Characteristics of absorption technologies [49]


System Operating Temp. ( ) Heat source Cooling Working Fluid Cooling Capacity (ton) COP Current Status Remark Simplest and widely used Using water as a refrigerant, cooling temperature is above 0C Negative system pressure Water cooled absorber required to prevent crystallization at high concentration Rectification of refrigerant required Working solution is environmental friendly Operating pressure as high as with NH3 No crystallization problem Suitable for use as heat pump due to wide operating range High performance cycle, commercially available Heat of condensation from first effect used as heat input for second stage Heat release from first stage absorber used for second stage generator High complexity control system Likely to be directfired, as input temp is very high Requires more maintenance as a result of high corrosion due to high operating temperature

Single effect cycle

80110

510

LiBr/ water

10-1500

0.50.7

Large water chiller

Single effect cycle

120150

<0

Water/ NH3

3-25

0.5

Commercial

Double effect (Series flow) Double effect (Parallel flow)

120150

510

LiBr/ water Water/ NH3

200-1500

0.81.2

Large water chiller

<0

Experimental unit

Triple effect cycle

200230

510

LiBr/ water

N/A

1.41.5

Computer model and experimental unit

Table 5. Characteristics of adsorption working pairs [42,44]


Heat of adsorption (kJ/kg) 2800 1000-1500 3300-4200 4000-6000 1200-1400 1800-2000 >2000 NH3 CH3OH 1368 N/A Release Temp. ( ) 70-100

Adsorbent

Adsorbate H2O CH3OH

Toxicity No Yes No Yes No Yes Yes Yes Yes

Vacuum Level High High High High Moderate High High Low Low

Heat Sources Solar energy, lowtemperature waste heat Hightemperature waste heat

Applications

Silica gel

Space cooling, refrigeration

Zeolite Activated Charcoal Charcoal fiber CaCl2

H2O NH3 C2H5OH CH3OH

>150 100 110 120 95

Space cooling, refrigeration

Solar energy, lowtemperature waste heat Solar energy, lowtemperature waste heat

Low temperature, ice making

Low temperature, ice making

Micro CCHP: Future Residential Energy Center

255

The most common working fluids for absorption chillers are water/NH3 and LiBr/water, although there are 40 refrigerant compounds and 200 absorbent compounds available in theory [49]. Lithiumbromide/water absorption chillers play a predominant role in the absorption chiller market in Asia-Pacific countries like China, Japan, Korea, and in the US. In contrast, ammonia/water absorptions chillers are more popular in Europe.

Adsorption Chillers [42-45] Adsorptioncooling technology is a novel, environmentallyfriendly and effective means of using lowgrade heat sources. Unlike conventional vapor compression systems which require a mechanical compressor assembly, this new technology uses a thermallydriven static sorption bed, saving as much as 90% of the required input power typically used to drive a mechanical compressor. The system takes advantage of the ability of certain sorbent material, stored in a sorption bed, to soak up a relatively large quantity of refrigerant vapor at some low temperature and pressure. At this stage, cooling capacity is achieved in the evaporator because of the evaporation of the refrigerant. The refrigerant is subsequently released to the condenser at a higher pressure simply by applying heat to the sorbent bed. To increase the efficiency and provide continuous cooling, more than one sorption bed is often used. A heat regeneration fluid also can be used to increase system efficiency by transferring heat from a hot to a cold bed. As a critical part of this technology, the characteristics of various adsorbentadsorbate working pairs are listed in the Table 5. Since there are no moving parts, except for valves, the sorption system is considerably simpler, requiring no lubrication and thus, little maintenance. Other advantages include quiet operation and modularity so it is readily scalable for increased heating and cooling capacity by additional beds. Furthermore, any heat source, such as waste heat or renewable energy, can be used, so energy saving can be potentially significant. Desiccant Dehumidifiers [3,40,41] Desiccant dehumidifiers can work in concert with sorption chillers or conventional air conditioning systems to significantly increase overall system energy efficiency by avoiding overcooling air and precluding oversized capacity to meet dehumidification loads. The desiccant process involves exposing the desiccant material (such as silica gel, activated alumina, lithium chloride salt or molecular sieves) to a moisture-laden process air stream, retaining the moisture of the air in desiccant and regenerating desiccant material via a heated air stream. System capacity is often expressed in volume of airflow or in moisture removal rate. Table 6 shows some specifications.
Table 6. Performance of desiccant dehumidification systems [41]
Flux (m3/min) 40-140 140-280 280+ Thermal Input (W/m3/min) 300-1000 300-1000 300-1000 Maximum Latent Removal (W/m3/min) 300-600 300-600 300-600

256

R. Z. Wang and D. W. Wu

Dehumidification technology is divided into two major types, solid desiccant dehumidifiers and liquid desiccant dehumidifiers; both are useful for the mitigation of indoor environmental quality and security problems and for humidity control in buildings. Dehumidification technology in the commercial sector remains a young technology with a premium price. Yet, commercial desiccant technologies have not been designed for integration into CCHP systems.

Other Options
Although thermallyactivated technologies indicate the trend in cooling and dehumidification options in CCHP systems, electric vapor-compression refrigeration systems still play an important role for their maturity and reliability. Therefore, quite a few CCHP systems in research and practical utilization still employ these conventional technologies as their cooling options. Nonetheless, it is unwise for a CCHP system to drive chillers using electricity generated by prime movers, since smaller prime movers have lower efficiency than larger types used in power plants. Enginedriven chillers emerge to substitute electric chillers in CCHP units, avoiding the losses in energy conversion. Enginedriven chillers, including reciprocating types, centrifugal types and screw types, are conventional chillers driven by an engine, in lieu of an electric motor. An advantage of enginedriven chillers is better variable speed performance, which improves partialload efficiency. Enginedriven chillers can also operate in a CCHP system for hot water loads when the waste heat produced by the engine is recovered.

RESEARCH TENDENCY OF MICRO RESIDENTIAL CCHP


Development of distributed CCHP systems has been undergoing for almost two decades, while many related technologies have been employed and ameliorated in this field through comprehensive researches. Micro residential CCHP systems for single- or multi-family buildings as new applications emerge in recent years. The technologies used in distributed CCHP systems are miniaturized and improved to better fit for micro residential utilizations [50,51,52,53]. Besides that, some novel technologies are also invented as new alternatives. Except for detail improvement or invention in specific technologies, new design methods for the whole micro CCHP systems are introduced in some literatures. Andrew Wright et al. [54] describe exploratory analyses of domestic electricity-profiles recorded at a high time resolution of 1 min on eight houses, while most load data are available at half-hour intervals. It is included that for dwellings with micro CCHP, a better understanding of electricity profiles is important for the economic analysis of systems, and to examine the effects of widespread onsite generation on local electricity-networks. In Adam Hawkess paper [55], it is also indicated that coarse temporal precision profiles of 1-h demand blocks in heat and power demands become questionable for applications where demand exhibits substantial volatility such as for a single residential dwellingan important potential market for the commercialization of small-scale fuel cells. Total CO2 emissions reduction is overestimated by up to 40% by the analyses completed using coarse demand data for a given micro CCHP

Micro CCHP: Future Residential Energy Center

257

unit. The economic difference is also significant at up to 8% of lifetime costs. H. Lund [56] presents the Danish experience with methodologies and software tools, which have been used to design investment and operation strategies for almost all small CCHP plants in Denmark during the decade of the triple tariff. Moreover, the changes in such methodologies and tools in order to optimize performance in a market with fluctuating electricity prices are discussed. For existed CCHP systems, various criteria are employed to compare different systems and obtain the best configuration of a typical application. Some simulation models are also presented in some papers to optimize operating performance, system configuration and control strategy. Aiying Rong [57] illustrates a long-term planning model based on hourly load forecasts to obtain cost-efficient operation of a CCHP system. This linear programming model with a joint characteristic for three energy components minimizes simultaneously the production and purchase costs of three energy components, as well as CO2 emissions costs. S.M. Ameli [58] presents the works which have been done and yielded results about the requirements of developing integrated distributed energy evaluation software. The comprehensive software package is for designing, optimizing and monitoring of distributed energy systems based on micro-turbine, fuel cell and internal combustion engine driven systems. A.D. Peacock et al. [59] employs a 50 dwelling data set of heat and power demands to investigate the implementation of various penetrations of micro CHP system on the resultant electrical load profile using two control methodologies: heat-led and a proposed method for modulating the aggregate electrical load. And they point out that further improvements in the modulating capability of this control approach may be realized if prime movers capable of rapid start-up, shut-down and cycling can be developed. A.D. Hawkes et al. [60] investigate cost effective operating strategies for three micro CCHP technologies; Stirling engine, gas engine, and solid oxide fuel cell. In this paper, central estimates of price parameters are used, which is shown that the least cost operating strategy for the three technologies is to follow heat and electricity load during winter months, rather than using either heat demand or electricity demand as the only dispatch signal. Least cost operating strategy varies between technologies in summer months. In another paper, A.D. Hawkes et al. [61] develop a techno-economic modeling of a solid oxide fuel cell stack for micro CHP system. Some literatures of internal combustion engine for residential sector also can be found. Hycienth I. Onovwiona et al. [62] present a parametric model that can be used in the design and techno-economic evaluation of internal combustion engine based CCHP systems for residential use. The model, which is suitable to provide system performance information in response to a buildings electrical and thermal demands, and is capable of simulating the performance of these systems in 15-min time steps. After focusing on the design, assessment and simulation of typical novel micro residential CCHP systems, some researchers begin to consider the future scenario of micro CCHP systems in residential sector, the relationship between distributed micro residential CCHP systems with centralized power plants and comparison in emission, economy and social costbenefit between them. Samuel Bernstein [63], H. Lund [64], Ineke S.M. Meijer [65], Jeremy Cockroft [66], discuss various aspect of the development situation of residential CCHP applications in US, Lithuania, Netherlands and UK, while Neil Strachan [67] and Francesco Gulli [68] provide a brand-new view point toward distributed CCHP systems including micro residential CCHP units, which deserves to investigate further.

258

R. Z. Wang and D. W. Wu

CONCLUSION
Micro residential CCHP systems share some important advantages with distributed CCHP systems: high overall energy efficiency ranging from 70% to more than 90%; less prime energy consumption; emission reduction and increase in the reliability of the energy supply. The electricity, heating and cooling demands of a family can be covered by micro CCHP systems simultaneously and independently, which is similar to larger systems, although larger systems have already well proven in quite a few applications of industrial, institutional and commercial buildings. A review of the current applications, demonstrations and experimental systems for residential single- or multi-family buildings has been presented. The prime movers applied in these systems include small traditional reciprocating internal combustion engine as well as micro-turbines, novel Stirling engine and fuel cells, especially PEMFC and SOFC. These technologies are suitable for residential applications to meet electricity demands. Although most systems listed are just cover the electricity and heat demands, some applications are also combined with certain thermally activated cooling options, such as small absorption chiller or adsorption chiller, to meet the space cooling demands in summer. In general, micro residential CCHP systems include both these two types of family energy supply centers. The review of micro CCHP applications illustrate that such kind of energy supply for families develops rapidly in recent years and many of them are in the edge of market entry. A further review analyzes the technologies used in micro residential systems comprehensively. Generally speaking, reciprocating engine based micro CCHP systems are the most realistic and reliable product for single- or multi- family buildings in current market. Micro turbine based systems are also feasible in technology, but too expensive to be afforded. Stirling engine based systems become an important competitor in the market, because of their versatility in fuel and other advantages compare with internal combustion engines. However, several obstacles need to be surmounted, especially high initial investment cost. In long views, fuel cell based micro CCHP systems are the most promising technologies. SOFC and PEMFC systems are in the beginning stage of commercialization. With technology improvement and large scale production, the capital costs of fuel cells will reduce rapidly in next decade. It is also critical to develop micro scale thermal initialized cooling technologies for single-family dwellings. Most demonstrations and applications of micro CCHP systems are actually only micro CHP (combined heating and power) systems. But in many places of the world, cooling demand of a family dwelling in summer is even larger than heating demand in winter. Consequently, micro scale absorption chiller, adsorption chiller and desiccant dehumidifiers are also an important part of future residential energy center. Research issues indicate the future tendency and direction of micro residential CCHP systems. Development of specific CCHP technologies is always the most critical research content, which is to find ways to overcome drawbacks of certain technology, pursuit higher energy efficiency, diminish emissions, reduce the capital cost and follow users electricity and heat profiles more precisely and economically. At the same time, new design methods for the whole micro CCHP system are necessary to be built up. While basic equipment technologies are available, a core thought is needed to optimize CCHP configuration, control strategies and economical operation. Besides that, various criteria are discussed to better estimate and compare different existed micro CCHP systems, which also stimulate and assist the

Micro CCHP: Future Residential Energy Center

259

development of the core designing thought. Similarly, simulation models of typical technologies or the whole systems also help to optimize operating performance, system configuration and control strategy. Diverse research results of the future scenario of micro residential CCHP systems indicate that distributed micro residential CCHP system is, to some extent, a controversial energy supply mode. The relationship between distributed micro residential CCHP systems with centralized power plants is necessitated to be further compared in energy, emission, economy, environment and social benefit aspects. Despite of some discussion, it is believed that micro residential CCHP systems would be the next generation of family energy supply center in decades.

REFERENCES
The European Educational Tool on Cogeneration, Second Edition. COGEN Europe (The European Association for the Promotion of Cogeneration, www.cogen.org). (December 2001). [2] The State-of-the Art of Tri-generation and Its Application in Tertiary Sector. Institute for Promotion of Innovative Technologies (Italy), Wen Guo. [3] Market Potential for Advanced Thermally-activated BCHP in Five National Account Sectors. Energy and Environmental Analysis, Inc. (May 2003). [4] Hawaii Commercial Building Guidelines for Energy Efficiency: Chapter 9 Building Cooling, Heating and Power Generation Systems. [5] Distributed Energy Resources and Combined Heat and Power: A Declaration of Terms. Anna Monis Shipley and R. Neal Elliott (The American Council for an EnergyEfficient Economy). (April 2000). [6] Assessment of Distributed Generation Technology Applications. Resource Dynamics Corporation. (February 2001). [7] Cooling Heating and Power for Industry a Market Assessment. Resource Dynamics Corporation. (August 2003). [8] Domestic-scale Combined Heat-and-Power System Incorporating a Heat Pump: Analysis of a Prototype Plant. Applied Energy, 70, 215-232. M.A. Smith, P.C. Few. (2001). [9] Feasibility of a New Domestic CHP Trigeneration with Heat Pump I. Design and Development. Applied Thermal Engineering, 24, 1409-1419. J.L.Mguez, S.Murillo, J.Porteiro, L.M.Lopez. (2004). [10] Feasibility of a New Domestic CHP Trigeneration With Heat Pump II. Availability Analysis. Applied Thermal Engineering, 24, 1421-1429. J.Porteiro, J.L.Mguez, S.Murillo, L.M.Lopez. (2004). [11] Performance Research of a Micro CCHP System with Adsorption Chiller. ISHPC-0442005. Proceedings of the International Sorption Heat Pump Conference, Denver, CO, USA. June 22-24, 2005. R.Z. Wang, X.Q. Kong, J.Y. Wu, Y. Huang Fu, D.W. Wu. (2005). [12] Micro-Cogen Systems from Marathon Engine. Diesel Progress North American Edition, August 2004. Marathon Engine Systems. (2004). [1]

260

R. Z. Wang and D. W. Wu

[13] An experimental investigation of a household size trigeneration. Applied Thermal Engineering, 27, 576585. Lin Lin, Yaodong Wang et al. (2007). [14] CHP Case Studies Saving Money and Increasing Security. Stephen F. Gillette. Capstone Turbine Corporation. (2004). [15] Solar energy-gas driven micro-CHP system for an office building. Applied Thermal Engineering, 26, 16041610. W. Yagoub, P. Doherty, S.B. Riffat. (2006). [16] A novel hybrid heat pipe solar collector/CHP systemPart 1: System design and construction. Renewable Energy, 29, 22172233. S.B. Riffat, X. Zhao. (2004). [17] Gas-Fired Distributed Energy Resource Technology Characterizations. The National Renewable Energy Laboratory (NREL), the Department of Energy (DOE), Office of Energy Efficiency and Renewable Energy (EERE) and the Gas Research Institute (GRI). (October 2003). [18] Energy efficiency and economic feasibility of CCHP driven by Stirling engine. Energy Conversion and Management, 45(9-10), 1433-1442. X.Q. Kong, R.Z. Wang and X.H.Huang. (2004). [19] Mini and Macro-CHP Market Assessment and Development Plan Summary Report. Faber Maunsell Ltd. London, UK. (2002). [20] Technical Documentation Solo Stirling 161 CHP Module. SOLO Stirling Engine. http://www.stirling-engine.de. Germany. [21] Micro-generation technology assessment for housing technology. Energy and Buildings, 36, 925931. Evgueniy Entchev, John Gusdorf, Mike Swinton, Mike Bell, Frank Szadkowski, Walter Kalbfleisch, Roger Marchand. (2004). [22] Field Performance of a Polymer Electrolyte Fuel Cell for a Residential Energy System. Renewable and Sustainable Energy Reviews, 9, 345-362. Yasuhiro Hamada, et al. (2004). [23] Commercial Unit of Residential Fuel Cell Cogeneration Systems Launch into the Market. Tokyo Gas Co., Ltd. (December 2004). [24] Demonstration of a residential CHP system based on PEM fuel cells. Journal of Power Sources, 131, 6268. G. Gigliucci, L. Petruzzi, E. Cerelli, A. Garzisi, A. La Mendola. (2004). [25] Design, assembly and operation of polymer electrolyte membrane fuel cell stacks to 1 kWe capacity. Journal of Power Sources, 125, 155165. S. Giddey, F.T. Ciacchi, S.P.S. Badwal. (2004). [26] Study of a small heat and power PEM fuel cell system generator. Journal of Power Sources, 156, 6470. Charles-Emile Hubert, Patrick Achard, Rudolf Metkemeijer. (2006). [27] Testing and model-aided analysis of a 2kWel PEMFC CHP-system. Journal of Power Sources, 145, 327335. P. Koenig, A. Weber, N. Lewald, T. Aicher, L. Joerissen, E. Ivers-Tiffee, R. Szolak, M. Brendel, J. Kaczerowski. (2005). [28] First UK domestic fuel cell cogen installation. Fuel Cells Bulletin, October 2005, 6. (2005). [29] Solid oxide fuel cell micro combined heat and power system operating strategy: Options for provision of residential space and water heating. Journal of Power Sources, 164, 260271. A.D. Hawkes, P. Aguiar, B. Croxford, M.A. Leach, C.S. Adjiman, N.P. Brandon. (2007).

Micro CCHP: Future Residential Energy Center

261

[30] Evaluation of system configurations for solid oxide fuel cell-based micro-combined heat and power generators in residential applications. Journal of Power Sources, 158, 12901305. R.J. Braun, S.A. Klein, D.T. Reindl. (2006). [31] Performance assessment of fuel cell micro-cogeneration systems for residential buildings. Energy and Buildings, 37, 11321146. V. Dorer, R. Weber, A. Weber. (2005). [32] Solid oxide fuel cell systems for residential micro-combined heat and power in the UK: Key economic drivers. Journal of Power Sources, 149, 7283. Adam Hawkes, Matthew Leach. (2005). [33] The financial viability of an SOFC cogeneration system in single-family dwellings. Journal of Power Sources, 158, 403416. Kari Alanne, Arto Saari, V. Ismet Ugursal, Joel Good. (2006). [34] Guide To Decentralized Energy Technologies. WADE (World Alliance of Decentralized Energy, www.localpower.org). (2002). [35] Sustainable small-scale CHP technologies for buildings: the basis for multi-perspective decision-making. Renewable and Sustainable Energy Reviews, 8, 401-431. Kari Alanne, Arto Saari (Laboratory of Construction Economics and Management, Helsinki University of Technology). (2004). [36] Technology Characterization: Reciprocating Engines. EPA (www.epa.gov). (March 2002). [37] Distributed Generation in Liberalized Electricity Markets. IEA (The International Energy Agency). (2002). [38] Stationary Fuel Cells: Future Promise, Current Hype. Anna Monis Shipley and R. Neal Elliott (The American Council for an Energy-Efficient Economy). (March 2004). [39] Integrated Energy Systems: Technologies, Program Structure, and Applications. Phillip Fairchild (Oak Ridge National Laboratory). (April 2002). [40] Combined Heat and Power Resource Guide. Midwest CHP Application Center (University of Illinois at Chicago) and Avalon Consulting, Inc. (September 2003). [41] Integrated Energy Systems (IES) for Buildings: A Market Assessment. Resource Dynamics Corporation. (August 2002). [42] Adsorption Refrigeration. China Machine Press. Ruzhu Wang, Jingyi Wu, Yanjun Dai. (2002). [43] Adsorption refrigeration-green cooling driven by low-grade thermal energy. Chinese Science Bulletin, 50(3), 193-204. Ruzhu Wang, Liwei Wang. (2005). [44] A Review of Adsorbents and Adsorbates in Solid-Vapour Adsorption Heat Pump Systems. Applied Thermal Engineering, 18, 707-714. N.C. Srivastava, I.W. Eames. (1998). [45] Heat exchanger design effect on the system performance of silica gel adsorption refrigeration systems. International Journal of Heat and Mass Transfer, 43, 4419-4431. K.C.A. Alam, B.B. Saha, Y.T. Kang, A. Akisawa, T. Kashiwagi. (2000). [46] Adsorption Air-Conditioning for Containerships and Vehicles. Craig Christy, Dave Fusco and Reza Toossi. (California State University Long Beach). (June 2001). [47] On-site Power Systems for Laboratories. Trina Masepohl, The National Renewable Energy Laboratory (NREL). (2003).

