INDUSTRIAL

COMBUSTION
TESTING
© 2011 by Taylor and Francis Group, LLC
INDUSTRIAL
COMBUSTION
TESTING
Charles E. Baukal, Jr.
CRC Press is an imprint of the
Taylor & Francis Group, an informa business
Boca Raton London New York
© 2011 by Taylor and Francis Group, LLC
CRC Press
Taylor & Francis Group
6000 Broken Sound Parkway NW, Suite 300
Boca Raton, FL 33487-2742
© 2011 by Taylor and Francis Group, LLC
CRC Press is an imprint of Taylor & Francis Group, an Informa business
No claim to original U.S. Government works
Printed in the United States of America on acid-free paper
10 9 8 7 6 5 4 3 2 1
International Standard Book Number: 978-1-4200-8528-0 (Hardback)
This book contains information obtained from authentic and highly regarded sources. Reasonable efforts have been made to publish reliable data and
information, but the author and publisher cannot assume responsibility for the validity of all materials or the consequences of their use. The authors and
publishers have attempted to trace the copyright holders of all material reproduced in this publication and apologize to copyright holders if permission
to publish in this form has not been obtained. If any copyright material has not been acknowledged please write and let us know so we may rectify in any
future reprint.
Except as permitted under U.S. Copyright Law, no part of this book may be reprinted, reproduced, transmitted, or utilized in any form by any electronic,
mechanical, or other means, now known or hereafter invented, including photocopying, microfilming, and recording, or in any information storage or
retrieval system, without written permission from the publishers.
For permission to photocopy or use material electronically from this work, please access www.copyright.com (http://www.copyright.com/) or contact
the Copyright Clearance Center, Inc. (CCC), 222 Rosewood Drive, Danvers, MA 01923, 978-750-8400. CCC is a not-for-profit organization that provides
licenses and registration for a variety of users. For organizations that have been granted a photocopy license by the CCC, a separate system of payment
has been arranged.
Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation
without intent to infringe.
Library of Congress Cataloging-in-Publication Data
Industrial combustion testing / edited by Charles E. Baukal, Jr.
p. cm.
“A CRC title.”
Includes bibliographical references and index.
ISBN 978-1-4200-8528-0 (alk. paper)
1. Furnaces--Testing. 2. Furnaces--Combustion. 3. Furnaces--Industrial applications. I. Baukal, Charles E.
TH7140.I47 2010
621.402’3--dc22 2010014467
Visit the Taylor & Francis Web site at
http://www.taylorandfrancis.com
and the CRC Press Web site at
http://www.crcpress.com
© 2011 by Taylor and Francis Group, LLC
v
Contents
Preface .........................................................................................................................................................................................ix
Editor...........................................................................................................................................................................................xi
Contributors ............................................................................................................................................................................ xiii
I Section General
1. Introduction ........................................................................................................................................................................ 3
Charles E. Baukal, Jr.
2. Testing Safety ................................................................................................................................................................... 41
Charles E. Baukal, Jr.
3. Experimental Design ...................................................................................................................................................... 63
Joseph Colannino
4. Fluid Flow ......................................................................................................................................................................... 77
Wes Bussman and Joseph Colannino
5. Temperature ...................................................................................................................................................................... 97
Charles E. Baukal, Jr.
6. Heat Flux ...........................................................................................................................................................................117
Charles E. Baukal, Jr.
7. Pollution Emissions........................................................................................................................................................141
Charles E. Baukal, Jr.
8. Combustion Noise ......................................................................................................................................................... 183
Mahmoud M. Fleifl, Carl-Christian Hantschk, and Edwin Schorer
9. Flame Impingement Measurements ...........................................................................................................................211
Charles E. Baukal, Jr.