262

R. Z. Wang and D. W. Wu

[48] Decentralized Generation Technologies - Potentials, Success Factors and Impacts in the Liberalized EU Energy Markets, Final Report. COGEN Europe (www.cogen.org). (October 2002). [49] A Review of Absorption Refrigeration Technologies. Renewable and Sustainable Energy Reviews, 5, 343-372. Pongsid Srikhirin, Satha Aphornratana, Supachart Chungpaibulpatana. (2001). [50] Absorption chiller crystallization control strategies for integrated cooling heating and power systems. International Journal of Refrigeration 1-8, (Article in press). Xiaohong Liao, Reinhard Radermacher. (2007). [51] The influence of feedstock drying on the performance and economics of a biomass gasierengine CHP system. Biomass and Bioenergy, 22, 271 281. J.G. Brammer, A.V. Bridgwater. (2002). [52] The influence of operating temperature on the efficiency of a combined heat and power fuel cell plant. Journal of Power Sources, 122, 3746. S.F. Au, S.J. McPhail, N. Woudstra, K. Hemmes. (2003). [53] Thermodynamic and economic performance of the LiBrH2O single stage absorption water chiller. Applied Thermal Engineering, 26, 21032109. Tomasz M. Mroz. (2006). [54] The nature of domestic electricity-loads and effects of time averaging on statistics and on-site generation calculations. Applied Energy, 84, 389403. Andrew Wright, Steven Firth. (2007). [55] Impacts of temporal precision in optimisation modeling of micro-Combined Heat and Power. Energy, 30, 17591779. Adam Hawkes, Matthew Leach. (2005). [56] Optimal designs of small CHP plants in a market with fluctuating electricity prices. Energy Conversion and Management, 46, 893904. H. Lund, A.N. Andersen. (2005). [57] An efficient linear programming model and optimization algorithm for trigeneration. Applied Energy, 82, 4063. Aiying Rong, Risto Lahdelma. (2005). [58] Integrated distributed energy evaluation software (IDEAS) Simulation of a microturbine based CHP system. Applied Thermal Engineering, (Article in press). S.M. Ameli, B. Agnew, I. Potts. (2005). [59] Controlling micro-CHP systems to modulate electrical load profiles. Energy, 32, 1093 1103. A.D. Peacock, M. Newborough. (2007). [60] Cost-effective operating strategy for residential micro-combined heat and power. Energy, 32, 711723. A.D. Hawkes, M.A. Leach. (2007). [61] Techno-economic modelling of a solid oxide fuel cell stack for micro combined heat and power. Journal of Power Sources, 156, 321333. A.D. Hawkes, P. Aguiar, C.A. Hernandez-Aramburo, M.A. Leach, N.P. Brandon, T.C. Green, C.S. Adjiman. (2006). [62] Modeling of internal combustion engine based cogeneration systems for residential applications. Applied Thermal Engineering, 27, 848861. Hycienth I. Onovwiona, V. Ismet Ugursal, Alan S. Fung. (2007) [63] MICRO-CHP: U.S. market potential and complex challenges. www.energyint.com. Samuel Bernstein. (2004). [64] Implementation strategy for small CHP-plants in a competitive market: the case of Lithuania. Applied Energy, 82, 214227. H. Lund, G. Siupsinskas, V. Martinaitis. (2005).

Micro CCHP: Future Residential Energy Center

263

[65] How perceived uncertainties influence transitions; the case of micro-CHP in the Netherlands. Technological Forecasting and Social Change, (Article in press). Ineke S.M. Meijer, Marko P. Hekkert, Joop F.M. Koppenjan. (2006). [66] A comparative assessment of future heat and power sources for the UK domestic sector. Energy Conversion and Management, 47, 23492360. Jeremy Cockroft, Nick Kelly. (2006). [67] Emissions from distributed vs. centralized generation: The importance of system performance. Energy Policy, 34, 26772689. Neil Strachana, Alexander Farrell. (2006). [68] Small distributed generation versus centralised supply: a social costbenefit analysis in the residential and service sectors. Energy Policy, 34, 804832. Francesco Gulli. (2006).

In: Leading-Edge Electric Power Research Editor: C. M. OSullivan, pp. 265-284

ISBN: 978-1-60021-981-8 2008 Nova Science Publishers, Inc.

Chapter 6

SENSITIVITY CALCULATION IN REAL TIME TRANSMISSION NETWORK AND ENERGY MARKETS


Jizhong Zhu
AREVA T & D Corporation 10865 Willows Rd. NE Redmond, WA 98052, USA

ABSTRACT
The calculations of the several sensitivities such as loss sensitivity, voltage sensitivity, generator constraint shift factor, and area based constraint shift factor become very important in energy management system (EMS) and energy markets. This chapter focuses on the analysis and implementation details of the above-mentioned sensitivities calculations in the practical transmission network and energy markets. The power operator uses them to study and monitor market and system behavior and detect possible problems in the operation. These sensitivities calculations are also used to determine whether the on-line capacity as indicated in the resource plan is located in the right place on the network to serve the forecasted demand. If the congestion or violation exists, the generation scheduling based on the sensitivities calculations can determine whether or not a different allocation of the available resources could resolve the congestion or violation problem. This chapter also comprehensively discusses how to compute and use the sensitivities under the different references such as the market-based reference, and the energy management system based reference. The calculation results of the several sensitivities are illustrated using the IEEE 14 bus system and AREVA T & D 60-bus system.

E-mail: jizhong.zhu@areva-td.com

266

Jizhong Zhu

I. INTRODUCTION
The electric power industry is being relentlessly pressured by governments, politicians, large industries, and investors to privatize, restructure, and deregulate. Despite the changes with different structures, market rules, and uncertainties, an energy management system (EMS) control center must always be in place to maintain the security, reliability, and quality of electric service [1]. It means that EMS in the open energy market must respond quickly, reliably and efficiently to the market changes. In order to achieve the optimal objective in energy markets, the calculations of the several sensitivities such as loss sensitivity, voltage sensitivity, generator constraint shift factor, and area based constraint shift factor become very important. This chapter focuses on the analysis and implementation details of the above-mentioned sensitivities calculations in the practical transmission network and energy markets. The power operator uses them to study and monitor market and system behavior and detect possible problems in the operation. These sensitivities calculations are also used to determine whether the on-line capacity as indicated in the resource plan is located in the right place on the network to serve the forecasted demand. If the congestion or violation exists, the generation scheduling based on the sensitivities calculations can determine whether or not a different allocation of the available resources could resolve the congestion or violation problem. In the early energy market, the transmission losses are neglected for reasons of computational simplicity, but are recently addressed in the Standard Market Design (SMD) [2-4]. The loss calculation is considered for the dispatch functions of SMD such as locationbased marginal prices (LMP). Loss allocation does not affect generation levels or power flows; however it does modify the value of LMP [5]. The early and classic loss calculation approach is the loss formula B coefficient method [6], which is replaced by the more accurate inverse Jacobian transpose method [7]. Numerous of loss calculation methods have been proposed in the literature and can be categorized into pro-rata [8], incremental [9], proportional-sharing [10], and Z-bus loss allocation [11]. The calculation of loss sensitivity is based on the distributed slack buses in the energy control center [6, 11-13]. In the real-time energy markets, LMP or economic dispatch is implemented based on market-based reference, which is an arbitrary slack bus, instead of the distributed slack buses in the traditional energy management system. Meanwhile, the existing loss calculation methods in traditional EMS systems are generally based on the generator slacks or references. Since the units with automatic generation control (AGC) are selected as the distributed slacks, and the patterns or status of AGC units are variable for the different time periods in the real time energy market, the sensitivity values will keep changing, which complicates the issue. This chapter presents a fast and useful formula to calculate loss sensitivity for any slack bus [14]. The simultaneous feasibility test (SFT) performs the network sensitivity analysis under the base case and contingency cases in the power system. The base case and post-contingency MW flows are compared against their respective limits to generate the set of critical constraints. For each critical constraint, SFT calculates constraint coefficients (shift factors) that represent linearized sensitivity factors between the constrained quantity (e.g. MW branch flow) and MW injections at network buses. The B-matrix used to calculate the shift factors is constructed to reflect proper network topology.

Sensitivity Calculation in Real Time Transmission Network and Energy Markets

267

The objective of SFT is to identify whether or not network operation is feasible for a real power injection scenario. If operational limits are violated, generic constraints are generated that can be used to prevent the violation if presented with the same network conditions. In the energy market systems, the trade is often considered between the source and the sink (i.e., the point of resource, POR and point of demand, POD). The source and the sink may be an area or any bus group. Therefore, the area based sensitivities are needed, which can be computed through the constraint shift factors within area. Voltage sensitivity analysis can detect the weak buses/nodes in the power system where the voltage is low. It can be used to select the optimal locations of VAR support service [1520]. According to the sensitivity values voltage benefit factor (VBF) and loss benefit factor (LBF), a ranking of VAR support sites can also be obtained. This chapter presents the implementation details of the several sensitivities calculations in the practical transmission network and energy markets. Section 2 describes the calculation of the market-based loss sensitivities. Section 3 describes the implementation of SFT and the calculation of the constraints shift factors. Section 4 describes the calculation of the voltage sensitivity. Section 5 shows the simulation results of the above-mentioned sensitivities.

II. LOSS SENSITIVITY CALCULATION


This section presents a fast and useful formula to calculate loss sensitivity for any slack bus. The formula is based on the loss sensitivity results from the distributed slacks without computing a new set of sensitivity factors through the traditional power flow calculation. Especially, the loads are selected as the distributed slacks rather than the usual generator slacks. The loss sensitivity values will be the same for the same network topology no matter how the status of the AGC units changes. In the energy market, the formulation of the optimum economic dispatch can be represented as follow:

Min F = C j Pj
j

j NG

(1)

such that

s.t.

P
ij

+ PL = PGj
j

j NG

(2) (3) (4)

S
j

Pj Pi max

j NG , i K max

PGj min PGj PGj max


where PD: The real power load.

j NG

268 Pimax: PGj: PGjmin: PGjmax: PL: Sij: Cj : Kmax: NG:

Jizhong Zhu The maximum requirement of power supply at the active constraint i. The real power output at generator bus j. The minimal real power output at generator j. The maximal real power output at generator j. The network losses. The sensitivity (shift factor) for resource or unit j and active constraint i with respect to the market-based reference. The real time price for the resource (or unit) j. The maximum number of active constraints. The number of units.

The Lagrangian function is obtained from equations (1) and (2).

FL = f i ( PDi ) + ( PDi + PL PGj )


i i j

(5)

Traditionally, generation reference (single or distributed slack) is used in the calculation of loss allocation. This works, but may be inconvenient or confusing for the users who frequently use the loss factors. The reason is that the AGC status or patterns of units are variable in the real time EMS or energy markets. The loss sensitivity values based on the distributed unit references will keep changing due to the change of unit AGC status. Thus, the distributed load slack or reference is used here. The optimality criteria of the Lagrangian function (5) are written as follow:

FL PL df = i + 1 + P = 0 i ND PDi dPDi Di P FL df = i + L 1 = 0 j NG P PGj dPGj Gj df i LDi = i ND dPDi

(6)

(7)

(8)

LDi =

1 P 1+ L PDi

i ND

(9)

dfi LGj = dPGj

j NG

(10)

Sensitivity Calculation in Real Time Transmission Network and Energy Markets

269

LGj =

1 P 1 L PGj

j NG

(11)

where, : the Lagrangian multiplier. the loss sensitivity with respect to load at bus i. the loss sensitivity with respect to unit at bus j.

PL : PDi
PL : PGj
We use both

PL , which is the loss sensitivity with respect to an injection at bus i, stand for Pi

PL PL and . Since the distributed slack buses are used here, all loss sensitivity PDi PGj

factors are non-zero. If an arbitrary slack bus, k, is selected, then Pk is the function of the other injections, i.e.

Pk = f ( Pi )

i n, i k

(12)

where n is the total number of buses in the system, and Pi is the power injection at bus i, which includes the load PDi and generation PGj. Actually, the load can be treated as a negative generation. Then equation (9) and (11) can be expressed as (13), and equation (8) and (10) can be expressed as (14).

Li =

1 P 1 L Pi

in

(13)

df i Li = dPi
Equation (2) will be rewritten as

in

(14)

PL = Pk + Pi
ik

in

(15)

270

Jizhong Zhu

The new Lagrangian function can be obtained from (1) and (15).

FL* = f i ( Pi ) + ( PL Pk Pi )
i in

(16)

The optimality criteria can be obtained from the Lagrangian function (16).

PL Pk FL* df i df k Pk + 1 = + P P = 0 in, ik Pi dPi dPk Pi i i


From (15), we get

(17)

PL P =1+ k Pi Pi
From (17) and (18), we get

(18)

df i * df k Li = dPk dPi
L* i = 1 P 1 L Pi i n, i k

(19)

(20)

It is noted that Li and Li* are similar, but they have different meaning [14]. The former is based on the distributed slack buses, and the latter is based on an arbitrary slack bus k. Similarly, the loss sensitivity in Li is based on the distributed slack, i.e.

PL Pi

(The
DS

subscript DS means the distributed slack); the loss sensitivity in Li* is based on an arbitrary single slack bus k, i.e.

PL . Note that the k-th loss sensitivity, with bus k as the slack bus, is Pi k

zero. From (14) and (19), we have the following equation.

L* i =

Li , L* k =1 Lk

(21)

From the above equations (13), (20) and (21), we get

Sensitivity Calculation in Real Time Transmission Network and Energy Markets

271

1 P 1 L Pi
1

1 =
k

PL Pk

DS

P 1 L Pi
PL Pi

(22)

DS

PL = Pi k

DS

P 1 L Pk

(23)

DS

Hence, with one set of the incremental transmission loss coefficients for the distributed slack buses, the loss sensitivity for an arbitrary slack bus can be calculated from the following formula.

PL = Pi k

PL Pi

DS

PL Pk

DS

P 1 L Pk

(24)

DS

The formula of loss sensitivity calculation is very simple, but is accurate and efficient for real-time energy markets. It will avoid computing a new set of the loss sensitivity factors whenever the slack bus k changes. Consequently, it means huge time savings. In addition, the loss factors based on the distributed load reference will not be changed no matter how the AGC statuses of units vary, as long as network topology is the same as before.

III. IMPLEMENTATION OF SFT


The objective of SFT is to identify whether or not network operation is feasible for a real power injection scenario. If operational limits are violated, generic constraints and the corresponding sensitivities (the shift factors) are generated, which can be used to prevent the violation if presented with the same network conditions. Meanwhile, the shift factors can also be used in the generation scheduling or economic dispatch to alleviate the overload of transmission lines. The SFT calculations include the contingency analysis (CA), in which the decoupled power flow (DPF) or DC power flow is used. The set of component changes that can be analyzed include transmission line, transformer, circuit breaker, load demand and generator outages. SFT informs the user of contingencies that could cause conditions violating

272

Jizhong Zhu

operating limits. These limits include branch overloads, abnormal voltages, and voltage angle differences across specified parts of the network. SFT reports the sensitivity (shift factor) of the constraint with respect to the controls. These controls include unit MW control, phase shifter, and load MW control.

3.1. Unit MW Control


The unit MW control is the most efficient and cheap control among these available controls. The formulation of sensitivity for unit can be written as follows.

S ij =

Ki Uj

i = 1,......, K max ,

j = 1,......U max

(25)

where, Sij: Ki: Uj: Kmax: Umax: The sensitivity of the constraint i to the unit MW control j. The constraint i. The unit MW control j. The maximum number of constraints. The maximum number of generator unit MW controls.

3.2. Phase Shifter Control


The phase shifter is another efficient control among these available controls. There are some assumptions for phase shifter in the SFT design. The phase shifter control variable is tap number. Normally tap number is an integer, but it can be handled as a real number in the practical SFT calculation. In addition, all opened phase shifters will be skipped over, that is, the sensitivity for the phase shifter that is open at any end will not be calculated. The step on the tap-type is the sensitivity of angle with respect to tap number. The formulation of sensitivity for phase shifter can be written as follows.

Sijp =
where, Sijp: Ki: PSjp: Kmax: PSmax:

Ki PS jp

i = 1,......, K max ,

jp = 1,......PSmax

(26)

The sensitivity of the constraint i to the phase shifter control jp. The constraint i. The phase shifter control jp. The maximum number of constraints. The maximum number of phase shifter controls.

Sensitivity Calculation in Real Time Transmission Network and Energy Markets

273

It is noted that there is a special branch in constraint logic that must be implemented when the phase shifter branch itself is in the constraint. Basically the artificial flow through transformer branch must be subtracted from constraint flow.

3.3. Load MW Control


The load MW control should be last control when other controls are not available. The formulation of sensitivity for load MW control can be written as follows.
S ijd = Ki LD jd i = 1,......, K max , jd = 1,...... LD max

(27)

where, Sijd: Ki: LDjd: Kmax: LDmax: The sensitivity of the constraint i to the load MW control jd. The constraint i. The load MW control jd. The maximum number of constraints. The maximum number of load MW controls in whole system.

It is noted that the sensitivity sign for load MW control is negative. The reason is that increasing load will cause more serious constraint violation, rather than reduce the constraint violation. According to the sensitivity relationship between the constraint and the load MW control, it is needed to reduce / shed load for alleviating or deleting the constraint violation.

3.4. Constraint Value


For each constraint, constraint value (DC value) is computed from the control values multiplied by sensitivities. The formulation can be written as follows.

DCVAL i =
where, DCVALi: VAL_Uj: Sij: Umax:

U max j =1

VAL _ U

(28)
j

* S ij

The constraint value for the constraint i. The value of control j. Here, controls including unit MW control, phase shifter and load MW control. The sensitivity or shift factor of the constraint i to the control j. The maximum number of controls.

274

Jizhong Zhu

The shift factors computed in SFT is based on the reference bus in EMS topology, but it can be easily converted to any market-based reference. Let k be market-based reference unit, and the shift factor of the constraint i with respect to any unit j that is obtained based on EMS reference bus is Sij. For unit k, the shift factor of the constraint i is Sik. Then, the shift factors after converting to market-based reference unit k can be computed as follows.

S ik ' = 0
S ij ' = S ij S ik
where,

i = 1,......, K max
i = 1,......, K max , jk

(29) (30)

Sij: The shift factor of the constraint i with respect to unit j that is based on the marketbased reference k. Sik: The shift factor of the constraint i with respect to unit k that is based on the marketbased reference k. Let PFj be the participation factor of unit j, and the shift factor of the constraint i with respect to any unit j is Sij. For area A, the area based shift factor of the constraint i is SiA, which can be computed as follows.

S iA =

(PF S )
j A j ij

PF
j A

i = 1,......, K max ,

j A

(31)

where, SiA: PFj: The area based shift factor of the constraint i. The participation factor of the unit j.

As we know that the shift factor of the constraint is related to the selected reference, i.e., the value of shift factor will be different if the reference is different even the system topology and conditions are the same. Sometimes the system operators would like to have the stable shift factor values without caring about the selection of reference bus/unit. Thus, the distributed load reference will be used to get the unique constraint shift factors if the system topology and conditions are unchanged. Let Sildref be the sensitivity of load distribution reference for the constraint i, and the shift factor of the constraint i with respect to any control j that is obtained based on EMS reference bus is Sij. Then, the shift factors based on the load distribution reference LDREF can be computed as follows.

Sij ' = Sij Sildref

i = 1,......,Kmax

(32)

Sensitivity Calculation in Real Time Transmission Network and Energy Markets where Sildref: the sensitivity of load distribution reference for the constraint i, that is,
LDmax

275

Sildref =

(S

jd =1 LDmax jd =1

ijd

LDjd ) i = 1,......, Kmax


jd

(33)

LD

In the practical energy markets such as independent system operator (ISO), the system consists of many areas but one is a major area in the ISO system that is called the internal area, and others are called as external areas. If the internal area is major concerned during the price calculation for this market system, the load distribution reference can be selected based on the internal area only. Similarly, Let LDAmax be the total number of load controls in the internal area of ISO system, which is less than the total number of load controls in whole ISO system LDmax. The shift factors based on the area load distribution reference LDAREF can be computed as follows.

Sij ' = Sij Sildaref


where

i = 1,......,Kmax

(34)

Sildaref: the sensitivity of load distribution reference in area A for the constraint i, that is,
LDAmax

S ildaref =

(S
jd =1

ijd

LD jd ) i = 1,......, K max LDAmax LDmax


jd

(35)

LDAmax jd =1

LD

LDAmax:

The maximum number of load MW controls in area A.

IV. VOLTAGE SENSITIVITY ANALYSIS


The purpose of the voltage sensitivity analysis is to improve the voltage profile and to minimize system real power losses through the optimal VAR control. These goals are achieved by proper adjustments of VAR variables in power networks. Therefore, if the voltage magnitude at generator buses, VAR compensation (VAR support) and transformer tap position are chosen as the control variables, the optimal VAR control model can be represented as:

276 min PL(QS, VG, T) such that Q(QS, VG, T, VD) = 0 QGmin QG(QS, VG, T) QGmax VDmin VD(QS, VG, T) VDmax QSmin QS QSmax VGmin VG VGmax Tmin T Tmax where VG: QS: QG: T: VD:

Jizhong Zhu (36)

(37) (38) (39) (40) (41) (42)

the voltage magnitude at generator buses. the VAR support in the system. the VAR generation in the system. the tap position of the transformer. the voltage magnitude at load buses

Two kinds of sensitivity-related factors can be computed through (36) (42). Here they are called as voltage benefit factors (VBF) and loss benefit factors (LBF), which are expressed as follows.

LBFi =

( PL 0 PL (Qsi )) Qsi

100% i ND

(43) (44)

VBFi =

(Vi (Q si ) Vi 0 ) Qsi

100% i ND

where Qsi: LBFi: VBFi: PL0: PL(Qsi): Vi0: the amount of VAR support at the load bus i. the loss benefit factors from the VAR compensation Qsi . the voltage benefit factors from the VAR compensation Qsi . power transmission losses in the system without VAR compensation. the power transmission losses in the system with VAR compensation Qsi. the voltage magnitude at load bus i without VAR compensation.

Sensitivity Calculation in Real Time Transmission Network and Energy Markets Vi(Qsi): ND: the voltage magnitude at load bus i with VAR compensation Qsi. the number of load buses.

277

V. SIMULATION RESULTS
The calculation results of the several sensitivities are illustrated using the IEEE 14 bus system and AREVA T&D 60-bus system. The one-line diagram of the AREVA T&D 60-bus system is shown in Figure 1. The 60-bus system, which has three areas, consists of 24 generation units (15 units are available in the tests), 32 loads, 43 transmission lines and 54 transformers.
CHENAUX CHFALLS MARTDALE NANTCOKE

ECAR

BRIGHTON

CEYLON

HUNTVTIL
RICHVIEW

MITCHELL
REDBRIDG

KINCARD HEARN HANDOVER

PICTON PARKHILL MTOWN


HOLDEN

STRATFRO LAKEVIEW

WEST
COBDEN BVILLE

JVILLE

EAST
DOUGLAS

GOLDEN

STINSON

WVILLE WALDEN

Figure 1. One-line diagram of AREVA T&D system (Area 1 - EAST, Area 2 - WEST, Area 3 ECAR).

278

Jizhong Zhu The following test cases are used to analyze the loss sensitivity in this chapter: Case 1: Case 2: Calculate loss sensitivities using the distributed generation slack and load slack, respectively. All units have AGC on. Calculate loss sensitivities using the distributed generation slack and load slack, respectively. All units have AGC on except the units under station Douglas in Area 1 Calculate loss sensitivities using the distributed generation slack and load slack, respectively. All units have AGC on except the units under station HEARN in Area 1 Calculate loss sensitivities using the distributed generation slack and load slack, respectively. All units have AGC on except the units in Area 2 Calculate loss sensitivities using the distributed generation slack and load slack, respectively. All units have AGC on except the units under station HOLDEN in Area 3 Calculate loss sensitivities for the selected single slack based on the loss factors under the distributed slack.

Case 3:

Case 4: Case 5:

Case 6:

The simulation results are shown in Table 16. All loss sensitivity factors for units and loads are computed. In order to reduce the length of the chapter, only loss sensitivities of generators are listed in Table 16, in which column 1 is the name of station and units. Column 2 is the area number that the unit belongs to. Column 3 is the AGC status of the unit. Tables 15 are the test results and comparison of loss sensitivity calculation based on the distributed generation reference and distributed load reference, respectively. The loss factors computed from the distributed unit reference are listed in column 4 of Table 15. The loss factors computed from the distributed load reference are listed in column 5 of Table 15. Generally, the values of loss sensitivities based on the generation reference are different from those based on the load reference, because the distribution of the units is not exactly the same as the distribution of loads in the power system. The loss factors will be close or equal if the units are close to the load locations. This can be observed from Table 1, where all units are on AGC status. For the 60-bus system, each load in area 3 has at least one unit connected, so the loss factors in area 3 are the same for both the distributed generation slack and distributed load slack. It is noted that from Table 1 5 that the loss sensitivity factors based on the distributed load slack are the same whether the status of the units is changed or not. But the loss factors based on the distributed generation references are changed since the AGC status of the units are different. Generally, the change of AGC status of the units only affects the loss sensitivities in the same area that these units belong to. It can be seen from Table 2 3 that, when AGC status of the units in area 1 changes, only the loss factors in area 1 is affected. The loss factors in the other areas are unchanged. For Table 5, when AGC status of the units in area 3 changes, only the loss factors in area 3 is affected. The loss factors in the other areas are unchanged. But for Table 4, when AGC status of the units in area 2 changes and all units in this area are not on AGC, it means that there is no unit reference in area 2. Then the units with AGC on in the other areas will pick up the

Sensitivity Calculation in Real Time Transmission Network and Energy Markets

279

mismatch (i.e. area 1 in this case). Thus, the loss factors in area1 and 2 are changed. The loss factors in the other areas are unchanged. Table I. Test Results and Comparison of Loss Sensitivity Calculation (Case 1: All units on AGC)
Loss Sensitivity Distributed generation Slack 0.015100 0.012100 0.009900 0.009900 0.009700 -0.016500 -0.016500 -0.018800 -0.001000 0.000700 -0.008900 0.021200 0.021200 0.001000 -0.012200 Loss Sensitivity Distributed load Slack 0.017000 0.014000 0.011800 0.011800 0.011600 -0.014600 -0.014600 -0.017000 -0.004200 -0.002500 -0.008900 0.021200 0.021200 0.001000 -0.012200

Station, Generator DOUGLAS, G2 DOUGLAS, G1 DOUGLAS, CT1 DOUGLAS, CT2 DOUGLAS, ST HEARN, G1 HEARN, G2 LAKEVIEW, G1 BVILLE, 1 WVILLE, 1 CHENAUX, 1 CHEALLS, 1 CHEALLS, 2 HOLDEN, 1 NANTCOKE, 1

Area No. 1 1 1 1 1 1 1 1 2 2 3 3 3 3 3

AGC Unit YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES

Table II. Test Results and Comparison of Loss Sensitivity Calculation (Case 2: All units on AGC except the units under station Douglas in Area 1)
Loss Sensitivity Distributed generation Slack 0.032800 0.029900 0.027800 0.027800 0.027600 0.001500 0.001500 -0.000800 -0.001000 0.000700 -0.008900 0.021200 0.021200 0.001000 -0.012200 Loss Sensitivity Distributed load Slack 0.017000 0.014000 0.011800 0.011800 0.011600 -0.014600 -0.014600 -0.017000 -0.004200 -0.002500 -0.008900 0.021200 0.021200 0.001000 -0.012200

Station, Generator DOUGLAS, G2 DOUGLAS, G1 DOUGLAS, CT1 DOUGLAS, CT2 DOUGLAS, ST HEARN, G1 HEARN, G2 LAKEVIEW, G1 BVILLE, 1 WVILLE, 1 CHENAUX, 1 CHEALLS, 1 CHEALLS, 2 HOLDEN, 1 NANTCOKE, 1

Area No. 1 1 1 1 1 1 1 1 2 2 3 3 3 3 3

AGC Unit NO NO NO NO NO YES YES YES YES YES YES YES YES YES YES

280

Jizhong Zhu Table III. Test Results and Comparison of Loss Sensitivity Calculation (Case 3: Only units under HEARN in Area 1 not on AGC)
Loss Sensitivity Distributed generation Slack 0.012600 0.009600 0.007400 0.007400 0.007200 -0.019000 -0.019000 -0.021300 -0.001000 0.000700 -0.008900 0.021200 0.021200 0.001000 -0.012200 Loss Sensitivity Distributed load Slack 0.017000 0.014000 0.011800 0.011800 0.011600 -0.014600 -0.014600 -0.017000 -0.004200 -0.002500 -0.008900 0.021200 0.021200 0.001000 -0.012200

Station, Generator DOUGLAS, G2 DOUGLAS, G1 DOUGLAS, CT1 DOUGLAS, CT2 DOUGLAS, ST HEARN, G1 HEARN, G2 LAKEVIEW, G1 BVILLE, 1 WVILLE, 1 CHENAUX, 1 CHEALLS, 1 CHEALLS, 2 HOLDEN, 1 NANTCOKE, 1

Area No. 1 1 1 1 1 1 1 1 2 2 3 3 3 3 3

AGC Unit YES YES YES YES YES NO NO YES YES YES YES YES YES YES YES

Table IV. Test Results and Comparison of Loss Sensitivity Calculation (Case 4: All units on AGC except the units in Area 2)
Loss Sensitivity Distributed generation Slack 0.015200 0.012200 0.010000 0.010000 0.009900 -0.016700 -0.016700 -0.019100 -0.021000 -0.019300 -0.008900 0.021200 0.021200 0.001000 -0.012200 Loss Sensitivity Distributed load Slack 0.017000 0.014000 0.011800 0.011800 0.011600 -0.014600 -0.014600 -0.017000 -0.004200 -0.002500 -0.008900 0.021200 0.021200 0.001000 -0.012200

Station, Generator DOUGLAS, G2 DOUGLAS, G1 DOUGLAS, CT1 DOUGLAS, CT2 DOUGLAS, ST HEARN, G1 HEARN, G2 LAKEVIEW, G1 BVILLE, 1 WVILLE, 1 CHENAUX, 1 CHEALLS, 1 CHEALLS, 2 HOLDEN, 1 NANTCOKE, 1

Area No. 1 1 1 1 1 1 1 1 2 2 3 3 3 3 3

AGC Unit YES YES YES YES YES YES YES YES NO NO YES YES YES YES YES

Sensitivity Calculation in Real Time Transmission Network and Energy Markets Table V. Test Results and Comparison of Loss Sensitivity Calculation (Case 5: All units on AGC except unit 3 under station HOLDEN in Area 3)
Loss Sensitivity Distributed generation Slack 0.015100 0.012100 0.009900 0.009900 0.009700 -0.016500 -0.016500 -0.018800 -0.001000 0.000700 -0.008500 0.021600 0.021600 0.001400 -0.011800 Loss Sensitivity Distributed load Slack 0.017000 0.014000 0.011800 0.011800 0.011600 -0.014600 -0.014600 -0.017000 -0.004200 -0.002500 -0.008900 0.021200 0.021200 0.001000 -0.012200

281

Station, Generator DOUGLAS, G2 DOUGLAS, G1 DOUGLAS, CT1 DOUGLAS, CT2 DOUGLAS, ST HEARN, G1 HEARN, G2 LAKEVIEW, G1 BVILLE, 1 WVILLE, 1 CHENAUX, 1 CHEALLS, 1 CHEALLS, 2 HOLDEN, 1 NANTCOKE, 1

Area No. 1 1 1 1 1 1 1 1 2 2 3 3 3 3 3

AGC Unit YES YES YES YES YES YES YES YES YES YES YES YES YES NO YES

Table VI. Test Results of Loss Sensitivity Calculation (Distributed Slack vs Single Slack)
AGC Unit YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES Loss Sensitivity Distributed Slack 0.017000 0.014000 0.011800 0.011800 0.011600 -0.014600 -0.014600 -0.017000 -0.004200 -0.002500 -0.008900 0.021200 0.021200 0.001000 -0.012200 Loss Sensitivity Single Slack, HOLDEN 1 0.016016 0.013013 0.010811 0.010811 0.010611 -0.015616 -0.015616 -0.018018 -0.005205 -0.003504 -0.009910 0.020220 0.020220 0.000000 -0.013213 Loss Sensitivity Single Slack, Douglas ST 0.005463 0.002428 0.000202 0.000202 0.000000 -0.026507 -0.026507 -0.028936 -0.015985 -0.014265 -0.020741 0.009713 0.009713 -0.010724 -0.024079

Station, Generator DOUGLAS, G2 DOUGLAS, G1 DOUGLAS, CT1 DOUGLAS, CT2 DOUGLAS, ST HEARN, G1 HEARN, G2 LAKEVIEW, G1 BVILLE, 1 WVILLE, 1 CHENAUX, 1 CHEALLS, 1 CHEALLS, 2 HOLDEN, 1 NANTCOKE, 1

Through the above comparisons, it can be observed that the method of the distributed load references for loss sensitivity calculation is superior to that of the distributed generation

282

Jizhong Zhu

references in the real time energy markets, since the AGC status of the units are changeable in the real time system. The results of loss sensitivity calculation for a single slack, which are computed from the proposed formula (24), are shown in Table 6. Column 3 in Table 6 is the set of the loss sensitivity coefficients for the distributed slack buses. Column 4 in Table 6 is the set of loss sensitivity factors with a single slack bus at the location of HOLDEN 1. Column 5 in Table 6 is the set of loss sensitivity factors with a single slack bus at the location of Douglas. It is noted that all the loss sensitivities are nonzero if the distributed slack is selected. If the single slack is selected, the loss sensitivity of the slack equals zero. Since the loss sensitivity values based on the distributed slacks from EMS are unchanged as long as the system topology is the same, the loss sensitivities for any market-based single slack can be easily and quickly acquired by use of the loss sensitivity formula (24). Therefore, a large amount of the computations are avoided whenever the loss sensitivities for a marketbased reference are needed in the real time energy markets. Table 7 8 are the results of the detected constraint and the corresponding shift factors. The results of one constraint that is branch T525 at Station CHENAUX are listed. Table VII. Example of the Active Constraint (Branch T525 At Station Chenaux)
Constraint name Branch T525 Rating (MVA) 1171.4 Actual Flow (MVA) 1542.7 Constraint deviation 371.3 Percent of Violation 131.7

Table VIII. Test Results of SFT (Shift Factors) Calculation for the Active Constraint T525 at Station Chenaux
Shift Factors on EMS Reference at Station DOUGLAS 0.000000 0. 000000 0. 000000 0. 000000 0. 000000 0. 000000 0. 000000 0. 000000 -0.013650 -0.024336 0.617887 0.521795 0. 521795 0.304269 0.291815 Shift Factors on Market Reference at Station HOLDEN -0.304269 -0.304269 -0.304269 -0.304269 -0.304269 -0.304269 -0.304269 -0.304269 -0.317919 -0.328605 0.313618 0. 217526 0.217526 0.000000 -0.012454

Station, Generator DOUGLAS, G2 DOUGLAS, G1 DOUGLAS, CT1 DOUGLAS, CT2 DOUGLAS, ST HEARN, G1 HEARN, G2 LAKEVIEW, G1 BVILLE, 1 WVILLE, 1 CHENAUX, 1 CHEALLS, 1 CHEALLS, 2 HOLDEN, 1 NANTCOKE, 1

Area No.

Unit in Serve YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES

1 1 1 1 1 1 1 1 2 2 3 3 3 3 3

Sensitivity Calculation in Real Time Transmission Network and Energy Markets

283

In Table 8, column 1 is the name of station and units. Column 2 is the area number that the unit belongs to. Column 3 is the AGC status of the unit. Column 4 is the set of the shift factors of the constraint T525 with respect to the units for the EMS-based reference at station DOUGLAS. Column 5 is the shift factors of the constraint T525 with respect to the units for the market-based reference at the location of HOLDEN 1. It is noted that all the shift factors are zero for the units in area 1 for the EMS-based reference since the reference is located in area 1 and all units in area 1 are close to the reference unit. If the market-based slack is selected, the shift factors for the market-based reference can be easily obtained from equations (29) and (30). Table 9 shows the major VAR support sites as well as the corresponding benefit factors LBF and VBF for the IEEE-14 bus system. Table IX. Voltage Sensitivity Analysis Results for IEEE 14 Bus Systems
VAR support Site Bus 4 Bus 5 Bus 8 Bus 9 Bus 10 Bus 11 Bus 12 Bus 13 LBFi 0.000376 0.000337 0.002309 0.007674 0.002618 0.007407 0.006757 0.008840 VBFi 0.000855 0.000884 0.001775 0.001989 0.002097 0.002175 0.002268 0.002122

CONCLUSION
This chapter presents a practical approach to compute the sensitivities in the practical transmission network and energy markets. The analysis and implementation details of the loss sensitivity, voltage sensitivity, generator constraint shift factor, and area based constraint shift factor are discussed. The chapter also comprehensively discusses how to compute and use the sensitivities under the different references such as the market-based reference, and the energy management system based reference, as well as how to convert the sensitivities based on EMS system reference into the ones based on the market system reference. These sensitivities calculations can be used to determine whether the on-line capacity as indicated in the resource plan is located in the right place on the network to serve the forecasted demand. The proposed approach is tested on IEEE 14-bus system and AREVA T&D 60-bus system. The test results show the reported approaches are very fast, useful and efficient for the practical transmission network and energy markets.

284

Jizhong Zhu

REFERENCES
[1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] T.E. Dy-Liyacco, Control Centers Are Here to Stay, IEEE Computer Applications in Power, Vol.15, No.4, pp18-23, 2002. N. Winser, FERC's Standard Market Design: the ITC Perspective, 2002 IEEE PES Summer Meeting, Chicago, IL. July 22 - 26, 2002. A. Ott, Experience with PJM Market Operation, System Design, and Implementation, IEEE Trans. on Power Systems, Vol.18, No.2, pp528-534, 2003. D. Kathan, FERCs Standard Market Design Proposal, 2003 ACEEE/CEE National Symposium on Market Transformation, Washington, DC, April 15, 2003. J.Z. Zhu, D. Hwang, and A. Sadjadpour, The Implementation of Alleviating Overload in Energy Markets, in Proc. IEEE PES 2007 General Meeting, Tampa, Florida, 2007. L.K. Kirchamayer, Economic Operation of Power Systems, New York: Wiley, 1958. H.W. Dommel, and W.F. Tinney, Optimal power flow solutions, IEEE Trans. on PAS, Vol.PAS-87, No.10, pp1866-1876, 1968. M. Ilic, F.D. Galiana, and L. Fink, Power Systems Restructuring: Engineering and Economics. Norwell, MA: Kluwer, 1998. D. Kirschen, R. Allan, and G. Strbac, Contributions of individual generators to loads and flows, IEEE Trans. Power Systems, Vol.12, No.1, pp52-60, 1997. F. Schweppe, M. Caramanis, R. Tabors, and R. Bohn, Spot Pricing of Electricity, Norwell, MA: Kluwer, 1988. J. Conejo, F.D. Galiana, and I. Kochar, Z-Bus loss allocation, IEEE Trans. Power Systems, Vol.16, No.1, pp105-110, 2001. F.D. Galiana, A.J. Conjeo, and I. Korkar, Incremental transmission loss allocation under pool dispatch, IEEE Trans. Power Systems, Vol.17, No.1, pp26-33, 2002. Elgerd, Electric Energy Systems Theory: An Introduction, New York: McGraw-Hill, 1982. J.Z. Zhu, D. Hwang, and A. Sadjadpour, Loss Sensitivity Calculation and Analysis, in Proc. 2003 IEEE General Meeting, Toronto, July 13-18, 2003. J.Z. Zhu and M.R. Irving, Combined Active and Reactive Dispatch with Multiple Objectives using an Analytic Hierarchical Process, IEE Proc. C, Vol.143, No.4, pp344-352, 1996. J.Z. Zhu, and J.A. Momoh, Optimal VAR pricing and VAR placement using analytic hierarchy process, Electric Power Systems Research, Vol.48, No.1, pp11-17, 1998. M.O. Mansour, and T.M. Abdel-Rahman, Non-linear VAR Optimization Using Decomposition and Coordination, IEEE Trans. PAS, Vol. 103, pp. 246-255, 1984. N.H. Dandachi, M.J. Rawlins, O. Alsac, and B. Stott, OPF for Reactive Pricing Studies on the NGC System, IEEE Power Industry Computer Applications Conference, PICA95, Utah, pp. 11-17, May 1995. Alsac and B. Sttot, Optimal Power Flow with Steady-State Security, IEEE Trans., PAS, Vol.93, pp745-751, 1974. J.A. Momoh and J.Z. Zhu, Improved Interior Point Method for OPF Problems, IEEE Trans. on Power Systems, Vol.14, No.3, pp1114-1120, 1999.

[16] [17] [18]

[19] [20]

In: Leading-Edge Electric Power Research Editor: C. M. OSullivan, pp. 285-316

ISBN: 978-1-60021-981-8 2008 Nova Science Publishers, Inc.

Chapter 7

WIDE-AREA MONITORING AND ANALYSIS OF INTER-AREA OSCILLATIONS USING THE HILBERT-HUANG TRANSFORM
A. R. Messinaa, M. A. Andradeb and E. Barocio c
a

The Center for Research and Advanced Studies (Cinvestav), Mexico b The Autonomous University of Nuevo Len, Mexico c The University of Guadalajara, Mexico

ABSTRACT
Many transient processes in power systems involve phenomena that vary in time and space in complicated ways. Comprehensive monitoring of large-scale power systems by means of properly placed time-synchronized phasor measurement units (PMUs) provides the opportunity to analyze and characterize complex inter-area swing dynamics involving all or most of the power system. Wide-area real-time monitoring may prove invaluable in power system dynamic studies by giving a quick assessment of the damping and frequency content of dominant system modes after critical contingencies. Measured data, however, may exhibit quite different dynamics at each system location or exhibit abrupt changes, dynamic irregularities, or be complicated by nonlinear trends or noise. Traditional Fourier and Prony methods for system identification are unable to resolve the localized nature of these processes and hence provide little useful information concerning the nature of noisy, time-varying oscillatory processes. In this Chapter, a new method for analyzing the temporal dynamics of nonlinear and non-stationary inter-area oscillations using a local empirical mode decomposition (EMD) method and the Hilbert transform is presented. Two novel algorithms are developed to address nonlinear and non-stationary issues. The first method is a local implementation of the empirical mode decomposition technique. The second is an algorithm to compute the Hilbert transform using finite impulse response (FIR) filters. By combining these approaches, the method can be used to analyze complex signals for which the conventional assumptions of linearity and stationarity may not apply and can be implemented for on-line estimation of modal damping and frequency using synchronized wide-area measurement systems.