10. Physical Modeling in Combustion Systems ............................................................................................................ 241
Christopher Q. Jian
11. Virtual Testing ............................................................................................................................................................... 251
Eddy Chui, Allan M. Runstedtler, and Adrian J. Majeski
I Section I Advanced Diagnostics
12. Laser Measurements ..................................................................................................................................................... 269
Michele Marrocco and Guido Troiani
13. CARS Temperature Measurements in Flames in Industrial Burners ................................................................ 289
Patrick M. Hughes, Thangam Parameswaran, and Richard J. Lacelle
© 2011 by Taylor and Francis Group, LLC
vi Contents
14. Diode Laser Temperature Measurements .................................................................................................................311
Thomas P. Jenkins and John L. Bergmans
15. Image-Based Techniques for the Monitoring of Flames ....................................................................................... 337
Javier Ballester and Ricardo Hernández
16. High Temperature Cameras ........................................................................................................................................ 355
William J. Lang
17. Liquid Fuel Atomization Testing ............................................................................................................................... 369
Khaled A. Sallam
II Section I Burner Testing
18. Process Burners .............................................................................................................................................................. 377
Jeffrey Lewallen, Thomas M. Korb, Jaime A. Erazo, Jr., and Erwin Platvoet
19. Commercial Boiler Burners ......................................................................................................................................... 395
Yaroslav Chudnovsky and Mikhail Gotovsky
20. Power Burners .................................................................................................................................................................411
Vit Kermes, Petr Be ˇ lohradský, Petr Stehlík, and Pavel Skryja
21. Regenerative Combustion Using High Temperature Air Combustion Technology (HiTAC) ....................... 429
Ashwani K. Gupta, Susumu Mochida, and Tsutomu Yasuda
22. Characterization of Ribbon Burners ......................................................................................................................... 449
Colleen Stroud Alexander and Melvyn C. Branch
23. Flameless Burners ......................................................................................................................................................... 471
Joachim G. Wünning and Ambrogio Milani
24. Radiant Tube Burners ................................................................................................................................................... 487
Michael Flamme, Ambrogio Milani, Joachim G. Wünning, Wlodzimierz Blasiak, Weihong Yang,
Dariusz Szewczyk, Jun Sudo, and Susumu Mochida
25. Metallic Mat Gas Combustion .................................................................................................................................... 505
Giuseppe Toniato, Andrea Zambon, and Andrea D’Anna
26. Performance Prediction of Duct Burner Systems via Modeling and Testing ................................................... 517
Steve Londerville
27. Oxy-Fuel and Oxygen-Enhanced Burner Testing ................................................................................................... 529
Lawrence E. Bool, III, Nicolas Docquier, Chendhil Periasamy, and Lee J. Rosen
I Section V Flare Testing
28. Large-Scale Flare Testing ............................................................................................................................................. 553
Charles E. Baukal, Jr., Jianhui Hong, Roger Poe, and Robert Schwartz
29. Flare Experimental Modeling ..................................................................................................................................... 571
Chendhil Periasamy and Subramanyam R. Gollahalli
© 2011 by Taylor and Francis Group, LLC
Contents vii
30. Flare Radiation ............................................................................................................................................................... 595
Wes Bussman and Jianhui Hong
V Section Testing in Combustors
31. Cement Kilns ...................................................................................................................................................................615
Eugen Dan Cristea and Givanni Cinti
32. Glass Furnaces ................................................................................................................................................................ 671
R. Robert Hayes and Charles E. Baukal, Jr.
33. Thermal Oxidizer Testing ........................................................................................................................................... 691
Bruce C. Johnson and Nate Petersen
34. Utility Boilers ................................................................................................................................................................. 705
Giuseppe Toniato and Silvio Rudi Stella
Appendix A: F-Distribution (99%, 95%, & 90% Confdence) ....................................................................................... 729
Appendix B: EPA Sample Methods ................................................................................................................................... 733
Appendix C: Common Conversions ................................................................................................................................. 735
Index ........................................................................................................................................................................................ 737
© 2011 by Taylor and Francis Group, LLC
ix
This book is intended to fll a gap in the literature for
books on industrial combustion testing. It should be of
interest to anyone working in or with the feld of indus-
trial combustion. This includes burner and furnace
designers, researchers, end users, government regula-
tors, and funding agencies. It can also serve as a refer-
ence work for those teaching and studying combustion.
The book covers a wide range of testing techniques used
in a broad array of applications in the metals, miner-
als, thermal oxidation, hydrocarbon/petrochemical,
and power generation industries. There are 61 authors
from 10 countries representing 33 prominent combus-
tion organizations, and these authors have hundreds of
years of combined experience with industrial combus-
tion testing.
The book contains 34 chapters divided into fve sec-
tions. Section I is a general section with 11 chapters:
Introduction, Testing Safety, Experimental Design, Fluid
Flow, Temperature, Heat Flux, Pollution Emissions,
Combustion Noise, Flame Impingement Measurements,
Physical Modeling in Combustion Systems, and Virtual
Testing. It is designed to provide some of the basic infor-
mation referenced in succeeding chapters. Section II
contains six chapters on advanced diagnostics: Laser
Measurements, CARS Temperature Measurements in
Flames in Industrial Burners, Diode Laser Temperature
Measurements, Image-Based Techniques for the Moni-
toring of Flames, High Temperature Cameras, and Liquid
Fuel Atomization Testing. Section III has ten chapters
on burner testing: Process Burners, Commercial Boiler
Burners, Power Burners, Regenerative Combus tion
Using High Temperature Air Combustion Technology
(HiTAC), Characterization of Ribbon Burners, Flameless
Burners, Radiant Tube Burners, Metallic Mat Gas
Combustion, Performance Prediction of Duct Burner
Systems Via Modeling and Testing, and Oxy-Fuel and
Oxygen-Enhanced Burner Testing. Section IV has three
chapters on fare testing: Large-Scale Flare Testing, Flare
Experimental Modeling, and Flare Radiation. Section V
has four chapters on testing in combustors: Cement
Kilns, Glass Furnaces, Thermal Oxidizer Testing, and
Utility Boilers.
The purpose of this work is to compile testing tech-
niques utilized in industrial combustion for use by
practitioners. No such book currently exists, which
means that those in this feld must consult a range of
sources such as journals, magazines, and conference
proceedings to get this kind of information. This is
generally impractical because those practicing in the
feld usually do not have the time or the resources to
extensively research this topic. While academics have
access to the information, they generally do not work
at the large scales associated with industrial combus-
tion and therefore may not be familiar with how tech-
niques are applied in production applications. This book
is designed to help practitioners both in the feld and in
academics.