286

A. R. Messina, M. A. Andrade and E. E. Barocio


The physical mechanism underlying nonlinear time-varying inter-area oscillations is investigated and methods to characterize the observed oscillatory phenomena in terms of physically meaningful modal components are proposed. Emphasis is placed on identifying modal content in the presence of noise and nonlinear trends. Issues concerning the implementation of the method and numerical considerations are also discussed. As specific applications, data obtained from PMU measurements from a real event in the northern systems of the Mexican interconnected system are used to examine the potential usefulness of nonlinear time series analysis techniques to characterize the spatio-temporal characteristics of the observed oscillations and to determine the nature and propagation of the system disturbance. The efficiency and accuracy of the method is demonstrated by comparison to other approaches.

BACKGROUND ON THE HILBERT-HUANG TECHNIQUE


Nonlinear, non-stationary behavior plays an important role in a variety of physical processes but it may be hard to identify and quantify. In this section, the combined use of Hilbert spectral analysis and the EMD method to characterize the time evolution of nonlinear, non-stationary processes is discussed.

The Empirical Mode Decomposition Technique


The empirical mode decomposition is a time-series analysis method that decomposes a signal, x(t ) , into essentially band-limited components or basis functions, a requirement to get meaningful instantaneous frequencies [1,2], using information from the data itself. The essence of this technique is to identify the basis oscillatory functions by their characteristic time scales and then decompose the signal into a series of temporal modes called intrinsic mode functions (IMFs) given by

x(t ) =

c (t ) + r (t ),
i i =1

(1.1)

where n is the number of IMF components, and r (t ) is the residue after the n IMFs have been extracted; the functions ci (t ) are nearly orthogonal and have zero local means. Each IMF is associated with a local, physical time scale and can be amplitude and/or frequency modulated and even non-stationary. The first IMF accounts for the higher frequency oscillations, while each succeeding component accounts for lower average frequencies. An IMF is defined as a wave in which [1]: (i) in the whole time span of the signal, the number of extremes, namely maxima and minima, and the number of zero crossings must be equal or differ at most by one, and (ii) at any time instant, the mean value of the amplitudes defined by the local maxima and minima, must be zero. In practice, however, only a set of IMFs contain relevant information to system behavior. As a result, we rewrite the basic model in (1.1) in the more useful form [3]

Wide-Area Monitoring and Analysis of Inter-Area Oscillations

287

x(t ) =

j =1

c j (t ) +

l = p +1

c (t ) + r (t ).
n l

(1.2)

where the terms c j (t ) , j = 1,K , p contain the physical behavior of interest, and the remaining n p terms contain uninteresting, non-sinusoidal characteristics. Given a model of the form (1.2), it is possible to apply Hilbert transform to determine local characteristics of the data. Following the work of Huang et al. [1], the original signal x(t ) can be expressed as the real part of the complex expansion

x(t ) =

j =1

c j (t ) + r (t ) = Re

j =1

Aj (t )e 0

i j ( t ) dt

(1.3)

where A j , j are the instantaneous amplitude and frequency of the j-th modal component. The IMFs are symmetric, have a unique local frequency, are nearly orthogonal, i.e.

IMFi , IMFj 0 , for i j and form a complete basis; the sum of the IMFs equals the
original series. Although the HHT technique can be efficiently used to characterize nonlinear, nonstationary oscillations several problems persist: (i) The IMFs are a mix of amplitude and frequency modulated signals. Extracting from these components the underlying dynamics is not easy (ii) Although experience suggest that the extracted IMFs usually represent recognizable physical characteristics of the data, the results are not supported by an underlying physical theory, (iii) Further, some IMFs may have no practical significance in the study of complex multi-component signals. This makes the analysis and interpretation of complex phenomena a difficult task. These basis functions are then processed through Hilbert analysis to obtain magnitude, phase and damping information as a function of time.

The Sifting Process


The basic method adopted to extract the IMFs consists of three steps [1,3]: a) Starting with the original signal, x(t ) , set hi (t ) = x(t ) , and extract the local minima and local maxima from hi (t ) ,

288

A. R. Messina, M. A. Andrade and E. E. Barocio b) Interpolate the local minima and local maxima with a cubic spline to form upper and lower envelopes respectively, and c) Obtain the mean of the envelopes, mi (t ) , and subtract it from hi (t ) to determine a new function hi +1 (t ) = hi (t ) mi (t ) . The three-step procedure is repeated until

hi +1 (t ) satisfies the criteria of an IMF and then c j (t ) = hi +1 (t ) .


This procedure is known as the sifting process. Although this method does not always guarantee a perfect instantaneous frequency over all conditions, the resulting instantaneous frequency is still consistent with the physics of the system under study. As pointed out in [3], an inherent aspect of the EMD is that each IMF represents a simple oscillatory mode as a counterpart of the simple harmonic function. Unlike other approaches, the decomposition of the original signal into intrinsic mode functions uses a direct, and adaptive method, which does not assume any basis. This makes it particularly attractive for the study of general signals. Once the original signal has been decomposed into set of intrinsic mode functions, the Hilbert transform can be applied to the IMF components to construct the energy-timefrequency distribution designated as the Hilbert spectrum. The following sections give a brief review of the Hilbert transform and describe the adaptation of proposed method to produce physically meaningful representations of nonlinear, and non-stationary data.

BACKGROUND: THE ANALYTIC SIGNAL OF GABOR


The Hilbert Transform
For a given real signal, u (t ) , its Hilbert transform is defined as

v(t ) =

u ( ) 1 d = t

u ( ) d , t

(1.4)

where P indicates that the Cauchy principal value of the integral is taken as the transform is an improper integral [2]. In a similar way, the Hilbert inverse transform is given by

u (t ) =

v( ) 1 d = P t

v( ) . t

(1.5)

These expressions can be written in a more convenient way in the form of convolutions as

v(t ) = u (t )

1 , t

(1.6)

Wide-Area Monitoring and Analysis of Inter-Area Oscillations

289

u (t ) = v(t )

1 . t

(1.7)

As shown in (1.6), the Hilbert transform returns a signal v(t ) with the same power as the original u (t ) but phase-shifted at each frequency by / 2 . Instead of evaluating the integral in (1.4), it is more practical to obtain the representation in the Gabor domain to take advantage of the analytical properties of the Frequency spectrum [4].

The Analytic Signal


The complex helical signal whose imaginary part is the Hilbert transform of the real signal,

(t ) = u (t ) + jv(t ),

(1.8)

is known as the analytic signal [4], where the real part, u (t ) , is the data itself and the imaginary part is given by the Hilbert transform of the signal in (1.4). The term analytic function is used in the sense of a complex function ( z ) of the complex variable

z = t + j .
In what follows, we briefly review existing techniques for computing the Hilbert transform, and propose a technique for its local calculation based on finite impulse response (FIR) filters.

CONTINUOUS-TIME ESTIMATION OF THE HILBERT TRANSFORM


Existing approaches to the numerical calculation of the Hilbert transform are based on the computation of the analytic signal using the Fourier transform. Details of this technique are given in [5], but a brief outline is provided here. Fourier-based techniques have a global character since they span the whole data range and hence, are not well adapted for characterization of local signal attributes. This limits its application to off-line studies requiring the full data set. Essentially, the practical implementation of the Hilbert transform using these approaches can be achieved by using the following steps [5]: Perform a Fourier transform of the data, and set all the Fourier coefficients with negative frequency to zero. Multiply the results by two, and perform and inverse Fourier transform. The result is the complex-valued Hilbert transform.

290

A. R. Messina, M. A. Andrade and E. E. Barocio

A major problem with this approach is that the Hilbert transform is a step function in frequency. This behavior can cause undesirable Gibbs phenomena resulting in ripples in the Hilbert spectrum at the end of the data set. This, in turn, creates errors in the instantaneous frequencies and amplitudes calculated from the affected regions. In the succeeding sections, a brief description of the method is discussed, followed by the mechanics of deriving the numerical approximations. First, some drawbacks and limitations of the existing frameworks for computation of the Fourier-based Hilbert transform are presented and the alternative algorithms are outlined. A new technique for the local computation of the Hilbert transform and the associated analytic signal is then introduced.

The Continuous-Time Hilbert Transform


The analytic signal (t ) associated with the signal x(t ) , is defined by

(t ) = x(t ) + xh (t ).
Taking the Fourier transform of (1.9) with respect to f , gives [5]

(1.9)

2 X ( f ), ( f ) = X (0), 0,
where

for f > 0 for f = 0, for f < 0

X(f ) =

x(t )e j 2 ft dt

is the Fourier transform of x(t ) . Since x(t ) is real, it is possible to show that its Fourier transform is complex conjugated and symmetric, i.e. X ( f ) = X ( f ) [2]. Based on these relationships, the analytic signal can be obtained, in continuous time, using the fast Fourier transform as follows: 1. Obtain the N-point FFT of the real signal x(t ) . Compute the Hilbert transform using the expression above. 2. Obtain the Fourier transform of the analytic signal as

X [0], 2 X [m], [m] = N X [ 2 ], 0,

for m = 0 for 1 m for m = N 2 for


N 2

N 2

+1 m N 1

(1.10)

Wide-Area Monitoring and Analysis of Inter-Area Oscillations

291

Compute the analytic signal in continuous-time using the N-point FFT inverse of (1.10). While the underlying idea is straightforward, this approach has some drawbacks that make it not too reliable. A further limitation is that it is necessary to use the whole of the signal to obtain instantaneous characteristics. In a variation to existing approaches a new algorithm for the local implementation of the Hilbert transform is proposed that circumvents these limitations and enables to track the temporal evolution of arbitrary signals, on an on-line basis.

REAL-TIME IMPLEMENTATION OF THE HILBERT TRANSFORM


Transient signals encountered in power systems and other applications are inherently non-stationary. This section explores approaches to extending Hilbert spectral techniques to analyze the local properties of general non-stationary signals. In this formulation, the Hilbert transform is developed using finite impulse response (FIR) filters whose frequency response is an approximation to the response of an ideal Hilbert transformer. Since the method is based on local information, this technique is well-suited for real-time applications.

The Discrete-Time Hilbert Transform


The linear time-invariant discrete-time system whose frequency response H (e ) is given by [6,7]
j

j, H (e j ) = j,

0 <, < 0.

(1.11)

is called a discrete-time Hilbert transformer, or an ideal 90-degree phase shifter as is has a unity magnitude and a phase angle of / 2 for 0 < < , and a phase angle of / 2 for < < 0 . The impulse response h[n] of a 90-degree phase shifter, or Hilbert transformer, corresponding to a frequency response

jX r (e j ) for 0 X i (e ) = j jX r (e ) for
j

is given by

h[n] =
or

1 2

je jn d

1 2

je j n d

292

A. R. Messina, M. A. Andrade and E. E. Barocio

2 sin 2 ( n / 2) , h[n] = n 0,

n 0, n = 0.
(1.12)

Figure 1.1 shows the impulse response of (1.12). Further, Figure 1.2 illustrates, in a qualitative manner, how the discrete-time Hilbert transformer may be used to form a complex analytic signal from a pair of real signals. The impulse response of the Hilbert transformer, given in (1.12) is not absolutely summable. Therefore,

H (e j ) =

h[n]e
n =

j n

(1.13)

converges to (1.13) just in the mean-square sense, making the Hilbert transformer a valuable theoretical concept corresponding to a no causal system with a system function existing only in a restricted sense. Nevertheless, it is possible to approximate the ideal Hilbert transformer by using FIR approximations with constant group delay. The characteristics of such approximations are illustrated by examples of Hilbert transformers applied to test signals.
0.8 2/ 0.6 0.4 0.2 h[n] 0 -0.2 -0.4 -0.6 -2/ -0.8 -8 -6 -4 -2 0 n 2 4 6 8 -2/7 -2/5 2/3 2/5 2/7

-2/3

Figure 1.1. Impulse response of an ideal Hilbert transformer or 90-degree phase shifter.

Wide-Area Monitoring and Analysis of Inter-Area Oscillations

293

Figure 1.2. Block diagram representation of the model used to create a complex sequence whose Fourier transform is one-sided.

Next, an efficient approach to compute the Hilbert transform using FIR filter is introduced. Also outlined are techniques to design the filters.

Hilbert Transformer Design and Implementation


Although, as pointed out in the previous section, ideal Hilbert transformers are no causal systems, and they are not realizable, an optimal approximation using the Parks-McClellan algorithm [6] was obtained using the Matlab function Remez. The Parks-McClellan algorithm designs optimum filters in the sense that the maximum error between the desired frequency response and the actual frequency response is minimized [7,8]. This algorithm, designed to make the error function of the filter to satisfy the set of necessary and sufficient conditions for optimality, is implemented in Matlab using either the remez function of the signal processing toolbox or the gremez function in the filter design toolbox. As suggested by the development above, the Parks-McClellan algorithm allows the design of high-order Hilbert transformers. In the practical implementation of the method, FIR filters of 100th order and 150th order have been found satisfactory for most practical applications in this work.

INSTANTANEOUS PARAMETERS OF THE ANALYTIC SIGNAL


(t ) = u (t ) + jv(t ) can be uniquely defined by introducing the notion of a rotating phasor
along the time axis in a three-dimensional space having u , v , and t axes, such that the length of the vector (i.e. its magnitude) is the magnitude of the instantaneous envelope and the rate of rotation of the vector corresponds to the instantaneous frequency. By changing from rectangular coordinates (u , v) to polar ones ( A, ) , it is obtained The concepts of instantaneous amplitude, phase and frequency of an analytic signal

in the Cartesian plane. (u , v) . The analytic signal is a vector that rotates about and advances

u (t ) = A(t ) cos [ (t ) ] ,

(1.14)

294

A. R. Messina, M. A. Andrade and E. E. Barocio

v(t ) = A(t ) sin [ (t ) ] ,


from which it follows that

(1.15)

(t ) = A(t )e j (t ) .
The analytic signal represents a time-dependent phasor with amplitude

(1.16)

A(t ) = u 2 (t ) + v 2 (t ),
and phase

(1.17)

(t ) = arctan

v(t ) . u (t )

(1.18)

Because an analytic signal is computed for each IMF, both local and global instantaneous characteristics can be defined. Figure 1.3 shows a two-dimensional representation of the rotating phasor in the complex plane. Here, successive pairs of values ( A(tk 1 ), (tk 1 )) ,

( A(tk ), (tk )) specify the trajectory of the rotating phasor. This notion forms the basis of the
developed algorithms in this research. Following Gabor [4], the time-dependent instantaneous angular frequency is the time derivative of the instantaneous phase, namely

(t ) =

&(t ) v(t )u & (t ) d v(t ) u (t )v = arctan . 2 2 dt u (t ) u (t ) + v (t )

(1.19)

The instantaneous frequency is then computed from (1.19) as

F (t ) =

(t ) 1 & (t ). = 2 2

(1.20)

On the basis of this representation, the use of Hilbert analysis to extract the instantaneous parameters of general nonlinear, non-stationary signals is discussed. The existing methods are generalized and techniques to compute energy relationships among system modes are proposed. In particular, a physically-based technique to extract the instantaneous signal parameters is proposed and a novel method to characterize synchrony among pairs of phase measurements is suggested.

Wide-Area Monitoring and Analysis of Inter-Area Oscillations

295

Figure 1.3. Rotating phasor in the complex plane representing the analytic signal.

Damping Evaluation
Nonlinear transient damping can play a crucial role in the long-term evolution of complex systems. In [3], a technique to compute transient damping was proposed using concepts from nonlinear mechanics. By extending this idea, a method is proposed for estimating linear and nonlinear damping parameters in non-linear, non-stationary signals. Consider a single-dof- system which has a linear spring and a general nonlinear element given by [9]

& & + 2ho ( A) x & + o ( A) x = 0 x


where

(1.21)

o ( A) is the instantaneous amplitude-dependent natural frequency of the system, and

ho ( A) is the amplitude-dependent instantaneous damping coefficient of the system. From


Hilbert spectral theory, the solution of the system can be expressed in the analytic signal form Z = x(t ) + j~ x (t ) , where ~ x (t ) is the Hilbert transform of x(t ) . In developing the solution for the nonlinear system (1.21) assume now that the the jth mode analytic signal (the solution) can be expressed in the form

Z (t ) = a (t )e i (t ) = j (t )e ( t ) +i ( t )

(1.22)

296

A. R. Messina, M. A. Andrade and E. E. Barocio

where j (t ) is the time-dependent amplitude of the jth IMF,

j (t ) is the exponential factor

characterizing the time-dependent decay of the waves in the j-th IMF due to damping, and

(t ) = (t )dt
0

is the exponential factor characterizing the time-dependent decay of the signal due to damping. Use of this assumption results in

& (t ) & (t ) = (t ) + + j (t ). (t ) (t )
Further, noting that

(1.23)

& (t ) & (t ) A Re , = (t ) A(t )


yields

(t ) =

& (t ) & (t ) A . A(t ) (t )

(1.24)

Straightforward computation shows that the critical damping ratio or damping loss factor is given by

(t ) =
where

(t , A) 2 o (t , A)

(1.25)

& (t ) / a(t ) . (t ) = 2a

By applying this procedure to each IMF, it becomes possible to predict or estimate local modal damping in complex nonlinear, non-stationary processes from measured or simulated data.

MODAL EXTRACTIONS VIA THE EMPIRICAL MODE DECOMPOSITION


Based upon previous results, let the EMD be split in an oscillatory part and a trend as follows:

Wide-Area Monitoring and Analysis of Inter-Area Oscillations

297

x(t ) =

c (t ) + r (t ) = c (t ) + c (t ) + r (t ),
j n j k n j =1 j =1 k = p +1

(1.26)

where the p dominant IMFs represent the oscillatory behavior of interest, while the rest of the terms are associated with featureless, often non-oscillatory characteristics. The Hilbert transform is then applied to the IMF components to determine the Hilbert spectrum. Application of the Hilbert transform to (1.26) yields

z (t )

j =1

c j (t ) + iH {c j (t )} =

A (t ) e
j j =1

i j ( t )

(1.27)

where H {c j (t )} is the Hilbert transform of c j (t ) , Aj (t ) =

c j (t ) 2 + H {c j (t )}2 and

j (t ) = arctan ( H {c j (t )} c j (t )) .
Differentiating (1.27) with respect to time yields

& (t ) i z

j =1

A j (t ) j (t )e

i j ( t )

e
j =1

i1 ( t )

& (t ), A j

(1.28)

where

j (t ) = d j (t ) / dt is the instantaneous frequency.


& (t ) & (t ) A z = + j (t ), z (t ) A(t )

Combining (1.27) and (1.28) gives (1.29)

& (t ) z (t )] . The same analysis is valid for the j-th (t ) = Im[ z & j (t ) z j (t ) . mode complex signal z
from which it follows that Approximate analysis of multi-component signals For many processes of interest, especially those arising from large perturbations in power systems, a signal is composed of a number of interacting sinusoidal functions. As a first step toward characterizing this behavior, assume that an observed signal, x(t ) , consists of a family of n oscillatory functions, with time-varying amplitudes A j (t ) , and phase the form

j (t ) in

x(t ) =

A (t ) cos( (t ) )
j j j =1

Using the definition of the Hilbert transform, the following relations can be derived

298

A. R. Messina, M. A. Andrade and E. E. Barocio

A(t ) =
as well as 1

j =1

Aj (t ) cos( j (t ) ) +

j =1

Aj (t ) ( sin j (t ) ) ,

(1.30)

& (t ) = (t ) =

& H (t ) x & (t ) * xH (t ) x(t ) * x . 2 A (t ) + 2

(1.31)

In terms of these variables it is possible to write the derivative of the analytic signal as

& (t ) = z
in which

n j =1

i j ( t ) dt i j ( t ) dt o & (t ) , j (t ) + e o A j iAj e
t t

(1.32)

& (t ) & (t ) = A(t ) Re z A z (t ) .


Further,

(1.33)

A( , t ) =

A
j =1

2 j

(t ) +

2 A A cos ( (t ) ) ,
k l kl k l k
t

(1.34)

is the asymmetric signal envelope, and

( A, t ) = Im

j =1

i j ( t ) dt i o j ( t ) dt & (t ) i j (t ) + e o A j Aj e
t

j =1

A e o
j

i j ( t ) dt

(1.35)

is the instantaneous frequency. Drawing upon this formulation, a systematic procedure has been derived for assessing the significance of modal components in system behavior.

.Following the discussion in this section and in the practical application of the method, a damping term, , can be added to reduce the effect of low-amplitude areas of the instantaneous frequency values.

Wide-Area Monitoring and Analysis of Inter-Area Oscillations

299

Quasi-Harmonic Approximation for Modal Analysis


An interesting particular case arises when, for a physical system, the system response can be approximated by a few, slowly varying functions, or modal components [3,9].In this section an approach to approximate the dynamic evolution of an observed data by a low order quasi-harmonic representation is described. This approach follows Feldmans treatment of nonlinear freely vibrating systems [9].