Besides providing a single-source reference, this book
also provides information for specifc applications. This
means that someone practicing in a particular area can
immediately go to their application, without necessarily
having to read through other chapters. They can deter-
mine for themselves what is useful for them. The reader
can save time and more quickly use the information pro-
vided by experts in each area. Nearly 1300 references and
over 800 fgures, and 140 tables are provided for those
that need further information on a particular topic.
The book provides case studies and examples to show
how to apply the information for particular applica-
tions. This includes identifying potential problems
that could be very costly if not avoided. For example,
failure to properly measure pollution emissions could
lead to large fnes from regulatory agencies. The book
is designed to be more hands-on and less theoretical so
the information can be easily applied to real situations
in a variety of industries.
This book tells the reader how to make measurements
and conduct tests in industrial combustion systems
including full-scale burners, furnaces, heaters, boilers,
fares, and thermal oxidizers.
There are some topics that are not covered and some that
are not treated extensively. Since the majority of indus-
trial applications use gaseous fuels, there is more treat-
ment of that type of fuel, with less discussion of liquid
and solid fuels. This book concerns atmospheric pressure
combustion, which is the predominant type used in most
industrial applications. There are some burner designs,
combustors, and applications that are not considered.
As with any book of this type, there are sure to be
author preferences and biases, but the coverage is fairly
extensive and comprehensive. There are also generous
discussions of many common industrial applications
to help the reader better understand the require-
ments for different types of tests. Particularly because
of the increasing emphasis on the environment, most
industrial tests include some type of pollution emission
measurements. While industrial combustion testing is
a dynamic area of continuing research, the principles
considered here are expected to be applicable well into
the foreseeable future.
Preface
© 2011 by Taylor and Francis Group, LLC
xi
Charles E. Baukal, Jr., PhD, is the Director of the
John Zink Institute for the John Zink Co., LLC (Tulsa,
Oklahoma) where he has been since 1998. He has also
been the Director of Research and Development and the
Director of the Research and Development Test Center
at Zink, which is a leading supplier of industrial com-
bustion equipment to a variety of industries. Previously,
Dr. Baukal worked for 13 years at Air Products and
Chemicals, Inc. (Allentown, Pennsylvania) in the areas
of oxygen-enhanced combustion and rapid gas quench-
ing in the ferrous and nonferrous metals, minerals, and
waste incineration industries. He worked for Marsden,
Inc. (a burner supplier in Pennsauken, New Jersey)
for fve years in the paper, printing, and textile indus-
tries, and Selas Corp. (a burner supplier in Dresher,
Pennsylvania) in the metals industry, both in the area
of industrial combustion equipment. He has 30 years
of experience in the felds of industrial combustion,
pollution control, and heat transfer and has authored
more than 100 publications in those areas. Dr. Baukal
is an adjunct instructor for Oral Roberts University
and the University of Tulsa, both in Tulsa, Oklahoma.
He is the author or editor of seven books in the feld
of industrial combustion including: Oxygen-Enhanced
Combustion (1998), Heat Transfer in Industrial Combustion
(2000), Computational Fluid Dynamics in Industrial
Combustion (2001), The John Zink Combustion Handbook
(2001), Industrial Combustion Pollution and Control (2004),
Handbook of Industrial Burners (2004), and Heat Transfer
from Flame Impingement Normal to a Plane Surface (2009).
Dr. Baukal has a PhD in mechanical engineering
from the University of Pennsylvania (Philadelphia,
Pennsylvania) and is a licensed Professional Engineer
in the state of Pennsylvania, a Board Certifed Environ-
mental Engineer, and a Qualifed Environmental
Professional. He has served as an expert witness in the
feld of combustion, has 11 U.S. patents, and is a mem-
ber of numerous honorary societies and Who’s Who com-
pilations. He is a member of the American Society of
Mechanical Engineers, the Air and Waste Management
Association, the Combustion Institute, and the American
Society for Engineering Education. He serves on sev-
eral advisory boards, holds offces in the Air and Waste
Management Association and the American Society for
Engineering Education, and is a reviewer for combus-
tion, heat transfer, environmental, and energy journals.
Editor
© 2011 by Taylor and Francis Group, LLC
xiii
Contributors
Colleen Stroud Alexander, PhD, received her doctorate in mechanical engineering from the University of Colorado
at Boulder. Her research focused on the heat transfer, fuid fow, and chemical kinetics involved in methane-air
combustion fame treatment processes. Her PhD research efforts resulted in fve technical publications in multiple
peer-reviewed journals. She carried out her postdoctoral work as a guest researcher at the National Institute for
Standards and Technology in Gaithersburg, Maryland, performing both experimental and numerical analysis in
the study of controlled combustion reactions within reacting fows. She most recently worked as a research engi-
neer studying the performance of advanced fuels in various combustion regimes at the National Renewable Energy
Laboratory in Golden, Colorado. Prior to receiving her degree, Dr. Alexander also worked in the information tech-
nologies sector as a project manager and technical support engineer at CSG Systems (Englewood, Colorado). She
also worked as a pneumatics engineer supporting the assembly of the Atlas Centaur Rocket at Lockheed Martin in
Denver, Colorado.