& (t ) 0 in (1.28), and using the above relations one obtains the following Setting A j
frequency-dependent amplitude and frequencies. For two modes (n=2)

A(t ) = A1 (t ) 2 + A2 (t ) 2 + 2 A1 (t ) A2 (t )cos ( 2 (t ) 1 (t ) )
and

(1.36)

(t ) = 1 (t ) +
where

A2 (t ) 2 + K (t ) (2 (t ) 1 (t )) , A2 (t )

(1.37)

K 2 (t ) = A1 (t ) A2 (t ) cos (2 (t ) 1 (t ) )
and

(1.38)

k (t ) =

(t )dt.
T k 0

(1.39)

In contrast to standard theory, equations (1.36) and (1.37) define nonlinear and nonstationary frequency (amplitude) modulated signals. Analogous expressions hold for third order interactions, and form the basis for the study of higher order interactions in later sections. Obtaining the frequency-dependent amplitude is straightforward. Solving for A(t ) as function of

(t ) in (1.37) and (1.38) yields


2 ( A12 (t ) + A2 (t ) (2 (t ) 1 (t )) for A1 A2 , 1 2 . 1 + 2 2

A( , t ) =

(1.40)

For three modes (n=3) Similar to the discussion above, we find that

300

A. R. Messina, M. A. Andrade and E. E. Barocio

A(t ) =

i =1

Aj (t ) 2 +

j =1 k =1

2 Aj (t ) Ak (t ) cos (k (t ) j (2) )

1/ 2

(1.41) (1.42)

2 A121 A2 2 + A32 (3 2 ) + K (t ) (t ) = 1 (t ) + , A2

where

K = A1 (t ) A2 (t ) cos (2 (t ) 1 (t ) ) 2 + A1 (t ) A3 (t ) cos (3 (t ) 1 (t ) ) [1 22 + 3 ] + A2 (t ) A3 (t ) cos (3 (t ) 2 (t ) ) [3 2 ].


Higher-Order Approximations Slightly more complicated expressions can be derived for the case of n modal components and are not included here. In the limit, when n = p , the signal can be fully reconstructed. By taking into account higher order effects, the above analysis provides a sound analytical basis to assess the number of modes that significantly interact. Estimation of Modal Coherency Knowledge of the instantaneous phase difference of oscillations that share the same frequency, can be used to estimate machines pulling apart from each other. In this section, an analytical approach to determine modal coherency is proposed based on the notion on the notion of synchronization in physical systems [10]. Given a data series x[k ] , k = 1, 2,K , n , the phase series is obtained from

[k ] = arctan

H { x[k ]} x[k ]

(1.43)

for k = 1, 2,K , n . It then follows that the angular frequency can be estimated from

[k ] =

[k ] [k 1]
t

(1.44)

where t = t k t k 1 is the sampling period. In practice, one is interested in the relative phase difference between two recordings. These notions are here used to define dynamic coherency between pairs of recordings. The proposed procedure consists of three main steps:

Wide-Area Monitoring and Analysis of Inter-Area Oscillations

301

(i) Decompose each signal of interest into intrinsic modal functions using the EMD technique, (ii) Having decomposed a signal of interest into modal components, express the phase angle of the i-th IMF as

k (t ) = arctan

H {IMFk (t )} IMFk (t )

(1.45)

where H {IMFk (t )} is the Hilbert transform of the k-th IMF. (iii) Relative phase deviations are then obtained as

(t ) for jk (t ) = j (t ) k

j = 1,..., n . Signals with similar phases are then grouped together to form temporal
clusters that identify the nature of the energy exchange. (iv) Using the results of the previous step, determine relative measures indicating the exchange of swing energy. (v) Determine the instantaneous frequency using (1.19) and (1.20). As a by-product, determine the time of arrival of particular frequencies. In our implementation of the method, the arctangent is calculated using the two-quadrant inverse function atan2 in Matlab. This function gives phases that range from to ; the phases reach and fall to once each cycle. The phase sequences are then straightened by adding 2 to the atan2 function at each discontinuity to get the unwrapped analytic phase.

NUMERICAL IMPLEMENTATION
Previously, a modeling framework describing the time evolution of nonlinear and nonstationary data has been proposed [11]. In this section, a variant to this approach that enables approximate analytical solutions to be obtained is proposed.

Empirical Mode Decomposition


In the present formulation, the original time series is approximated by the real part of a truncated analytic function of the form

x(t ) = Re

j =1

Aj (t )e 0

i j ( t ) dt

k = p +1

Ak (t )e 0

i k ( t ) dt

(1.46)

302

A. R. Messina, M. A. Andrade and E. E. Barocio

where p is the unknown number of relevant modal components. The basic EMD method adopted to extract the IMFs essentially consists of three steps [3]: Step 1. Starting with the original signal x(t ) , set ro (t ) = x(t ) , and j = 1 . Step 2. Extract the j-th IMF using the following sifting procedure a) Set ho (t ) = rj (t ) , and i = 1 . b) Identify the successive local minima and the local maxima for ho (t ) . The time spacing between successive maxima is defined to be the time scale of these successive maxima. c) Interpolate the local minima and the local maxima with a cubic spline to form an upper emaxi1 (t ) , and lower emin i1 (t ) envelope for the whole data span. d) Compute

mi 1 (t ) = emaxi1 (t ) + emini1 (t ) / 2 , and subtract it from hi 1 (t ) ; determine a new


estimate hi (t ) = hi 1 (t ) mi 1 (t ) , such that emini1 (t ) hi (t ) emaxi1 (t ) for all t . Set i = i + 1 . The four step procedure described above (2(b) to 2(d)) is then repeated until hi (t ) satisfies a predetermined stopping criterion (described below). Then, set c j (t ) = hi (t ) . Step 3. Obtain an improved residue rj (t ) = rj 1 (t ) c j (t ) . Repeat steps 2(a) through 2(d) with j = j + 1 until the number of extrema in rj (t ) is less than 2. This approach allows elimination of low-amplitude riding waves in the time series and eliminates asymmetries with respect to the local mean, i.e. it makes the wave profile more symmetric. Figure 1.4 illustrates the first step in the application of the EMD technique to a tie-line power signal obtained from a transient stability simulation. The key aspect is that in each step of the procedure, the number of extrema is reduced with the first IMF containing the fastest time scale and each subsequently extracted IMF has a higher time scale than the previous one. At the end of the procedure, the residue becomes either a monotonic function or a constant and the signal is represented by (1). In the implementation of the method, the iterative procedure is stopped when the last ci (t ) or r (t ) become smaller than a predetermined value or when the residue r (t ) becomes a monotonic function from which no further IMFs can be extracted [1]. Then, using the procedure described in Section 1.4, the Hilbert transform is applied to study the effect of specific mode combination on system behavior. This is implemented in three steps: a) Decompose the observed oscillation into a set of n IMFs using the EMD technique b) Apply the Hilbert transform to each IMF, j = 1,.., p , k = p + 1,..., n and compute the amplitude and phase of each component.

the

instantaneous

mean

of

the

envelopes,

Wide-Area Monitoring and Analysis of Inter-Area Oscillations c) Determine the specific effect of each mode combination on system behavior.
800 750 700 Upper envelope

303

Mean value

Power (MW)

650 600 550 500 450 400 350 0 200 400 600 800 1000 1200 Lower envelope

Time point

Figure 1.4. First step of sifting process applied to a test signal. The solid black line represents the original time series. The red and green curves correspond to the upper and lower envelopes.

The above procedures have been implemented in a Matlab-based computer software for the characterization of the temporal spectral characteristics of general nonlinear, nonstationary signals.

APPLICATION TO MEASURED DATA


Due to their complexity, wide-area power oscillations are particularly good candidates for Hilbert analysis. In this section, the developed algorithms are applied to measured data obtained from PMUs of a real event in the Mexican interconnected system (MIS). The nonstationary character of the recorded oscillations provides the thrust to examine the application of the method for on-line estimation of instantaneous attributes.

Description of the Event


At the local time of 06:27:35 on January 1, 2004, undamped oscillations were observed throughout the northern systems of the Mexican interconnected system [12]. The results presented in this work are based on PMU measurements of these oscillations from widely separated areas of the MIS. The main event that originated the oscillations was a failed temporary interconnection of the Northwest regional system to the MIS through a 230-kV line between MZD and DGD substations. Prior to this event, the northwestern system operated as an electrical island.

304

A. R. Messina, M. A. Andrade and E. E. Barocio

Figure 1.5 shows a geographical diagram of the MIS showing the location of the initiating event and the specific location of PMU sites. These devices are part of the existing phasor measurement network in the Mexican interconnected system [13]. The automatic synchronization equipment had a failure and the interconnection was carried out with the systems operating out of phase. This started power and frequency oscillations throughout the northern systems. Following the failed interconnection, no damped inter-area oscillations were observed throughout the northern area, involving severe frequency, voltage and power changes and resulting in the tripping of about 140 MW of load as the result of the operation of protective relays. The condition prevailed for some minutes until the line was disconnected. As a result of topological changes and load shedding, the observed oscillations exhibit highly complex phenomena including modal interaction, and transient motions characterized by changing frequency content. The following is a summary of relevant instants of system behavior in the context of this study [12]. At 06:27:52, large amplitude oscillations are observed at various locations in the system. At 6:28:06 a local generator in the northwestern system trips out resulting in the operation of load shedding schemes and under frequency relays At 06:28:54, the system stabilizes at a frequency of about 60.05 Hz to decrease then slowly to about 59.959 Hz. At 06:29:39, the northwestern system disconnects from the MIS. The frequency at Hermosillo and Mazatlan Dos substations in the northwestern system drops to about 58.5 Hz and then slowly recovers to the nominal value.

Phasor Measurement Unit (PMU)

Phasor Measurement Unit (PMU)

Phasor Measurement Unit (PMU)

Phasor Measurement Unit (PMU)

Phasor Data Concentrator

Figure 1.5. Schematic of the Mexican system showing the location of PMUs and the disturbance area.

Wide-Area Monitoring and Analysis of Inter-Area Oscillations

305

During the time interval 06:27:52 06:28:54 the system experienced severe fluctuations in frequency, power, and voltage resulting in the operation of protection equipment with the subsequent disconnection of load, independent generation and major transmission resources. Among the existing PMU locations, we selected frequency and voltage measurements at four major substations round the north and northeastern systems: Hermosillo (HLT) in the extreme of the northwestern system, Mazatlan Dos in the interconnection point between the northwestern and north regional systems, and Tres Estrellas (TTE) in the on the most extreme part of the north-south corridor. Measurements from PMUs were recorded and later analyzed for evaluation of system dynamic behavior. Figure 1.5 shows the location the initiating events and disturbance area indicating the location of selected PMU sites. The stations are deployed on the backbone of the 400/230 kV system of the north systems which allows analysis of inter-area oscillations and mode propagation. The first set of response data is collected from bus frequencies at strategic system substations; the measurements examined extend from 06:27:35 local time to 06:28:55. Measurements were recorded over 400 s collected at a rate of 0.20 samples per second for a total of 2002 samples. The recordings have been extensively checked for trends, noise, and other relevant information using various techniques. Figure 1.6 shows a plot of the corresponding time-evolution of the recorded bus voltage frequencies at selected system locations. Extracting from these recordings temporal behavior is a challenging problem that requires the use of advanced signal processing techniques. Of particular relevance to this study is the detection and characterization of temporal changes in the spectral content of the record. Visual inspection of these figures provides little insight into system dynamic behavior. Further, the non-stationary character of these oscillations makes spectral analysis a difficult task. The following sections describe the application of non-stationary techniques to the measured data, including the estimation of instantaneous parameters, the accuracy of modal estimates that can be achieved, and the study of mode propagation. The accuracy of the method is demonstrated by comparison to wavelet and Fourier analysis.
60.8 60.6 60.4

06:28:06 06:28:54 06:29:59

Frequency (Hz)

60.2 60 59.8 59.6 59.4 59.2 59 0 50 100 150 200 250 300 350 400

06:27:42

Time (s)

a. Hermosillo (HLI). Figure 1.6. (Continued on next page).

306

A. R. Messina, M. A. Andrade and E. E. Barocio

b. Mazatlan Dos (MZD).

c. Tres Estrellas (TTE). Figure 1.6. Time traces of recorded bus frequencies.

Wavelet Analysis of the Recorded Oscillations


Based on the frequency data collected by PMUs, wavelet analysis and the EMD method were used to determine non-stationary characteristics. For illustrative purposes, the frequency signal at the Mazatlan Dos (MZD) substation was chosen, since this substation is located near the electrical center of the northwestern and north systems. As discussed in our numerical results, this signal exhibits a complex temporal dynamics including inter-modulation and frequency generation. Figure 1.7 shows the wavelet spectrum for the Mazatlan Dos signal together with the original signal. The original signal in Fig. 1.7a is plotted against the nonlinear trend of the signal, to identify the long-term behavior. For comparison, Fig. 1.7 also shows the

Wide-Area Monitoring and Analysis of Inter-Area Oscillations

307

continuous-time wavelet spectrum of the recording. Note the strong trend in the dynamic record. In order to have meaningful results for wavelet analysis, the mean value and the standard deviation were calculated and removed from the data. Further, because of the strong underlying trend, a high-order (cubic) model was needed to remove any nonlinear trend in the data. These observations imply significant nonlinearity in the measured oscillations and bring into question the use of conventional analysis techniques for practical on-line assessment of temporal characteristics. The wavelet spectrum displays a number of interesting features including frequency variations and harmonic generation. Three distinct periods of activity can be clearly identified: Period 1 (06:27:52-06:27:49). An initial transient period in which a high frequency mode at about 0.22 Hz is seen to dominate the system response. Presumably, this mode corresponds to the north-south inter-area mode found in the analytical model.
60.5 60.5

60 Frequency (Hz)

60 Frequency (Hz)

59.5

59.5

59

59

58.5 0 100 200 Time (s) 300 400

58.5

Figure 1.7. (Continued on next page).

308

A. R. Messina, M. A. Andrade and E. E. Barocio

450 350

0.61961

45 35

250
0.21576

25

150

0.039081

15

50 0 0.5 1

1.2254

5 1.5 2 2.5

Frequency (Hz)
Figure 1.7. The wavelet spectrum of the frequency oscillation measured at MZD substation.

Period 2 (06:27:52-06:27:49). A mid-term period dominated by a low frequency mode at about 0.62 Hz. This episode is characterized by harmonic and sub harmonic generation. Of particular interest is the presence of higher frequency components (1.22 Hz) most likely a second harmonic of the 0.61 Hz mode. Another interesting feature of the plot is the regime of high-frequency behavior near sub harmonic modes at about 0.40 Hz and 0.125 Hz). Period 3 (06:27:52-06:27:49). A long-term period in which the frequency reduces to the original value (0.22 Hz). These time windows coincide with major system events described in this section.

Real-Time Adaptive Filtering by the EMD


To visualize the complex spatial and temporal dynamics that takes place in the system following the failed interconnection, the results in [12] were extended to provide on-line analysis of the Hilbert transform. As discussed in the theoretical formulation in section 1.1.1, the local EMD allows the adaptive decomposition of nonlinear and non-stationary signals, making it possible to separate high-frequency noise from components containing useful, physical information. In this analysis, the transient response is expressed as a sum of nearly sinusoidal oscillations (the empirically derived basis functions). Using the EMD method, the MZD frequency signal was decomposed into nine IMFs and a trend. These IMFs were then used to extract the time-varying amplitude and phase of the intrinsic oscillations. Figure 1.8 shows the IMFs decomposed from the frequency signal in Figure 1.6b. As emphasized in the previous results, each component represents a different oscillation mode with different amplitude and frequency content. Further, because each IMF is uncorrelated to each other, the dominant frequency of each component can be estimated. This information is used below to estimate the mode propagation along the system and to determine modal coherency.

Integrated Power

Time (s)

Wide-Area Monitoring and Analysis of Inter-Area Oscillations

309

In order to assist in the interpretation of results, and to aid in the physical interpretation of IMF as they relate to the underlying process giving rise to the oscillations, the original signal was reconstructed using the various components. Figure 1.9 provides a comparison of the original series with the signal reconstructed using various IMFs illustrating the dominant frequencies determined using conventional Fourier and Prony analysis. As shown in this figure, the accuracy of the reconstructed signal improves gradually by increasing the number of IMF components.
0.5 0 -0.5 0.2 IMF2 0 -0.2 0.2 IMF3 0 -0.2 IMF4 0.1 0 -0.1 0.1 0 -0.1 0.2 0 -0.2 0.2 IMF7 0 -0.2 0.1 0 -0.1 0.05 IMF9 0 -0.05 59.95 59.9 59.85 59.8 59.75 50 100 150 200 250 300 350 400 IMF10 50 100 150 200 250 300 350 400 IMF8 50 100 150 200 250 300 350 400 50 100 150 200 250 300 350 400 50 100 150 200 250 300 350 400 50 100 150 200 250 300 350 400 50 100 150 200 250 300 350 400 50 100 150 200 250 300 350 400

IMF5

IMF1

50

100

150

200

250

300

350

400

IMF6

50

100

150

200 Time (s)

250

300

350

400

Figure 1.8. The 10 IMFs components obtained from the MZD frequency oscillation trough the EMD method.

310

A. R. Messina, M. A. Andrade and E. E. Barocio

Figure 1.9. Comparison of the original signal with the signal reconstructed using dominant IMFs. Dotted square boxes indicate dominant frequencies.

The results clearly show that the obtained IMFs are capable of capturing the timefrequency evolution of the observed oscillations. A prominent feature of the observed dynamics is the presence of nonlinear, time-evolving frequency composition. It is apparent from these simulations, that most of the temporal characteristics are captured by the first two modes. The first IMF is seen to capture the dynamic behavior of the mid part of the study region (time interval 06:27:52-06:27:49), while the second IMF appear related to the initial oscillation period (06:27:52-06:27:49). The method successfully extracts the temporal information and the instantaneous frequency and phase, even when several oscillatory modes are present. In addition, the timestamp of the arrival of every mode can be used to detect boundaries of oscillatory events and abrupt changes in temporal behavior on a real-time basis. From this representation, instantaneous phase and frequency information was obtained using the local Hilbert transform method. Figure 1.10 shows the Hilbert amplitude spectrum, whereas Figure 1.11 shows the time-dependent amplitude and frequency of the first four modes of concern for periods 1 and 2 above, calculated without using any smooth function. Close examination of the time-dependent frequency shows the presence of two dominant modes. The first mode is seen to exhibit a mean frequency of 0.62 Hz in close agreement with wavelet results in Figure 1.7. The second mode at about 0.27 Hz exhibits nearly stationary characteristics. These changes in frequency content are consistent with control actions taking place in the system.

Wide-Area Monitoring and Analysis of Inter-Area Oscillations


2.5 1

311

0.9

0.8

0.7

Frequency (Hz)

0.5

0.4

0.3

0.5

0.2

0.1

50

100

150

200 Times (s)

250

300

350

400

Figure 1.10. The Hilbert amplitude spectrum for the MZD frequency oscillation and its projection onto the amplitude-time plane.

1 0.8

06:28:06 06:27:42

Amplitude (Hz)

1.5

0.6

Amplitude

0.6 0.4 0.2 0 120 2.5

IMF 1 IMF 2 IMF 3 IMF 4

125

130

135

140

145

150

155

160

165

170

Frequency (Hz)

2 1.5 1 0.5 0 120 125 130 135 140 145 150 155 160 165 170

0.61 Hz 0.22 Hz

Time (s)

Figure 1.11. Instantaneous amplitude and frequency of the dominant modes. Prominent variations in system frequency are indicated.

312

A. R. Messina, M. A. Andrade and E. E. Barocio

In interpreting these results we point out that, unlike wavelet analysis, nonlinearity is captured as inter-modulation. While not discussed in this work, analysis of the temporal evolution of the frequency signals shows to interacting frequencies; a dominant frequency at about 0.62 Hz and a second frequency at about 0.50 Hz. The physical implication of this behavior is currently being investigated. It is of interest to note that straightforward application of the Hilbert technique introduces large-amplitude, high frequency spikes when the modal amplitude is small (refer to 1.19). A better estimate of the instantaneous frequency can be obtained by modifying this basic approach as discussed in our footnote on page 14 or using other specialized techniques This is, however, not applied in the present work and forms the basis of ongoing research. Examination of the time evolution of these modes in Fig. 1.11 reveals the time of arrival of the dominant modes and enables the identification of abrupt changes in system behavior. The timing of these variations can be accurately singled out by non-stationary analysis techniques. With this basis, we next use quasi-harmonic approximations to approximate the spatiotemporal dynamics.

Low-Order System Approximations


Previous analytical work has concentrated on the evaluation of the full EMD decomposition to analyze system behavior. This evidence suggests that transient oscillations may involve a limited number of interacting modes and that this motion can be described by quasi-periodic oscillation. The studies reported below examine the application of reducedorder system representations to approximate system behavior with the objectives of identifying their applicability in avoiding the use of multi-component analysis.

b. Instantaneous frequency.
Figure 1.12. Comparison of the instantaneous parameters of the original signal with

Figure 1.12 compares the time evolution of the dominant IMF with that computed using Eqs. (1.37) and (1.42). Similar results can be found for the amplitude of the signals and are

Wide-Area Monitoring and Analysis of Inter-Area Oscillations

313

not included here. The key point of importance here is that by retaining only the first two modes in the Hilbert representation, the model is able to accurately replicate both amplitude and frequency behavior. This also demonstrates the power of the empirical mode decomposition as a filter.