Javier Ballester, PhD, is currently a professor in fuid mechanics at the University of Zaragoza (Spain), where he has
been since 1997. He received his degree in electrical engineering from the University of Zaragoza in 1992. Previously,
he was hired as a researcher at the Laboratory of Research on Combustion Technologies by the Technological
Institute of Aragon (1991–1992) and by the Spanish Council of Scientifc Research (1992–1997). His areas of exper-
tise are fuid mechanics and combustion, and his current research interests include the combustion of solid fuels,
advanced monitoring and control of industrial fames, and combustion instabilities. He has three patents and has
authored over 90 papers in international journals and conferences. He has participated, in most cases as principal
investigator, in more than 90 research projects, including contracts with private companies and projects funded by
the Spanish and European administrations.
Petr Beˇlohradský, MS, is currently a postgraduate student and he works as a technician at the Institute of Process
and Environmental Engineering at Brno University of Technology (Czech Republic). He holds his degree in math-
ematical engineering from Brno University of Technology. His work is directed toward the research of combustion
on gaseous fuels with special focus on modeling by using statistical methods and computational fuid dynamic
methods. He is an author or co-author of several papers related to combustion modeling presented at international
conferences.
John L. Bergmans, MEng, is the principal engineer and owner of Bergmans Mechatronics LLC (Newport Beach,
California). Bergmans received his degree in mechanical engineering from Carleton University (Ottawa, Canada)
in 1995. He founded Bergmans Mechatronics in 2003 and has since developed data acquisition and control systems
for several rocket motor test stands and an oxyfuel combustion system. Bergmans is also active in the development
and testing of tunable-diode, laser-based instrumentation for large-scale combustion applications. Prior to found-
ing BML, Bergmans was employed for eight years by CFD Research Corp. (Huntsville, Alabama), where he devel-
oped closed-loop pressure controllers for solid-propellant rocket and air-breathing propulsion systems.
Wlodzimierz Blasiak, PhD, is head and professor in the Division of Energy and Furnace Technology, Royal Institute
of Technology, Sweden. He has his degree of applied thermodynamics from Technical University of Czestochowa
(Poland). He has carried out research on heat and mass transfer processes in boilers and furnaces and published
around 200 papers since 1993. For the last ten years the main research themes of his work are high performance
industrial furnaces, high temperature air combustion-HiTAC/fameless combustion for gaseous and solid fuel,
high temperature air/steam gasifcation of biomass and wastes-HTAG, oxyfuel, and fameless oxyfuel combustion.
He also carried out and managed many research projects fnanced by Swedish and international agencies in coop-
eration with European and Japanese industry. He has four patents (three of them are PCT, and two of them are U.S.
provisional-pending) on solid fuel thermal conversions.
Lawrence E. Bool III, PhD, is a senior development associate in the combustion research and development group
for Praxair, Inc. (Tonawanda, New York) where he has been since 1997. Dr. Bool received his doctorate in chemi-
cal engineering from the University of Arizona in 1993. His work focuses on using basic science to develop new
© 2011 by Taylor and Francis Group, LLC
xiv Contributors
oxyfuel applications for industry. Recent examples include a novel process to reduce pollutants from power plants
and a process to produce activated carbon. Dr. Bool holds 20 patents and has authored several peer-reviewed
publications.
Melvyn C. Branch, MS, PhD, is the Joseph Negler Professor of Mechanical Engineering, Emeritus at the University
of Colorado at Boulder. He received his degrees in mechanical engineering from the University of California at
Berkeley. He has previously served as associate dean of Engineering for Research and Administration, associate
dean of the graduate school, and Director of the Center for Combustion Research. He has taught graduate and
undergraduate courses on combustion fundamentals, fuid mechanics, heat transfer, applied thermodynamics,
and fuel technology. His research activity in these areas includes experimental and theoretical studies of combus-
tion-generated air pollutants, fuel effciency, fame processing, metal burning, and aircraft and rocket combustion.
His recent consulting activity includes the 3M Company, the Combustion Research Division of Sandia National
Laboratories, the Air Pollution Control Division of the State of Colorado and the U.S. Federal Trade Commission.
Dr. Branch has served as a member and Chair of the Colorado Air Quality Control Commission, the state agency
responsible for promulgating state regulations relating to air quality, and as a member of the Research Committee
of the Health Effects Institute. He is a member of the Combustion Institute, Tau Beta Pi, Pi Tau Sigma, and a Fellow
of the American Society of Mechanical Engineers. He is a past chairman of the Western States Section-Combustion
Institute. He has been honored with the Society of Automotive Engineers Ralph Teetor Award for engineering educa-
tors and the University of Colorado Teacher Recognition Award for outstanding teacher during the year. His research
awards include the American Society of Mechanical Engineers Gustus L. Larson Award, the Fulbright Fellowship,
the University of Colorado Faculty Fellowship, and the Associated Western Universities Faculty Fellowship. He has
authored over 90 technical articles and supervised 15 students to completion of their PhD.