Estimation of Dynamic Modal Coherency


Further insight into the temporal behavior of the observed oscillations can be found by examining the phase difference of oscillations sharing the same frequency. Another related issue is how these modes propagate within the system. This latter aspect is particularly important when designing the optimum location of measurement devices and the design of wide-area monitoring and control systems. To analyze the extent to which an oscillation is synchronous at different locations within the system the individual decompositions can be analyzed and compared. Knowledge of the temporal evolution of the phases aids in identifying machines behaving coherently which is important for the understanding of complex motion and proper control action. This approach can be used to detect rapid changes in phase that may be overlooked by other analytical techniques. Building on the concept of a local analytic signal outlined in Section 1.1.2, a procedure is developed to determine efficiently dynamic modal coherency technique in time-domain. In this approach, the phase difference between multiple oscillations, which share the same frequency, was estimated by following the procedure in Section 1.6.3. Figure 1.13 shows the phase difference of the 0.22 Hz modal components of the Hermosillo, Mazatlan Dos and Tres Estrellas extracted from the EMD technique. For the numerical simulations in the present study, the phase of the frequency at the 230 kV Hermosillo substation is taken as a reference but this selection has no bearing on the presented results. This gives a measure of instantaneous modal coherency.

a. Absolute angles. Figure 1.13. (Continued on next page).

314

A. R. Messina, M. A. Andrade and E. E. Barocio

b. Relative angles. Figure 1.13. Instantaneous phase difference for IMF1.

Examination of the results in Figure 1.13 reveals that the proposed algorithm is able to identify precisely the exchange of energy taking place in the system. Prior to the major initiating event, (06:25:36-06:27:42), the analysis of the instantaneous phases in Figure 1.13a shows that the frequencies at Hermosillo and Mazatlan Dos substations oscillate coherently maintaining a relatively constant phase difference. This is further confirmed in Figure 1.13b that shows a constant value of the relative phase angle. By contrast, the analysis of the phase of the Tres Estrellas frequency signal with respect to the Hermosillo signal shows a linearly increasing characteristic. This is consistent with theoretical expectations. By monitoring the time evolution of phase angles and system damping in near real-time, the present method can be used to trigger remedial control actions and to aid in the monitoring of specific aspects of the temporal evolution of complex oscillations. Such analysis may be useful, for example, to study the characteristics of the inter-area mode separation phenomenon. These results have particular relevance for power system dynamic analyses, where vast volumes of data require processing, and it is critical that machines losing stability can be detected on a real-time basis.

CONCLUSION
Transient processes in power systems are inherently non-stationary. In this chapter, nonlinear time-series analysis techniques based on the combined application of the empirical mode decomposition and Hilbert analysis have been employed to characterize the dynamic behavior of measured data. The results provide a number of insights into the nature of temporal spectral variations, suggesting the technique is effective for detecting both, oscillatory and abrupt power system transients. Two new approaches to dynamic behavior assessment, one of temporal behavior, and the other of temporal synchrony have been compared with existing formulations. Non-stationary

Wide-Area Monitoring and Analysis of Inter-Area Oscillations

315

signal analysis methods applied to measured and simulated data reveal a wealth of interactions occurring between major systems modes. The evidence provided by Hilbert and wavelet analysis of simulated and measured data indicates that the interactions result in significant modulation of the amplitude and frequency of the nonlinear components. These interactions play an important role in the dynamics of the system and certainly contribute significantly to the variability of the observed oscillations at the frequencies of concern. Efforts have also been directed towards the incorporation of new features to existing approaches. These include the analysis of synchronicity among multiple nonlinear, nonstationary records and the analysis of quasi-harmonic approximations to system dynamic behavior. Several lines of investigation open from these studies. First, further examination is needed of the effect of noise in system temporal behavior. Further study is also required to understand and characterize the temporal and geographical variation of such phenomena (mode propagation), especially in large interconnected power systems. These results are an initial step in this direction. Finally, studies are underway to help in the identification of suitable locations to install new PMUs when damping of wide-area power oscillations is a major concern.

REFERENCES
[1] Huang, N. E. , Sheng, Z., Long, S. R. , Wu, M. C. , Shih, H. H., Zheng, Q. , Yen, N. C., Tung, C. C. , and Liu, H. H., The empirical mode decomposition and the Hilbert spectrum for nonlinear and non-stationary time series analysis, Proceedings of the Royal Society of London. A, 1998, 454, 903-995. Hahn, Stefan L. Hilbert transforms, The Transforms and Applications Handbook, A. D. Poularikas, Ed. Boca Raton, CRC Press, 1996, pp. 463-629. Messina, A. R., Vittal Vijay, Nonlinear, non-stationary analysis of inter-area oscillations via Hilbert spectral analysis, IEEE Trans. on Power Systems, 2006, 21 (3), 1234-1241. Gabor, D.; Theory of communication, in J. of the Inst. E. E, 1946, 93, 429-457. Marple, S.L., "Computing the discrete-time analytic signal via FFT," IEEE Trans. on Signal Processing, 1999, 47 (9), 2600-2603. McClellan, J. H. , Parks, T. W., Rabiner, L. R., A computer program for designing optimum FIR linear phase digital filters, IEEE Trans. on Audio and Electroacoustics, 1973, 21, 506-526. Parks, T. W., McClellan, J. H., Chebyshev approximation for nonrecursive digital filters with linear phase, IEEE Trans. on Circuit Theory, 1972, 19, 189-194. L. R. Rabiner, J. H. McClellan, and T.W. Parks, FIR digital filter design techniques using weighted Chebyshev approximations, in Proceedings of the IEEE, vol. 63, 1975. Feldman, M., Non-linear free vibration identification via the Hilbert transform, Journal of Sound and Vibration, 208 (3), 1997, 475-489.

[2] [3]

[4] [5] [6]

[7] [8] [9]

316

A. R. Messina, M. A. Andrade and E. E. Barocio

[10] Hurtado, J.M. Rubchinsky, L.L., Sigvardt, K. A., Wheelock, V. L., Pappas, C. T. E, Temporal evolutions of oscillations and synchrony in GPi/muscle pairs in Parkinsons disease, J. Neurophysiology, 2005, 93, 1569-1584. [11] Andrade, M. A., Messina, A. R., Olguin, D., Rivera S., C. A. ,Identification of instantaneous attributes of torsional shaft signals using the Hilbert transform , IEEE Trans. on Power Systems,2004 , 19 (3),1422-1429. [12] Messina, A.R., Vittal, V., Ruiz-Vega, D., Enrquez Harper, G., Interpretation and visualization of wide-area PMU measurements using Hilbert analysis, IEEE Trans. on Power Systems, 2006, 21 (3), 1763-1771. [13] Silva Peruyero, M. A., Melendez Roman, C. G., Phasor measurement unit (PMU) applications in the transmission network of CFE, Cigre Conference 2006, Paris France.

In: Leading-Edge Electric Power Research Editor: C. M. OSullivan, pp. 317-330

ISBN: 978-1-60021-981-8 2008 Nova Science Publishers, Inc.

Chapter 8

UNCONVENTIONAL PROBLEMS IN POWER SYSTEMS PROTECTION


Mahmoud Gilany and Mohamed A. Mahmoud
College of Technological Studies, Kuwait
Of the numerous electric power faults an Electric Engineer comes across in a life time, only a few of these faults are memorable- the rest being routine ones. In this chapter, some of those unconventional faults, which are mainly related to power system protection, are presented. The chapter presents five case studies of actual field incidents rather than hypothetical scenarios. The objective of the chapter is to present a typical approach for analyzing the faults in power systems.

1. CASE-STUDY NO. 1
In this case, a fault occurred on one of the outgoing feeders tapped from 11 kV distribution board, DB1 as shown in Figure 1. The Board is supplied from a nearby 66/11 kV substation which is supplying other industrial loads. Well-coordinated overcurrent protective relays, RA, RB and RC were installed as shown in Figure1. The exact fault-scenario was as follows: a) A fault occurred on feeder FD-X. b) The protective relay RA tripped after 0.4 second and disconnected the faulty feeder FD-X. c) All other relays (RB, RC, etc) were stable and did not respond to the fault as expected. d) However, what was not expected is that shortly after clearing the fault, the main overcurrent relay, denoted RC in Figure 1 that is protecting the 66/11 kV transformer tripped without any other fault recorded in the network.

318

Mahmoud Gilany and Mohamed A. Mahmoud

66 kV
1.1 Sec RC

DB2 RB 0.7 Sec

11 kV
RD RE

DB1

FD1 RA 0.4 Sec

FD2

Industrial Loads

FD-X Industrial Loads Industrial Loads


Figure 1. The network considered in Case-1.

Generally, six questions need to be answered before starting to analyze a fault: 1. 2. 3. 4. 5. 6. What is the faulty element? What is the type of fault? Where is the fault located? What is the protection system being used? What are the settings of the protective relays involved in the fault? What kind of loads are supplied from this part of network?

In many cases, answers to only some of these questions are enough to analyze the fault. In this case, a short circuit fault occurred on a power cable (FD-X). The fault location was close to distribution boards DB1 and DB2. The settings of the protective overcurrent relays used in the network are shown in Figure 1. So far, there is no useful information to explain the subsequent apparently false tripping of relay RC. However, in response to question No. 6, there are industrial loads supplied from the outgoing feeders. Let's have a detailed look at the effect of these industrial loads. The loads are mainly large electrical motors. These motors usually cause troubles with the protective relays during the starting periods. This is because the motor current at starting is a function of the motor speed as explained in Equation-1 derived from the simplified motor's equivalent circuit shown in Figure 2.

IM =

Vs Emf Vs kn ... = Xm Xm

(1)

Unconventional Problems in Power Systems Protection

319

where IM is the motor current, Vs is the supply voltage, n is the motor speed, is the flux, k is a constant and XM is the motor reactance.

Figure 2. Simplified motor's equivalent circuit.

At the instant of starting, the speed n is zero, and hence the motor current is high. After a few seconds, the motor reaches its normal speed and the motor's high-starting-current drops to the motors normal rated value. It seems so far that this discussion is not related to our problem since the apparently false trip of relay RC occurred while the motors were already running and consequently there were no starting currents. In fact, there were starting currents, but the reason for them needs explanation. After the inception of the original fault, and as the fault location is near to the busbar DB2, there was a large reduction in the busbar voltage. This reduction causes the motors supplied from the healthy feeders FD1 and FD2 to run into a braking mode. The speed of the motors dropped to zero but the motors didn't disconnect totally. The motors under such conditions draw higher current than normal levels. Once the original fault at feeder FD-X is cleared, the voltage on DB2 returned to its normal value and the motors connected to the healthy feeders were released. The motors started to withdraw high starting current as if they were just starting. The starting currents in this case occurred simultaneously in all the motors connected to the bus bar of DB2. Consequently, it withdrew severely high starting current and as a result, the instantaneous overcurrent relay RC tripped causing the main transformer to trip. The following questions may be raised: Q1. Why did Relays RD , RE and RB not trip like RC ? Although there were high starting currents in the motors supplied from the feeders protected by these relays, they didnt trip as only small part of the total starting current is supplied through each of them. Relay RC is the only relay supplying all motors, hence it tripped. Q2. What about normal conditions? Will relay RC trip under such conditions? The probability of having all motors starting simultaneously is very small, and hence such condition does not occur under normal operation conditions. It is worth mentioning that what happened in the present case-study is similar to a case known in the literature as Cooled Inrush [1].

320

Mahmoud Gilany and Mohamed A. Mahmoud

Cooled Inrush is a particular condition for distribution circuits that serve residential and commercial loads. It is characterized with high starting current that occurs when a feeder is energized after a prolonged outage. For this cold load condition, the diversity of intermittent loads is lost. This is because consumers tend to leave more than the normal load connected, and thermostatically controlled equipment will start simultaneously as soon as the voltage is restored. The overall effect is that a very high initial current (cold-load-inrush) is drawn on the restoration of the power. The typical fixed-setting of the relay may cause it to trip on the cold-load-inrush. The settings of the relays that protect the power distribution network may need to be changed to suit the different power system conditions. The concept of adaptive digital relay [2-5] provides an optimum solution for this problem by using two groups of settings. A simple adaptive solution for cold-inrush problem is presented in reference [5]. In this proposed algorithm, the relay examines first if the feeder is off. If the feeder is not off, or if it has not been just restored, then the relay algorithm follows the normal protection routine. On the other hand, if yes (i.e., has been just restored) then there is a probability that a cold inrush may occur just after restoring the power. The relay then starts with a curve that has a high TMS (Time Multiplier Setting) among those of the relay characteristic's curves shown in Figure 3. For example, it may start with the curve having a TMS = 0.8 for the first 3 - 4 seconds and then switches back to the original curve on which the coordination is based.

Figure 3. Typical Inverse overcurrent relay characteristic.

Unconventional Problems in Power Systems Protection

321

2. CASE-STUDY NO. 2
In this case, a single-line-to-ground fault occurred on one of the four parallel underground cables (feeder D2) supplied from a 132/66 kV station as shown in Figure 4. The fault was properly cleared by the corresponding protection of that feeder. However, instead of only tripping the faulty feeder, D2, all other feeders (D1, D3, and D4) were also tripped simultaneously without any recorded fault on any of them!

Figure 4. The circuit considered for Case-2.

In this system, each cable is protected by an overcurrent/earth fault relay at the supply side. It is also protected by directional overcurrent relay at the other side (load side). The earth-fault-relay is connected to a core-type current-transformer. In such failure cases, cable charging currents are usually the main suspect. Cables are characterized by very high shunt capacitance values, which lead to high charging currents compared with those of overhead transmission lines. This is due to cable construction and nature of materials used for insulation and shielding. The high capacitance of the cables results in a comparatively high charging current that is affected by many parameters. It increases with the increase of cable diameter, length and operating voltage [6]. The charging current IC, is calculated from the well known equation:

IC = V 2 C L L

(2)

where V is the phase voltage, C is the cable capacitance per unit length, L is the cable length, and is the radian frequency. Calculations of protective relay settings for power-cable networks should take into account the phenomenon of charging currents that may lead to false tripping of relays. To explain the reasons for the false trips that occurred in the healthy feeders, the current distribution in all feeders is presented in Figure 5 where the values of these currents are presented in the phasor diagram shown in Figure 6.

322

Mahmoud Gilany and Mohamed A. Mahmoud

Transformerdelta-side

C B A

Healthy Phase

Healthy Phase

Healthy Phase

The Faulty Feeder Fault current


Figure 5. Distribution of the fault currents in isolated system.

A Healthy Feeder

Figure 6. Phasor diagram for the voltage and current signals.

Healthy Phase

Unconventional Problems in Power Systems Protection

323

The feeders are located on the delta-side of the transformer. It is well known that for a single-line-to-ground fault in isolated-systems, the only path for ground currents to complete the circuit is to flow through the distributed line-to-ground capacitances of the healthy phases. The capacitive currents pass through the healthy phases of both the faulty and healthy circuits as shown in Figure 5. It is possible in isolated systems to continue operation with the existence of a ground fault. However, the system must have a phase-to-earth insulation level as that of a phase-tophase insulation level. Figure 5 shows that the earth-fault current in the feeders depends on the value of the capacitive currents through the sound feeders (ICA and ICB in Figure 6) which in turn depend on the feeders length according to equation- 2. Under normal operating conditions, the capacitive current in each feeder sums up to zero as the vector sum of any three phase symmetrical currents is zero. Under fault conditions, say on phase-A, the capacitive current in phase-A (IA) vanishes; and the other healthy phasecurrents are increased by 3 factor. Their sum will have a value equal to three times the phase current in healthy conditions. From Figure 6, it can be seen that the fault currents have the following features: 1. The earth fault current through the faulty feeder has the largest value because it is the sum of all capacitive currents in the other adjacent feeders. 2. A high capacitive current in a healthy feeder appears when the phase currents are added together. 3. The fault current of the faulty feeder is in opposite direction to the currents in the healthy feeders. Regarding the studied case, a core type current transformer is used for earth fault protection. The phase-currents in the healthy feeders are added vectorially resulting in high current in these feeders. Another factor that causes the healthy feeders to trip is the use of undirectional overcurrent relays at the supply side. One possible way to detect similar faults is to raise the setting of all earth fault relays, which may not be acceptable in many cases as it reduces the sensitivity of the protection system. The optimum solution for such a condition is based on the third feature mentioned above for the fault current. The fault current of the faulty feeder is in the opposite direction of the currents in the healthy feeders. This feature can be used to prevent the false trip especially if digital relays are used.

3. CASE-STUDY NO. 3
This case was recorded in a 66/11 kV transformation station. Three transformers were connected in parallel as shown in Figure 7. One of the three transformers, Tr-1, was disconnected for maintenance purposes. The disconnected transformer was put back in service successfully. However, a few seconds after restoring the transformer, the three transformers were all tripped although there were no faults inside any transformer or on the external circuit.

324

Mahmoud Gilany and Mohamed A. Mahmoud

Figure 7. The circuit considered in Case-3.

Before searching for an explanation for such false trips, it is useful to review the typical mechanism for restoring a transformer. The proper circuit-closing procedure is illustrated graphically in Figure 8 and summarized below: 1. Close the isolator switches of the high-voltage-side (denoted as No.-1 and No.-2 in Figure 8). 2. Close the low voltage-side isolator (No-3). 3. Close the high-voltage-side circuit breaker (No- 4). 4. Close the low-voltage-side circuit breaker (No.-5). The opening process is opposite to the closing process, i.e, open the circuit breaker No. 5, then breaker No. 4, then open the isolators No.-3, No.-2, and No.-1 respectively. It is worth mentioning that an isolator must be opened or closed only under no-load conditions, otherwise it will be damaged. If any one of the two circuit breakers (after certain delay time) failed to close completely, then the control system will open the isolator in the order mentioned above in order to be ready to re-try the closing process. In the case under consideration, the two breakers at both sides of the transformer were successfully closed. However, as a result of a fault in the control system, it did not recognize this fact. As a result, it started the opening operation of the isolators by sending a signal to the motor of isolator No.-3. Note that the breakers No.-4 and No.-5 were incorrectly identified by the control system to be in open status.

66 kV

11 kV

1 4

2 5

Figure 8. A transformer circuit-closing procedure.

Unconventional Problems in Power Systems Protection

325

The problem now is that the isolator No.-3 is forced to open under load-condition, which is against its specifications. Consequently, a severe arc occurred inside the isolator, which is a fault. This fault was fed from the three parallel transformers as shown in Figure 9, and consequently they were all tripped.

Figure 9. Fault-current path.

The importance of such a fault is that it reminds us to pay close attention to the relationship between control and protection systems. Such a fault would be avoided if the circuit supplying the motor of isolator No.-3 was provided with a normally-closed auxiliarycontact from current-transformer of the protection system as shown in Figure 10. With this auxiliary contact, the motor of Iso-3 will never operate as long as there is a current in the transformer.

Figure 10. A simple circuit to prevent isolator wrong operation.

4. CASE-STUDY NO. 4
Another case related to the problems of control circuits is presented below. In this case, a fault occurred inside one of three parallel transformers (namely Tr-1) as shown in Figure 11. The fault was successfully detected and the faulty transformer was tripped. However, right after tripping the transformer, the two incoming feeders supplying the transformer station, FD1 and FD2 as well as the other healthy transformers, Tr-2 and Tr-3, were all tripped without any recorded fault on either the transformers or the feeders.

326

Mahmoud Gilany and Mohamed A. Mahmoud Such complete trip may correctly occur only in one of two possible cases: 1. In case of a bus bar fault, or 2. In case of a breaker-failure

In the present situation, there was neither a bus bar fault nor a clear breaker-failure!. To explain such a fault, the mechanism for breaker-failure system needs to be reviewed. The mechanism is shown schematically in Figure 12.

Figure 11.The circuit considered in Case-4.

Figure 12. Breaker failure mechanism.

Unconventional Problems in Power Systems Protection

327

After any fault is detected, the relay R usually sends a trip signal to the circuit breaker CB and another signal to trigger the breaker-failure circuit, BF. When the circuit breaker is successfully opened, the timer T within the breaker-failure circuit is reset by a signal from an auxiliary contact from the current transformer CT. If the timer, for any reason, is not reset, all the breakers connected to the bus bar must be tripped, which represents a breaker-failure case. In the recorded case, the circuit breaker CB opened successfully, but the timer was not reset as a result of an internal fault in the timer itself. Consequently, the breaker-failure system was falsely invoked and all breakers were incorrectly tripped. Installing an auxiliary contact from the circuit breaker or from the current transformer in series with the timers contact would prevent such a problem. Again, the importance of such a fault is that it reminds us to pay more attention to the relation between control and protection systems.

5. CASE-STUDY NO. 5
Most of the faults on transmission lines are shunt faults, for example (SLG, DLG, etc). These faults are easy to be detected as they are associated with large increase in the current. Series faults (resulting from an open-conductor) result in unbalanced fault conditions without increase in the current. The case of open-conductor takes place generally in about 5% of fault cases. The percentage is increased to about 10% of the faults in areas near the sea as a result of higher corrosion rates. The extent of power outage resulting from an open circuit condition may be larger than that resulting from a short circuit fault. It seems that this fact is recognized by the "highvoltage-thieves" who want to steal the high voltage transmission wires. Many attempts to steal high-voltage wires are recorded in third world countries. Some attempts were made by hanging a thick wire on the overhead line such that a permanent short circuit fault is detected by the protection system at both sides of the line as shown in Figure 13. The line is consequently opened from both sides. It would be now safe for the thieves to climb the tower and cut the wires.

Figure 13. Making a short circuit fault.

thievies also succeeded in making open-circuit faults to do the same job. It seems that they discovered that this is safer since an open-circuit fault may result in disconnecting the

328

Mahmoud Gilany and Mohamed A. Mahmoud

two parallel circuits on the same transmission tower. In case of making short-circuit, only the faulty circuit is tripped while the other healthy circuit on the same tower continues carrying power. But, can an open-circuit fault be made at will? In many cases, the overhead transmission lines are required to pass a road or a lake. This requires a special type of towers called tension towers. These tension-towers are characterized by horizontal insulator strings as shown in Figure 14. An open conductor typically occurs when the bridle (part of the line shorting the insulation string shown in Figure 14) is cut. Shooting this bridle with a gun will result in an open-circuit. Of course, aging and wind may also lead to the same effect. [7]. Detecting such an open-circuit fault as well as estimating the impedance to the open circuited point is not an easy task as the fault current after opening the circuit is near zero. Only a very small current through the stray capacitors may be noticed. Traditional distance relays are not capable of detecting an open-circuit condition. Usually, the distance relay at the open-side does not respond to this fault since the measured impedance is beyond its zonelimits [7, 8]. To explain why the power outage resulting from an open-circuit fault may be wider than that resulting from a short circuit fault, consider a failure case recorded in a 220kV shown in Figure 15. An open circuit fault occurred on one of two parallel circuits as shown. The fault was not detected ,as expected, by the distance relays at both sides of the line. Instead, it was detected by the earth fault relays, EFR-1 and EFR-2; as the open-circuit fault represents an unbalance condition. The unbalance level at relay EFR-2 (at 220 kV side) is much higher than the unbalance level at relay EFR-1 (at 66 kV side). Consequently, the earth fault relay EFR-2 tripped faster than the earth fault relay EFR-1. Therefore, the major 220 kV transformation station was totally disconnected resulting in a major shut down in that area. In this case, the two circuits on the TL were disconnected making it safe for thievies to streal the wire.

Figure 14. The bridle in tension tower.

Unconventional Problems in Power Systems Protection

329

Some researches introduced techniques for detecting cases of open-conductors and high impedance faults by monitoring non nominal frequency parameters. These techniques lack the selectivity feature [9, 10]. Other techniques used for distribution feeders depend on locating a number of voltage sensors along the feeder to sense unbalance voltages that result from a case of open conductor. However, this is not suitable for transmission lines [11]. There are two other methods widely-used to detect an open-circuit condition. In the first method, the open-circuit fault is detected by the earth fault relays since the open-circuit fault represents a type of unbalance condition. However, the resulting unbalance depends on the value of the pre-fault current [12]. With heavy pre-fault load, the probability of detecting such an unbalance is low. Also, the earth fault relay is used as a back-up relay, and consequently it takes longer time to operate. To avoid relying on the pre-fault load, the second method is based on a ratio between the negative sequence current I2 and the positive sequence current I1. This ratio, (I2/I1) is used as a relaying quantity. This ratio eliminates the effect of pre-fault load. However, the ratio (I2/I1) is a function of the location of the open-circuit. The setting value of this function of the relay is obtained by a trial and error routine and hence is not accurate. This may lead to mal-operation with any other unbalanced condition. Again, there is also a long delay time. Open circuit faults present a major challenge to electric power engineers and needs further studies to improve the reliability of the protection systems used.

Higher Unbalance EFR-2

2. 5 Sec

220 kV

66 kV EFR-1 Open Circuit 1. 5 Sec

Figure 15. Network considered in Case-5.

330

Mahmoud Gilany and Mohamed A. Mahmoud

REFERENCES
Westinghouse Electric Corporation, Applied Protective Relaying, Relay Instruments Division, Coral Springs, Florida, 1982, PP. 10-7:10-8. [2] M.S. Sachdev, J.Singh and R.J. Fleming, "Mathematical models Representing TimeCurrent Characteristic of Overcurrent Relays for Computer Applications", IEEE PES Winter Meeting, New York, Paper No. A78 131-5, pp 1-8, Jan.-Feb. 1978. [3] Horowitz, S.H., Phadke, A.G., Throp, J.S., "Adaptive Transmission System Relaying", IEEE Transactions on Power Delivery, Vol. 3, No. 4, pp 1436-1445, Oct. 1988. [4] Jampala, A.K., Venkata, S.S., Domborg, M.J., "Adaptive Transmission Protection: Concepts and computational issues", IEEE Transactions on Power Delivery, Vol. 4, No. 1, pp 177-185, Jan. 1989. [5] M. Gilany , A New Digital Overcurrent Relay for Distribution Systems, Journal of Engineering and Applied Science , Cairo University , Vol. 44, No. 6, pp.1091-1105, Dec.1997. [6] C.S. Shifreen, W.C. Marbile, "Charging Current Limitations in Operation of HighVoltage Cable Lines", AIEE Transactions, Oct. 1956, pp 803-813. [7] Electricity Council, "Power System Protection", Peter Peregrinus, U.K.,1981. [8] Gerhard Ziegler,Numerical Distance Protection, Publicis, MCD, Munich and Erlangen, 1999. [9] E.C.Senger,W.Kaiser, "Broken Conductor Protection System Using Carrier Communication", IEEE Trans. Power Delivery, vol.15, No.2 , April, 2000 . [10] C. L. Benner and B. D. Russel, Practical High Impedance Fault Detection on Distribution Feeders, IEEE - IA, vol. 33, no. 3, May/June 1997. [11] Detection of Downed Conductors on Utility Distribution Systems, IEEE Tutorial Course, course text 90EH0310.3.PWR. [12] M. Gilany , E. Aboul-Zahab, Bahaa Hassan, M. Alhadidy, "A New Method For Enhancement of Distance Relay Performance Against Open-Conductor Conditions", 10th International Middle East Power System Conference, MEPCON 2005, Egypt, pp. 613-618, Dec. 13-15, 2005. [1]

INDEX
A
access, 9, 174 accounting, 188, 210 accuracy, x, 2, 42, 56, 208, 215, 286, 305, 309 acid, 209, 249, 251 actuators, 215 adaptation, 288 adjustment, 209, 219 adsorption, ix, 170, 239, 242, 244, 247, 252, 254, 258, 261 afternoon, 138, 153 aggregation, 182 aging, 328 agriculture, 194 air emissions, 248 airports, 251 alcohol, 248 alcohols, 249 algorithm, x, 182, 208, 216, 237, 262, 285, 291, 293, 314, 320 alloys, 251 alternative, 176, 179, 183, 185, 194, 200, 201, 203, 209, 290 alternatives, 177, 179, 181, 182, 183, 184, 186, 192, 194, 200, 201, 202, 256 ammonia, 255 amplitude, 1, 208, 286, 287, 293, 294, 295, 296, 298, 299, 302, 304, 308, 310, 311, 312, 315 arithmetic, 48, 63, 107 ash, vii, 5, 6, 7, 11, 53, 75, 76, 92, 93, 94, 98, 109, 111, 118, 120 Asia, 255 assessment, x, 174, 176, 178, 180, 181, 184, 200, 257, 260, 261, 263, 285, 307, 314 assignment, 184 assumptions, x, 210, 218, 227, 272, 285 atmospheric pressure, 210 attacks, 241 attention, 175, 176, 192, 325, 327 attractiveness, 251 autonomy, 193 availability, 174, 192, 227, 244, 247 averaging, 262 avoidance, 245 awareness, 175

B
banks, 224 barriers, 174 basic services, 174 batteries, 245 behavior, x, 1, 2, 3, 215, 219, 222, 227, 238, 245, 265, 266, 286, 287, 290, 297, 298, 302, 303, 304, 305, 306, 308, 310, 312, 313, 314, 315 Beijing, 109, 120 bending, 20 bias, 6, 9, 75, 88, 89, 90, 91, 94, 200 biodiesel, 247 biodiversity, 180 biomass, 130, 157, 158, 159, 167, 199, 200, 262 birds, 200 blocks, 256 boilers, vii, 5, 6, 7, 52, 66, 71, 92, 93, 97, 121, 122 buffer, 158 buildings, 129, 131, 186, 241, 243, 256, 258, 261 bureaucracy, 187 burn, 81, 91, 92, 108, 109, 120 burner cone angle, vii, 5 burning, 26, 52, 118, 120, 148, 149, 209, 217, 240 burnout, 7, 52, 72, 75, 107, 108, 109, 110, 111, 113, 116, 117, 118, 121, 125

332

Index

C
cables, 321 California, 189, 217, 236, 261 Canada, 244 candidates, 303 capacitance, 321 capital cost, 245, 246, 251, 258 carbon, vii, 5, 11, 54, 55, 71, 75, 76, 78, 81, 93, 98, 111, 117, 120, 121, 129, 130, 134, 147, 149, 151, 153, 155, 176, 249, 251 carbon dioxide, 129, 130, 134, 147, 149, 151, 153, 155, 251 carbon monoxide, 134, 147, 149 carrier, 41 case study, 158, 183 cast, 14 catalyst, viii, 127, 129, 134, 147, 158, 248, 251 C-C, 98 CCHP systems, ix, 239, 241, 245, 252, 256, 257, 258 CEE, 284 cell, viii, ix, 127, 129, 130, 134, 136, 147, 151, 155, 158, 161, 186, 239, 245, 246, 247, 248, 251, 257, 258, 260, 261, 262 ceramic, 14, 95, 249, 251 ceramics, 249 certainty, 177 channels, 7, 8, 9 chemical energy, 210 China, 5, 6, 52, 53, 86, 121, 122, 124, 188, 239, 255, 261 Chinese, 121, 122, 124, 261 chloride, 255 chromium, 110 climate change, 176 closure, 72 clusters, 301 CO2, 77, 115, 116, 118, 121, 128, 145, 149, 150, 151, 153, 155, 194, 241, 243, 246, 249, 250, 251, 256, 257 coal, vii, ix, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 23, 26, 28, 29, 31, 33, 37, 38, 40, 41, 44, 47, 49, 51, 52, 53, 54, 55, 56, 58, 59, 62, 63, 64, 66, 67, 68, 69, 70, 71, 72, 75, 76, 77, 78, 79, 81, 82, 84, 85, 86, 89, 91, 92, 93, 94, 95, 96, 97, 99, 100, 108, 109, 111, 112, 114, 115, 116, 117, 118, 119, 120, 121, 122, 123, 124, 205, 209 coal particle, 11, 28, 52, 53, 75 combustion, vii, ix, 5, 6, 7, 8, 9, 11, 18, 20, 23, 26, 29, 31, 33, 38, 40, 52, 55, 56, 66, 72, 74, 75, 76, 78, 79, 81, 84, 85, 86, 87, 88, 89, 90, 91, 92, 94, 95, 98, 99, 109, 111, 112, 115, 116, 117, 118, 121, 122, 123, 124, 130, 134, 149, 158, 162, 192,

206, 210, 211, 239, 240, 242, 243, 244, 247, 248, 257, 258, 262 combustion chamber, 9, 130, 158, 210, 211 combustion characteristics, 109 combustion processes, 11 commerce, 221 commodities, 175 communication, 176, 315 communication technologies, 176 community, 129, 138, 178 compensation, 275, 276, 277 competition, 174, 175, 209, 217 competitiveness, 176, 187 competitor, 243, 258 complexity, 183, 210, 253, 303 components, ix, x, 1, 14, 81, 98, 111, 174, 182, 209, 210, 217, 234, 244, 248, 251, 252, 257, 286, 287, 288, 297, 298, 299, 300, 301, 302, 308, 309, 313, 315 composition, 11, 54, 71, 78, 81, 92, 95, 123, 124, 142, 145, 154, 251, 310 compounds, 255 compressibility, 207, 218 computation, 231, 289, 290, 296 computer simulations, 246 computer software, 303 computing, 208, 267, 271, 289 concentration, 6, 8, 10, 11, 13, 15, 19, 20, 26, 35, 36, 37, 38, 41, 43, 46, 47, 48, 49, 52, 55, 63, 72, 73, 74, 75, 77, 78, 81, 82, 83, 84, 85, 86, 89, 91, 92, 93, 95, 100, 108, 110, 113, 114, 115, 116, 118, 119, 121, 134, 147, 158, 253 concordance, 85 condensation, 253 conditioning, 255 conductor, 327, 328, 329 configuration, 41, 121, 130, 210, 242, 243, 257, 258 conflict, 6, 201 consciousness, 170 consensus, 178 conservation, 210 constant load, 131, 134 constraints, ix, 205, 215, 266, 267, 268, 271, 272, 273 construction, 192, 193, 251, 260, 321 consumers, vii, 174, 209, 320 consumption, viii, ix, 112, 148, 149, 170, 173, 180, 205, 206, 209, 211, 217, 219, 235, 236, 258 contingency, 221, 266, 271 continuity, 193, 200 control, ix, 1, 2, 3, 6, 48, 92, 95, 130, 161, 163, 164, 165, 166, 167, 174, 205, 207, 208, 209, 210, 211, 212, 213, 214, 216, 223, 224, 237, 238, 247, 248,

Index
253, 256, 257, 258, 262, 266, 272, 273, 274, 275, 310, 313, 314, 324, 325, 327 convergence, 218 conversion, ix, 124, 176, 186, 191, 205, 210, 256 cooking, 244 cooling, ix, 96, 128, 134, 138, 158, 160, 166, 176, 190, 239, 240, 241, 242, 243, 244, 247, 250, 252, 253, 254, 255, 256, 258, 261, 262 Copenhagen, 203 corrosion, viii, 5, 7, 11, 92, 93, 94, 95, 109, 118, 248, 253, 327 cost curve, 187 cost minimization, 238 cost-benefit analysis, 177 costs, viii, 127, 173, 175, 176, 179, 180, 187, 188, 192, 193, 199, 200, 204, 209, 241, 245, 246, 247, 248, 250, 251, 252, 257, 258 coupling, 1, 210 coverage, 12 credibility, 200 criticism, 180 crops, 192, 199 crystallization, 253, 262 customers, 176, 216, 217 cycles, 134, 187 cycling, 245, 257

333

D
damping, x, 285, 287, 295, 296, 298, 314, 315 data set, 257, 289, 290 decay, 26, 27, 28, 29, 58, 296 decision makers, 176, 178, 179, 181, 182 decision making, viii, 173, 174, 175, 176, 177, 178, 179, 180, 181, 182, 183, 201, 202, 203 decision-making process, 178 decisions, 176, 177, 178, 180, 181, 182 decomposition, 285, 286, 288, 308, 312, 313, 314, 315 definition, 42, 56, 178, 297 degradation, 2 delivery, vii, 208, 209, 211, 217, 248 demand, viii, x, 6, 97, 127, 129, 130, 131, 137, 138, 139, 140, 142, 143, 145, 147, 151, 152, 153, 156, 158, 161, 166, 167, 168, 173, 174, 175, 178, 180, 190, 206, 208, 209, 216, 217, 218, 221, 246, 256, 257, 258, 265, 266, 267, 271, 283 demand characteristic, viii, 127, 129 Denmark, 188, 257 density, 14, 22, 41, 42, 207, 217 Department of Energy, 260 deposition, 209 deregulation, ix, 205

designers, 190 detection, 1, 2, 305 developing countries, 188 deviation, 207, 219, 226, 227, 282, 307 devolatilization, 72 diesel fuel, 247 differential equations, 210 diffusion, 27, 29, 33, 36, 38, 58, 72, 76, 101, 102, 105, 123, 202 discharges, 130 discipline, 208 discontinuity, 200, 301 discount rate, 180 distortions, 180 distribution, vii, ix, 3, 6, 11, 18, 23, 26, 27, 35, 41, 42, 89, 105, 107, 121, 142, 143, 145, 155, 174, 175, 178, 205, 208, 209, 217, 221, 224, 226, 227, 228, 229, 230, 231, 233, 234, 235, 236, 237, 241, 274, 275, 278, 288, 317, 318, 320, 321, 329 divergence, 55 diversification, 176 diversity, 202, 320 division, 29, 35, 36, 37, 49, 50, 51, 52, 53, 54, 55, 66, 76, 98 Doppler, 41, 42 drying, 262 durability, 245 duration, 2, 208

E
earth, 321, 323, 328, 329 eating, ix, 134, 239 economic growth, 175, 203 economic performance, 262 economic problem, viii, 173 economics, vii, 179, 262 economies of scale, 175, 188 Education, 121, 170 efficiency level, 193 Egypt, 330 EIA, 237, 238 electric energy, 142, 143 electric power, 130, 228, 236 electric power production, vii Electric power transmission, vii electrical power, vii, 244, 245 electricity, vii, viii, ix, 6, 128, 129, 134, 136, 148, 149, 160, 173, 174, 175, 176, 186, 187, 188, 190, 191, 192, 193, 204, 205, 209, 217, 221, 227, 228, 231, 234, 235, 236, 239, 240, 241, 242, 243, 244, 245, 246, 247, 248, 251, 252, 256, 257, 258, 262 electricity system, ix, 205

334

Index
exercise, 178 experimental condition, 110 exploitation, 202 extinction, 52, 55, 120 extraction, 7, 192

electrodes, 251 electrolyte, 245, 248, 251, 260 electromagnetic, 41 emergence, 241 emission, vii, ix, 5, 6, 7, 8, 11, 26, 53, 55, 75, 78, 85, 86, 92, 93, 109, 119, 121, 128, 130, 150, 151, 239, 241, 243, 246, 257, 258, 259 empirical mode decomposition (EMD), x employees, 193 endothermic, 149 energy, viii, ix, x, 26, 127, 129, 130, 142, 143, 155, 158, 159, 160, 170, 173, 174, 175, 176, 178, 180, 186, 187, 188, 189, 190, 191, 192, 193, 194, 199, 200, 201, 202, 203, 204, 205, 208, 209, 210, 211, 218, 219, 221, 237, 239, 240, 241, 243, 245, 246, 247, 248, 250, 252, 254, 255, 256, 257, 258, 260, 261, 262, 265, 266, 267, 268, 271, 275, 282, 283, 288, 294, 301, 314 energy consumption, viii, 170, 173, 209, 258 energy efficiency, viii, ix, 127, 176, 239, 241, 255, 258 energy markets, ix, 265, 283 energy supply, viii, ix, 155, 173, 174, 175, 190, 193, 200, 239, 241, 245, 258, 259 England, 237 environment, 74, 109, 175, 176, 177, 178, 180, 193, 200, 218, 241, 259 environmental advantage, 248 environmental factors, viii, 173 environmental impact, viii, ix, 127, 129, 174, 180, 200, 203, 204, 209 Environmental Impact Assessment, 179 environmental issues, 175 environmental protection, viii, 173, 175 EPA, 261 equilibrium, 218 equipment, viii, 128, 129, 130, 131, 136, 145, 146, 147, 158, 161, 192, 242, 247, 258, 304, 305, 320 erosion, 95, 97, 98, 122 estimating, 3, 179, 295, 328 ethanol, 127 EU, 188, 262 Euro, 193 Europe, 176, 187, 188, 191, 209, 242, 245, 255, 259, 262 European Commission, 176, 203, 204 European Investment Bank, 179 European Parliament, 202 evaporation, 255 evening, 138, 153 evolution, 1, 2, 286, 291, 295, 299, 301, 305, 310, 312, 313, 314 excitation, 223

F
failure, 95, 304, 321, 326, 327, 328 family, ix, 138, 140, 143, 183, 203, 214, 239, 241, 242, 244, 245, 256, 258, 259, 261, 297 farms, 188 feedback, 161 FFT, 290, 291, 315 fidelity, 210 film, vii, 5, 20 filters, 192, 285, 289, 291, 293, 315 finite impulse response (FIR) filters., x firms, 188 first generation, 7 flame, viii, 5, 6, 8, 10, 11, 12, 15, 23, 26, 28, 33, 48, 52, 54, 55, 66, 72, 91, 92, 93, 94, 95, 98, 108, 109, 119, 120, 123, 125 flame propagation, 33, 91 flexibility, 243, 248, 251 flight, 200 flow field, 11, 21, 24, 26, 27, 59, 70, 122 fluctuations, 118, 167, 218, 305 flue gas, 7, 85 fluid, 24, 190, 192, 204, 237, 255 focusing, 257 forests, 209 fossil, viii, ix, 173, 193, 205, 239 fossil fuels, viii, ix, 173, 205 Fourier analysis, 305 Fourier and Prony methods, x, 285 France, 203, 257, 263, 316 freedom, 1 frequency distribution, 288 frequency resolution, 2 friction, 160, 206, 217, 218, 219 fuel cell, viii, ix, 127, 129, 130, 134, 136, 147, 151, 155, 158, 161, 209, 239, 241, 245, 246, 247, 248, 249, 251, 256, 257, 258, 260, 261, 262 fuel efficiency, 247 fulfillment, 170 fusion, 6, 92, 109, 118

G
gas phase, 27

Index
gas turbine, ix, 122, 205, 208, 209, 210, 211, 212, 213, 214, 220, 223, 224, 226, 227, 228, 230, 234, 236, 237, 238, 252 gases, 81, 109, 206, 210, 211 gasoline, 247, 248, 251 gel, 254, 255, 261 generation, viii, ix, x, 7, 127, 128, 129, 130, 131, 134, 136, 137, 142, 143, 145, 149, 150, 153, 155, 160, 174, 175, 192, 204, 205, 208, 209, 222, 227, 236, 237, 238, 239, 240, 241, 244, 245, 247, 248, 251, 256, 259, 260, 262, 263, 265, 266, 268, 269, 271, 276, 277, 278, 279, 280, 281, 305, 306, 307, 308 generator constraint shift factor, ix, 265, 266, 283 Germany, 187, 188, 237, 260 glass, 41, 42, 43, 49, 50, 56, 86, 100, 188 goals, 175, 176, 275 government, viii, 173, 187, 266 GPC, 216 graph, 201 gravity, 206 grid technology, 129 grids, 128, 129, 131, 137, 142, 143, 145, 174 groups, 179, 320 growth, ix, 175, 187, 203, 205, 208, 209 growth rate, ix, 187, 205 guidance, 40, 52, 55, 63, 107, 109 guidelines, 201, 203