Wes Bussman, PhD, is a senior research and development engineer for the John Zink Co., LLC (Tulsa, Oklahoma)
where he has been since 1981. He received his degree in mechanical engineering from the University of Tulsa
(Tulsa, Oklahoma). Dr. Bussman has 19 years of basic scientifc research work, industrial technology research and
development, and combustion design engineering. He holds ten patents, has authored several published articles
and conference papers, and has been a contributing author to several combustion-related books. He has taught
engineering courses at several universities and is a member of Kappa Mu Epsilon Mathematical Society and Sigma
Xi Research Society.
Yaroslav Chudnovsky, MS, PhD, is a senior staff member of research and development at the Gas Technology
Institute (Des Plaines, Illinois). Dr. Chudnovsky received his degrees in 1982 and 1990, respectively, from Bauman
Technical University (Moscow, Russia). He conducts research and development of advanced, low-emissions, high-
effciency, and high heat transfer combustion systems and technologies for industrial applications. He has over 25
years of combined basic and applied research and development experience in engineering, design, and laboratory/
feld evaluation of advanced energy exchange and combustion systems and technologies. Prior to joining the Gas
Technology Institute in 1995, he worked as a head of the research laboratory at Power Machinery Research Institute
(Moscow, Russia) where he developed solutions for energy, space, and military applications. His areas of interest
include: heat transfer enhancement and waste heat recovery, convective heat transfer and heat exchangers, advanced
combustion and environmental technologies, and smart thermal management. He has over 100 publications and
six patents. He is the editor of the Heat Exchanger Design Handbook and the Journal of Enhanced Heat Transfer.
Eddy Chui, PhD, is currently a senior research scientist with CanmetENERGY, the clean energy research and tech-
nology development centre of Natural Resources Canada of the Canadian Government (Ottawa, Canada). Dr. Chui
received his degree in mechanical engineering from the University of Waterloo (Ontario, Canada) in 1990. Prior to
joining CanmetENERGY in 1993, he had worked for Bechtel Canada in project engineering, University of Alberta in
acoustic research, and Advanced Scientifc Computing Ltd. (presently ANSYS Canada) in numerical modeling. At
CanmetENERGY, he is responsible for directing and conducting research on various aspects of combustion mod-
eling technology and utilizing the model to assist industries in practical applications. Past achievements include
the development of new modeling strategies to predict NO
x
formation in coal fames and natural gas fames, deter-
mination of the sensitivity of combustion performance to coal blending, design of a new generation of furnace
model for process simulation, evaluation of combustion performance in appliances using biomass, development of
strategies for industrial processes to convert to a lower-carbon fuel, and the successful implementation of the model
to resolve combustion-related problems in full-scale units like utility boilers, coke ovens, refnery furnaces, blast
© 2011 by Taylor and Francis Group, LLC
Contributors xv
furnaces, and industrial furnaces for metal processing. Current research efforts are focused on developing clean
coal technologies: oxy-coal combustion and coal gasifcation for CO
2
capture, modeling CO
2
storage in subsurface
environments, new computational fuid dynamic tools for nonexpert users, and assisting the power sector in China
and Canada to burn coals more cleanly and effciently through the use of simulation. Also, a new capability of
microscale modeling has been developed under his supervision, presently being implemented on investigating
solid oxide fuel cells. He has authored and co-authored over 150 publications in international journals, conference
proceedings, industrial reports, and government departmental reports.
Giovanni Cinti is the technology department manager for the Technical Centre of Italcementi Group (CTG), located
in Bergamo (Italy). He received his certifcate degree at Politecnico di Milano in 1973 as a chemical engineer. In 1975
he started his professional activity in Italcementi SpA in the central headquarters as a member of the combustion
department, dealing with all the aspects of cement kiln burners and related combustion pollutants. He is a member
of the International Flame Research Foundation, holding the chairmanship of the Italian Association for four years
(2004–2008). He has represented the company in the Associazione Tecnico-Economica del Cemento (AITEC) and in
the European Cement Association (Cembureau) and was the cement expert of the Italian delegation in the meeting
for the defnition of Best Available Technologies for Cement Manufacturing in 2001 and 2007.
Joseph Colannino, BS, MS, is director of engineering for John Zink Co., LLC, where he has worked for the
last 12 years. He received his degree in chemical engineering with minors in materials and chemistry from the
California Polytechnic University at Pomona and a degree in knowledge management with emphasis in orga-
nizational dynamics from the University of Oklahoma. He is a registered professional engineer in the state of
California. He has been engaged in combustion research for more than 20 years and has authored many papers
and presentations. His book, Modeling of Combustion Systems: A Practical Approach (Taylor & Francis), was published
in 2006. Besides the John Zink Handbook, Joseph has also contributed book chapters in other volumes including the
Industrial Combustion Handbook (CRC Press), and the Air and Pollution Control Equipment Selection Guide, (Lewis). He
is an adjunct faculty member at Tulsa University and Oral Roberts University (both in Tulsa, Oklahoma) teaching
combustion and engineering courses. Colannino is listed in several Who’s Who compilations.
Eugen Dan Cristea, PhD, MEng, has worked since 1987 in the cement and lime industry in the positions of techni-
cal director of Cimprogetti SpA, an engineering company located in Bergamo (Italy) and today as a process function
manager of Italcementi Group in Bergamo. He received his doctorate degree in thermal sciences from the Politehnica
University of Timisoara (Romania) and his engineering degree in power generation engineering from Politehnica
University of Bucharest (Romania). He did postdoctoral work as a visiting adjunct assistant professor at Montana
State University (Bozeman, Montana) performing numerical simulation combustion for MHD combustor fred on
natural gas. He has served as head of Combustion and MHD laboratory of Scientifc Research Division (formerly
the Power Institute of Romania Academy) of the Institute of Scientifc Research and Technological Engineering for
Power Equipment in Bucharest. He conducted some fundamental and mainly applied research works in all areas of
thermal sciences including combustion science and combustion engineering, heat and mass transfer, fuid mechanics,
thermodynamics and chemical thermodynamics, and direct energy conversion, with particular emphasis on experi-
mental as well as computational approaches. He is a member of the American Society of Mechanical Engineers, of
the International Flame Research Foundation at Pisa (Italy), and has served on the Italian Flame Research Committee.
Dr. Cristea has authored and co-authored two combustion-related books, over 20 journal articles, over 20 conference
papers and holds four patents for novel burner development. He has delivered seminar lectures at the International
Flame Research Foundation.
Andrea D’Anna, PhD, is an associate professor of chemical engineering at Università “Federico II” di Napoli
(Napoli, Italy) where he has been since 2001. He has a degree in chemical engineering from Università “Federico II”
di Napoli (Napoli, Italy). He was a researcher at Istituto Ricerche Combustione, CNR and at Fertimont, Montedison
SpA. His research interests include combustion chemistry, chemical kinetics, combustion-formed particles and
their effects on health and climate, nano-material synthesis, characterization and modeling, transport properties of
nano-materials, and fltration procedures. He is the author of over 100 technical publications.
Nicolas Docquier, PhD, is the general manager of ACI (Atlanta, Georgia), a division of Air Liquide Advanced
Technologies US, specializing in combustion equipment for steel, nonferrous, and glass industries. He has a doctoral
degree in energy sciences from the Ecole Centrale de Paris (France), a degree in fuid mechanics from the von Karman
© 2011 by Taylor and Francis Group, LLC
xvi Contributors
Institute (Belgium), and an engineering degree from Université de Liège (Belgium). He has been a combustion spe-
cialist with Air Liquide since 2003, in France and in the United States. He also worked at IFP Powertrain Engineering
(Paris, France) and for Rolls-Royce Industrial and Marine Gas Turbines (Coventry, United Kingdom). He has a strong
industrial and research background in combustion, industrial heating and melting processes, fuid mechanics, heat
transfer, fuels and emissions, and has developed several combustion test platforms. His experience includes oxy-
burners, furnaces and melting processes, safety systems and practices, emission measurements and sensors, optical
diagnostics, internal combustion engines, and turbomachinery. He is the author or co-author of over 20 publications
on these topics, has nine patents for novel burner and sensor development and has taught graduate courses on fuid
mechanics.
Jaime A. Erazo, Jr., MS, is a design/test engineer at the John Zink Co., LLC, Tulsa, Oklahoma. He has worked for
the John Zink Company Process Burners group for one year. He graduated with a degree in mechanical engineer-
ing from the University of Oklahoma in 2008. He authored fve combustion related technical publications and
presentations during his time at Oklahoma University.
Michael Flamme, PhD, is internationally known for his work on gas-fred technology over a period of more than 20
years with Gaswärme-Institut (Essen, Germany). In 1989 he received his degree from Bochum University, Germany
for his work focused on high temperature processes using high preheated combustion air. He has particular exper-
tise and knowledge of high temperature industrial processes, combustion technologies for gas turbines and boil-
ers, and waste and biomass conversion to energy. He authored over 110 publications in national and international
journals and conference proceedings. His scientifc achievements were rewarded by the Wilhelm Jost Medal of
the German Section of the Pittsburgh Combustion Institute in 1993. He currently manages his own independent
energy consultancy (FlammeConsulting) in Essen, Germany.
Mahmoud M. Fleifl, BS, MS, PhD, is a senior thermoacoustic and vibration engineer in the research and develop-
ment department of John Zink Co., LLC (Tulsa, Oklahoma). He has been with the company since 1999. Dr. Fleifl
graduated from Ain Shams University, Cairo, Egypt with his degrees in mechanical engineering, and his doc-
toral degree in mechanical engineering from a co-supervisory program between Ain Shams University and MIT.
His areas of expertise are fuid dynamics, combustion instabilities, and noise control. He published eight journal
articles and over 20 conference papers. He has over 13 years of experience in advanced techniques of acoustically
driven combustion instability and noise control. He is a member of ASME and AIAA. He is an honored member
of several Who’s Who compilations.