335

housing, 260 human activity, 177 humidity, 22, 256 hybrid, 130, 157, 260 hydrazine, 249 hydrocarbon fuels, 251 hydrocarbons, 109, 174, 250 hydrogen, 121, 136, 146, 149, 151, 155, 158, 170, 191, 248, 249, 251 hydrogenation, 130, 148, 149, 155 hydroxide, 249 hypothesis, 227

I
identification, x, 208, 213, 238, 285, 312, 315 IEEE 13 node power distribution system, ix, 205 IEEE 14 bus system, x, 265, 277 images, 119 imbalances, 180 IMF, 286, 288, 294, 296, 297, 301, 302, 308, 309, 310, 312 implementation, ix, x, 2, 3, 202, 208, 223, 257, 265, 266, 267, 283, 285, 286, 289, 291, 293, 301, 302 import prices, 246 impurities, 248, 251 in situ, 73, 74, 82, 83 incentives, 187 incidence, 220 incomplete combustion, 75 India, 188 indicators, ix, 174 industrial application, 92 industrialisation, 192 industry, ix, 6, 123, 175, 176, 187, 188, 205, 216, 217, 244, 266 inertia, 9, 15, 20, 35, 51, 62, 64, 84, 89, 106, 107, 210, 211 infinite, 208 infrastructure, 178, 236, 238 injections, 207, 220, 222, 266, 269 innovation, 176 input, 111, 166, 208, 212, 213, 215, 216, 222, 223, 224, 244, 251, 253, 255 insight, 305, 313 instability, 1, 2, 38, 209 institutions, 178, 179 instruments, 183 insulation, 245, 321, 323, 328 integrated unit, 178 integration, 128, 208, 256 intensity, 9, 26, 27, 176 interaction, ix, 178, 205, 304

H
Hawaii, 259 health, 175, 176, 180 heat, viii, ix, 6, 7, 23, 26, 28, 52, 91, 108, 127, 128, 129, 130, 131, 134, 136, 146, 148, 149, 151, 153, 157, 158, 160, 162, 170, 188, 190, 191, 192, 204, 206, 207, 218, 239, 240, 241, 242, 243, 244, 245, 246, 247, 248, 250, 251, 252, 253, 254, 255, 256, 257, 258, 260, 261, 262, 263 heat loss, 7 heat transfer, 163, 190, 204, 218 heating, ix, 6, 52, 54, 66, 71, 78, 81, 94, 108, 111, 120, 134, 166, 170, 176, 206, 239, 240, 244, 245, 246, 252, 255, 258, 260, 262 height, 20, 42 helium, 191, 248 Hilbert transform, x, 285, 287, 288, 289, 290, 291, 293, 295, 297, 301, 302, 308, 310, 315, 316 hip, 158 hospitals, 138, 140, 142, 153, 241, 243, 251 hotels, 138, 140, 142, 153, 154, 241, 243, 251 House, 145, 146, 147, 157, 171 households, 138, 140, 143

336
interactions, 1, 178, 299, 315 interface, 25, 28 interpretation, 178, 287, 309 interval, 197, 208, 213, 222, 305, 310 intervention, 180 inversion, 216 investment, 176, 179, 199, 200, 208, 209, 243, 250, 257, 258 investors, 187, 266 ions, 249 isolation, 15, 18 isothermal, 37, 218 Italy, 173, 188, 203, 245, 259

Index

M
machinery, 145, 160 magnet, 223, 224 management, viii, ix, x, 173, 177, 179, 181, 183, 247, 265, 266, 283 manufacturing, 188 mapping, ix, 174 market, viii, ix, x, 173, 174, 175, 180, 187, 193, 200, 205, 209, 217, 242, 243, 244, 245, 246, 248, 255, 256, 258, 262, 265, 266, 267, 268, 274, 275, 282, 283 market penetration, viii, 173, 175 market share, 175 market value, 180 markets, ix, 174, 175, 180, 187, 237, 265, 266, 267, 268, 271, 275, 282, 283 Massachusetts, 123 matrix, 183, 194, 195, 199, 207, 220, 249, 266 measurement, x, 1, 18, 21, 41, 43, 70, 75, 110, 285, 304, 313, 316 measures, 11, 53, 72, 92, 95, 183, 187, 192, 193, 247, 301 mechanical energy, 188, 191, 192 Mediterranean, 199 metals, 249 methane, 246 methanol, 249, 251 Mexico, 203, 285 micelles, 24 micro-grid system, viii, 127, 129, 154 Middle East, 330 Ministry of Education, 121 missions, 130, 149, 151, 153, 155, 176, 194, 246, 247, 250, 256, 257 mixing, vii, 5, 6, 7, 8, 9, 11, 12, 13, 15, 16, 17, 18, 20, 24, 26, 27, 28, 29, 36, 37, 38, 39, 40, 41, 44, 49, 51, 52, 54, 56, 58, 59, 62, 63, 64, 66, 70, 72, 75, 79, 80, 82, 86, 88, 89, 90, 99, 101, 102, 105, 106, 107, 109, 112, 115, 123 modeling, ix, 2, 43, 55, 70, 76, 203, 205, 208, 210, 218, 237, 257, 262, 301 models, ix, 49, 50, 139, 152, 153, 173, 176, 177, 178, 185, 202, 208, 210, 215, 216, 237, 257, 259, 330 modern society, 174 modules, 187 moisture, 6, 54, 76, 255 momentum, 20, 21, 31, 51, 70, 76, 79, 87, 210 money, 175, 179, 180, 243 monopoly, 174, 175, 180 morning, 138, 140, 142, 153 motion, 90, 312, 313

J
Japan, 127, 138, 155, 169, 170, 171, 188, 245, 255 jobs, 194 judgment, 193

K
kerosene, 134, 155 Korea, 255 Kuwait, 317

L
laminar, 123, 219 land, 194, 200 land use, 194 laws, 178, 218 leisure, 241 liberalisation, 175, 176, 187 lifespan, 188 lifetime, 257 limitation, 291 linear model, 208, 215 linear programming, 257, 262 lithium, 251, 255 Lithuania, 257, 262 location, x, 42, 70, 202, 210, 217, 222, 266, 282, 283, 285, 304, 305, 313, 318, 319, 329 long distance, vii LPG, 251 LTD, 120 lubricating oil, 242, 248 lying, 178

Index
motivation, ix, 205 movement, 106, 190 multiplier, 269

337

oxides, 247 oxygen, 11, 48, 53, 72, 73, 75, 81, 84, 91, 114, 248, 251

N
nanotechnology, 187 natural gas, ix, 189, 205, 209, 216, 217, 218, 221, 223, 234, 236, 237, 238, 242, 244, 245, 247, 248, 251 natural resources, 174 neglect, 218 negotiating, 177 Netherlands, 257, 263 network, vii, ix, 130, 157, 175, 176, 218, 219, 220, 221, 228, 230, 237, 265, 266, 267, 268, 271, 272, 283, 304, 316, 317, 318, 320 New York, 122, 203, 237, 238, 284, 330 next generation, 259 nickel, 110 nitrogen, 53, 109, 123, 124, 247 nitrogen oxides, 247 nodes, 206, 217, 218, 220, 227, 230, 267 noise, x, 194, 213, 222, 244, 247, 248, 285, 286, 305, 308, 315 nonlinear systems, 3, 208, 238 non-renewable resources, viii, 173 North America, 259 Nuevo Len, 285 numerical analysis, 130, 161

P
Pacific, 255 pacing, 302 parameter, 17, 76, 128, 161, 164, 169, 187, 193, 206 Pareto, 180 Pareto optimal, 180 Parliament, 202 particle mass, 41, 49, 86, 100 particles, 9, 11, 14, 15, 20, 29, 35, 36, 41, 42, 43, 46, 49, 50, 51, 52, 53, 55, 56, 58, 59, 62, 63, 64, 75, 86, 87, 88, 89, 90, 91, 92, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 121 pathways, 178 payback period, 246 performers, 200, 201 petroleum products, 247 phase diagram, 98 phasor measurement units (PMUs), x physics, 210, 288 pitch, 66, 68 planning, viii, 127, 173, 176, 178, 179, 181, 183, 202, 203, 217, 257 planning decisions, 178 plants, 6, 11, 52, 92, 176, 187, 193, 194, 202, 203, 204, 209, 217, 236, 241, 247, 251, 256, 257, 259, 262 platinum, 95, 248 policy makers, 178 pollutants, 6, 192, 200 pollution, 6, 92, 95, 180 polymer, 129, 245, 246, 260 population, 194, 200 ports, 110 Portugal, 122 potassium, 249, 251 power generation, viii, 127, 128, 129, 130, 131, 134, 136, 137, 142, 143, 145, 153, 155, 160, 174, 175, 204, 236, 238, 247, 248, 251 power plants, 6, 11, 52, 92, 176, 193, 202, 204, 217, 236, 241, 251, 256, 257, 259 power transmission system, vii prediction, 3, 122, 208, 214, 215, 216, 223 preference, 182, 183, 184, 185, 186 present value, 179, 180 pressure, 14, 21, 22, 55, 109, 111, 118, 119, 124, 128, 148, 158, 178, 206, 207, 209, 210, 217, 218, 219, 221, 234, 252, 253, 255 pressure gauge, 128

O
observations, 74, 307 oil, ix, 55, 66, 93, 94, 98, 120, 121, 174, 190, 205, 209, 242, 247, 248, 250 oligopoly, 180 one dimension, 211 operator, ix, 265, 266, 275 opportunity costs, 179 optimal performance, 223, 224 optimization, viii, 127, 130, 203, 215, 238, 262 optimization method, 215 ordinary differential equations, 210 oscillation, ix, 1, 205, 227, 302, 308, 309, 310, 311, 312, 313 output, 134, 136, 142, 145, 156, 157, 158, 160, 161, 178, 208, 212, 213, 215, 216, 224, 243, 244, 251, 268 overload, 130, 271 ownership, 174 oxidation, 147

338

Index
relationship, 1, 84, 182, 218, 257, 259, 273, 325 relationships, 2, 290, 294 relevance, 305, 314 reliability, ix, 188, 193, 227, 239, 241, 245, 247, 251, 256, 258, 266, 329 renewable energy, viii, 129, 173, 174, 175, 176, 190, 202, 255 reparation, 93 reserves, 209 residential buildings, 241, 245, 261 residues, 192 resistance, viii, 5, 13, 15, 16, 17, 18, 21, 22, 28, 86, 92, 93, 95, 98, 99, 124, 219, 221, 222, 251 resolution, 2, 183, 256 resources, viii, x, 1, 173, 174, 179, 265, 266, 305 response time, 161 restitution, 14 returns, 289 rhodium, 95 rice, 180, 268 rings, viii, 173 risk, 193, 194, 200, 245 risk profile, 200 root-mean-square, 24 rotations, 134 roughness, 22, 217, 218 Royal Society, 315 Russia, 8

prices, 180, 187, 188, 246, 257, 262, 266 probability, 177, 319, 320, 329 probe, 20, 71, 72, 110 producers, 175, 217 production, viii, 6, 128, 134, 136, 148, 149, 160, 173, 175, 178, 187, 188, 193, 194, 199, 204, 240, 243, 251, 257, 258 production costs, 187 profit, 179 program, 137, 138, 153, 210, 315 programming, 215, 216, 223, 238, 257, 262 propagation, x, 33, 91, 286, 305, 308, 315 propane, 242, 249 proportionality, 128 proton exchange membrane, viii, 127, 129 prototype, 192, 244, 245 public interest, 174 public sector, 180

Q
quadratic programming, 215, 223 quality of life, 175, 176, 180

R
radiation, 91, 108, 128, 158, 160, 188, 190, 191, 193, 199 Radiation, 127 radio, 12, 13, 17 radius, 12, 15, 20, 21, 44, 46, 59, 102, 105, 107 range, ix, 1, 18, 20, 28, 29, 35, 38, 41, 42, 43, 46, 48, 52, 55, 62, 63, 85, 89, 91, 92, 102, 105, 107, 108, 148, 149, 160, 174, 177, 181, 183, 208, 224, 234, 241, 243, 244, 248, 250, 253, 289, 301 rating scale, 194 rationality, 177 reactants, 210 reaction temperature, 158 real time, 166, 175, 266, 268, 282 reality, 181 real-time basis, 310, 314 reconstruction, 9 recovery, 241, 242, 243, 244, 245, 247 reduction, ix, 18, 234, 239, 241, 247, 252, 256, 258, 319 refractive index, 41, 42 regeneration, 255 regional, 202, 247, 303, 305 regulations, 175, 187, 247 regulators, 217 rejection, 250

S
safety, 175, 176, 215 salt, 189, 204, 251, 255 salts, 190 sample, 72, 111, 200, 224 sampling, 85, 110, 128, 136, 137, 138, 145, 150, 151, 153, 161, 213, 222, 223, 224, 300 savings, viii, 173, 241, 271 scheduling, x, 265, 266, 271 school, 170 search, 174, 181 searches, 137 searching, 324 SEC, 202 security, 3, 208, 256, 266 seed, 41 seeding, 41 selecting, 177, 192 selectivity, 329 sensitivity, ix, 201, 246, 248, 265, 266, 267, 268, 269, 270, 271, 272, 273, 274, 275, 276, 278, 281, 282, 283, 323 sensors, 329

Index
separation, 18, 314 series, x, 2, 3, 190, 227, 286, 287, 300, 301, 302, 303, 309, 314, 315, 327 service provider, 174 SES, 204 settlements, 200 SFT, 266, 267, 271, 272, 274, 282 shadow prices, 180 shape, 6, 20 shares, 251 sharing, 145, 149, 153, 154, 266, 313 shear, 33 shock, 247 shoot, 165 sign, 273 signals, x, 2, 41, 208, 213, 214, 285, 287, 288, 291, 292, 294, 295, 297, 299, 303, 308, 312, 316, 322 silica, 14, 255, 261 silicon, 110, 186, 187 similarity, 70 simulation, 1, 108, 122, 178, 203, 209, 210, 217, 218, 224, 227, 236, 237, 238, 257, 259, 267, 278, 302 sites, 190, 192, 209, 217, 267, 283, 304, 305 slag, 6, 7, 118 smoke, 192 social factors, 180 society, 122, 174, 179 sodium, 190 software, 192, 257, 262, 303 solar collectors, 189 solar energy, 187, 189, 190, 191, 243 solid oxide fuel cells, 251 solid phase, 75 sorption, 255 Soviet Union, 53 Spain, 187, 188, 190, 202, 204, 205 species, 77, 78, 110, 111, 121, 123 specific gravity, 206 specific heat, 206, 207, 211 spectral techniques, 291 spectrum, 288, 289, 290, 297, 306, 307, 308, 310, 311, 315 speed, 53, 93, 109, 164, 167, 188, 206, 207, 210, 214, 221, 223, 224, 225, 226, 227, 248, 256, 318, 319 speed of light, 221 speed of response, 164, 167 spin, 187 stability, vii, viii, 3, 5, 6, 8, 11, 26, 28, 55, 66, 91, 92, 93, 95, 98, 108, 109, 120, 125, 175, 193, 197, 201, 208, 226, 236, 237, 242, 251, 302, 314 stabilization, 10, 11, 53, 54, 208

339

stages, 174 stakeholders, 178 standard deviation, 307 standards, 6, 175 statistics, 3, 262 steel, 14, 95, 97, 110 STM, 244, 250 storage, 120, 131, 146, 157, 175, 189, 190, 193, 199, 200, 204, 208, 209, 211, 217, 221, 248 strain, 251 strategies, 176, 178, 194, 257, 258, 262 streams, 7, 8, 9, 251, 252 strength, 185, 201 stress, 26, 28, 87 structuring, 202 subjectivity, 203 substitution, 227 summer, ix, 138, 151, 239, 257, 258 Sun, 122, 123, 124 suppliers, 248 supply, vii, viii, ix, 5, 23, 28, 81, 85, 86, 93, 109, 127, 128, 130, 131, 134, 136, 142, 145, 148, 151, 154, 155, 157, 158, 162, 167, 170, 173, 174, 175, 176, 178, 180, 190, 192, 193, 194, 200, 205, 217, 239, 241, 244, 245, 252, 258, 259, 263, 268, 319, 321, 323 supply chain, 194 supply curve, 180 suppression, 216, 224 surface area, 190 surplus, 179, 208, 244 sustainability, 176, 193, 209 swirl burners., vii, 5 swirl coal burners, vii, 6 swirl coal combustion, vii, 5 swirling vane angle, vii, 5, 21, 22, 23, 24, 26 Switzerland, 238 symbols, 112, 113, 114, 115, 116, 117, 119 synchronization, 300, 304 systems, viii, ix, x, xi, 1, 2, 3, 76, 127, 136, 145, 146, 155, 174, 175, 176, 177, 178, 188, 190, 191, 192, 193, 194, 203, 205, 208, 209, 211, 215, 217, 218, 221, 222, 223, 227, 235, 236, 238, 239, 240, 241, 242, 243, 244, 245, 246, 247, 248, 250, 252, 255, 256, 257, 258, 261, 262, 266, 267, 285, 286, 291, 293, 295, 297, 299, 300, 303, 304, 305, 306, 313, 314, 315, 317, 323, 325, 327, 329

T
tanks, 221 targets, 175 tariff, 257

340

Index
uncertainty, 55, 76, 177, 202 uniform, 6, 27, 35, 36, 81, 107, 109, 211 unit cost, 192 United States, 186, 191 urban areas, viii, 127 users, 175, 186, 209, 217, 236, 241, 243, 247, 258, 268

technological revolution, 174 technology, 2, 5, 11, 129, 130, 174, 176, 186, 188, 190, 192, 193, 199, 202, 209, 242, 243, 244, 245, 247, 248, 251, 255, 256, 258, 260 temperature, vii, 5, 6, 7, 11, 20, 26, 31, 47, 52, 53, 54, 55, 57, 66, 68, 70, 71, 72, 75, 77, 78, 79, 81, 82, 84, 85, 91, 92, 93, 94, 95, 96, 97, 108, 109, 110, 111, 112, 118, 121, 123, 124, 138, 158, 160, 188, 191, 192, 206, 207, 218, 224, 233, 242, 244, 248, 250, 251, 252, 253, 254, 255, 262 tension, 328 terrorism, 241 Texas, 238 theory, viii, 5, 11, 92, 95, 174, 177, 178, 179, 180, 210, 237, 255, 287, 295, 299 thermal energy, 240, 241, 261 thermodynamic cycle, 192 threat, 176 three-dimensional space, 293 threshold, 185 thresholds, 185, 201 time periods, 266 time resolution, 256 time series, x, 2, 3, 286, 301, 302, 303, 315 timing, 312 titanium, 42, 43 topology, 1, 222, 266, 267, 271, 274, 282 total energy, ix, 239 tracking, 1, 2, 191, 192 trade, 267 traits, 252 trajectory, 214, 223, 224, 294 transformation, 3, 323, 328 transition, 219 transitions, 263 transmission, ix, 130, 162, 174, 178, 205, 207, 208, 209, 216, 217, 218, 221, 234, 236, 238, 241, 265, 266, 267, 271, 276, 277, 283, 284, 305, 316, 321, 327, 328, 329 transmission path, vii transport, 33, 46, 48, 90, 92, 129, 158, 193, 211, 236 transportation, ix, 176, 205, 217, 219, 221, 238, 248 trend, viii, 6, 173, 188, 256, 296, 306, 307, 308 trial, 165, 329 trial and error, 165, 329 trust, 221 turbulence, 24, 26, 29, 43, 46, 48 turbulent mixing, 24, 26 Turkmenistan, 209

V
validation, 3 validity, 2, 187 values, 20, 23, 26, 29, 41, 44, 49, 52, 58, 59, 62, 63, 75, 88, 89, 101, 102, 121, 166, 178, 180, 181, 193, 197, 217, 266, 267, 268, 273, 274, 278, 282, 294, 298, 321 vapor, 252, 255, 256 variability, 315 variable, 187, 212, 248, 256, 266, 268, 272, 289 variables, viii, ix, 2, 173, 205, 215, 224, 275, 298 variation, 18, 55, 120, 291, 315 vector, 184, 206, 207, 220, 293, 323 vehicles, 251 velocity, 7, 9, 14, 16, 18, 21, 22, 23, 24, 26, 27, 28, 29, 31, 33, 38, 40, 41, 43, 44, 46, 49, 51, 52, 53, 55, 59, 72, 77, 79, 86, 88, 91, 97, 101, 105, 108, 217 versatility, 258 vibration, 158, 160, 248, 315 viscosity, 217 visualization, 316 volatility, 53, 93, 256 voltage management., vii voltage sensitivity, ix, 265, 266, 283

W
wages, 193 wall temperature, 95, 96, 97 water absorption, 255 wavelet, 3, 305, 306, 307, 308, 310, 312, 315 wavelet analysis, 306, 307, 312, 315 wealth, 315 wear, 14, 224, 248, 251 welfare, 179 wells, ix, 205, 216, 217 wind, 41, 66, 70, 175, 187, 188, 193, 194, 200, 203, 209, 328 wind turbines, 187, 188, 200 windows, 308 winter, ix, 138, 151, 239, 246, 257, 258 wires, 327

U
UK, 188, 246, 257, 260, 261, 263

Index
World Bank, 179

341

Z Y
zirconia, 249

yield, 191, 192