Subramanyam R. Gollahalli, MASc, PhD, is a professor and holds the Lesch Centennial Chair in the School of
Aerospace and Mechanical Engineering at the University of Oklahoma (Norman, Oklahoma) where he has been
since 1976. He received his Master’s degree in 1970 and his doctoral degree in 1973 both in mechanical engineering
from the University of Waterloo (Waterloo, Canada). He has held the positions of assistant professor, associate pro-
fessor, professor, and Lesch Centennial professor, and academic director. He also worked as a research assistant
at the University of Waterloo (Waterloo, Canada) and as a lecturer at the Indian Institute of Science (Bangalore,
India). He has developed and taught courses in the combustion and energy areas at both undergraduate and
graduate levels. He investigated the combustion of emulsifed fuels and synthetic fuels, the publications based on
his research, which are cited frequently. He served on the following technical committees: Combustion and Fuels
Committee, Gas Turbine Division, ASME; Terrestrial Energy Committee, AIAA; Propellants and Combustion
Committee-AIAA; Fuels and Combustion Technologies Committee, ASME; Emerging Energy Technologies
Committee, ASME; and the Technical Program Committee, Combustion Institute. He has been recognized with
the following awards: Regents Award for Superior Teaching at the University of Oklahoma, Energy Systems
Award-AIAA, George Westinghouse Gold Medal-ASME, AIAA Sustained Service Award Robert Angus Medal,
Engineering Institute of Canada, and Three Best Paper Awards, ASME Fellow, AIAA Associate Fellow. He was
the associate editor of the Journal of Energy Resources Technology (1994–2000) and the associate editor of the Journal
of Engineering for Gas Turbines and Power (2000–2006).
Mikhail Gotovsky, MS, PhD, ScD, is a leading researcher at NPO CKTI (1963 to present) and a professor at
GTURP (State University of Plant Polymer Technologies). He received his frst degree in 1963 from St-Petersburg
State Polytechnic University (Russia) and his other two degrees in 1970 and 2000, respectively, from Central
Boiler-Turbine Institute (Russia). After graduating from St-Petersburg State Polytechnic University, Dr. Gotovsky
© 2011 by Taylor and Francis Group, LLC
Contributors xvii
joined the team of highly experienced and motivated professionals at NPO CKTI and grew from a junior engineer
to the leading research and development professional in the area of heat and mass transfer for industrial applica-
tions. His areas of interest include: a variety of heat transfer problems (liquid metals, forced convection, high heat
fux boiling), two-phase fow hydrodynamics and heat transfer, heat transfer enhancement, and thermal problems
of nuclear waste transportation and storage. He has over 120 publications and 12 patents.
Ashwani K. Gupta, PhD, DSc, is a distinguished university professor at the University of Maryland (College Park,
Maryland) where he has been a professor of mechanical engineering. Prior to this he was a member of the research
staff at MIT, and an independent research worker at the University of Sheffeld, United Kingdom. His main research
interests have been in the felds of combustion, air pollution, propulsion, high temperature air combustion, swirl
fows, diagnostics, fuel sprays, fuel reforming, sensors, microscale combustion, and wastes to clean energy conver-
sion. He has co-authored three books on swirl fows, and fowfeld modeling and diagnostics, and high temperature
air combustion: from energy conservation to pollution reduction. In addition he has authored nine book chapters
and published over 450 archival papers in journals, refereed symposia, and conference proceedings. His honors
and awards include: AIAA Energy Systems Award, AIAA Propellants and Combustion Award, ASME George
Westinghouse Gold Medal, ASME James Harry Potter Gold Medal Award, ASME James N. Landis Medal Award,
ASME Worcester Reed Warner Medal Award. Dr. Gupta received the University of Maryland President Kirwan
Research Award and College of Engineering Research Award. He received eight Best Paper Awards from ASME
and AIAA for his research contributions. He is the founding co-editor of the Energy Engineering and Environment
Series published by CRC Publishers. He is an associate editor of the Journal of Propulsion and Power, Journal of Applied
Energy, International Journal of Sprays and Combustion Dynamics, and International Journal of Reacting Systems. He has
served as chair of AIAA Terrestrial Energy Systems Technical Committee, chair of Propellants and Combustion
Technical Committee, deputy director of Energy Group, and director of Propulsion and Energy Group. At ASME
he served as chair for Fuels and Combustion Technology Division, and Computers in Engineering Division. He is
cited in Who’s Who in America, Engineering, Technology, American Education, and Aviation in the United States, and The
Men of Achievement in the United Kingdom.
Carl-Christian Hantschk, PhD, has been working as a consulting engineer in industrial acoustics for Müller-
BBM GmbH (Munich, Germany) since 2001. He was promoted to managing director in 2009. He works on indus-
trial acoustics in general, including theoretical and applied acoustics, environmental acoustics, aero-acoustics and
numerical acoustics, with special focus on the interdisciplinary feld between combustion and acoustics. He holds
a diploma in mechanical engineering and received his doctorate in thermodynamics from the Technical University
Munich, Germany. His research focused on combustion-driven acoustic oscillations in burners and combustion-
acoustic interactions. He gave lectures on chemical thermodynamics, thermal radiation, and heat transfer and
acoustics at his university, international conferences, and for industrial clients. His work resulted in 30 publications
and four invention disclosures. As one of his main research projects, he codeveloped an active acoustic feedback
control for industrial combustion systems.
R. Robert Hayes, MS, is the vice president of El Dorado Engineering, Inc. (Salt Lake City, Utah) where he has
been since 2006. He also worked at the National Renewable Energy Laboratory and the John Zink Co., LLC. He
received his degree in mechanical engineering from Brigham Young University. He has a strong technical back-
ground in combustion, heat transfer, fuels, and emissions formation/reduction, with experience in both industrial
and research facilities. He led a world-class alternative fuels and advanced vehicle technology research facility
at the National Renewable Energy Laboratory. His experience includes internal combustion systems, alternative
fuels, furnaces, burners, air/exhaust handling systems, pollution control systems, emissions measurements, instru-
mentation, safety systems, and combustion of energetic materials (propellants and explosives). He is the author
or co-author of over 20 publications on these topics. He has three patents for novel burner development and has
taught professional courses on burners, formation and control of combustion emissions, heat transfer, and fuid
mechanics.
Ricardo Hernández, graduated in 2003 with a degree in physics at the University of Zaragoza (Zaragoza,
Spain), is currently a PhD student there and works in the Laboratory of Research on Combustion Technologies.
He works on research of advanced monitoring and control of industrial fames and combustion instabilities,
has participated in some research projects and is co-author of several papers in international journals and
conferences.
© 2011 by Taylor and Francis Group, LLC
xviii Contributors
Jianhui Hong, BS, PhD, is a fare process engineer at John Zink Co., LLC (Tulsa, Oklahoma). He is the princi-
pal investigator and lead inventor of the ultra-stable WindProof™ pilot, the triple-redundancy InstaFire
TM
fare
pilot, and the ultra-effcient, steam-assisted Steamizer-XP™ fare. He received a degree from Tsinghua University,
Beijing, China and his doctorate from Brigham Young University, Provo, Utah, both in chemical engineering. He
has other U.S. and foreign patents including low NO
x
incinerator apparatus and control method and air-assisted
fare. He also worked as a research and development engineer at John Zink. His other areas of expertise include
ground fare design and optimization, kinetic simulation involving NO
x
, SO
x
, and soot; global optimization of
steel stack structure considering structural and process constraints; phased array of thermal radiometers for
measuring the fame epicenter and radiant fraction of industrial fares; fare smoke control method; fare control
for over-steaming/over-aeration prevention. He has authored and co-authored over 15 journal articles and book
chapters. His personal interests include heli-plane design, aircraft emergency landing system, and personal aerial
vehicle design.
Patrick M. Hughes, MSc, is the group leader for measurement systems and combustion kinetics at Canmet ENERGY
Ottawa (Ottawa, Canada). He has been a research scientist with CanmetENERGY since 1982. Before that he was a
defense scientist for six years with National Defence Canada. He has his degree in mechanical engineering from
the University of Waterloo (Waterloo, Canada). Throughout his career he has developed advanced measurement
techniques to study combusting fow felds. His research with Natural Resources Canada has involved the use of
laser-based and other optical techniques to study industrial burner technology. He is also involved in the devel-
opment of techniques to characterize coal combustion kinetics and deposition in power boilers. His publications
cover optical measurement techniques and their application to industrial burners, data packages for evaluation of
computational fuid dynamic models, advanced characterization techniques for coal combustion and deposition,
and rocket motor instabilities. He is currently the editor-in-chief of the Combustion Journal of the International Flame
Research Foundation.
Thomas P. Jenkins, PhD, is a senior scientist at MetroLaser, Inc., where he has worked since 2000. He received his
degree in mechanical engineering from the University of California at Davis. He has been principal investigator
on eight research programs for the DoE, Air Force, NASA, Army, and private industry to develop laser-based diag-
nostics for studying combustion and fuid mechanics. He has demonstrated several frst-of-a-kind measurements,
including quantitative nonintrusive measurements of soot concentration in an aircraft engine exhaust, nonintrusive
temperature and H
2
O concentration in an industrial glass furnace, and a large area fow velocimetry system for
studying parachutes. He is a member of the American Institute of Aeronautics and Astronautics (AIAA), and is an
active member of the AIAA Aerodynamic Measurement Technologies committee. He has been the primary author
on more than 30 journal articles and conference papers. Prior to coming to MetroLaser, Dr. Jenkins worked for three
years as a research associate at Stanford University, where he developed soot diagnostics for advanced propulsion
systems.
Christopher Q. Jian, PhD, is the director of the Simulation Technology Solutions Group (STS) at John Zink Co.,
LLC (Tulsa, Oklahoma) where he has been since 2004. He received his degree in mechanical engineering from the
University of Maryland at College Park (College Park, Maryland). Prior to joining John Zink, he was the business
research manager at Owens Corning responsible for asset utilization and customer proftability analyses and merg-
ers and acquisitions. He was the research and development manager at Vortec Corporation before he joined Owens
Corning in 1995. He served as past chair of the production effciency subcommittee of the Glass Manufacturing
Industry Council and a member of its executive advisory committee. Dr. Jian’s research areas include fossil fuel
combustion, glass melting and delivering, computational fuid dynamics and physical modeling, as well as low
level radioactive material vitrifcation. He holds four U.S. patents and has authored and/or co-authored over 80
technical publications.
Bruce C. Johnson, MSc., PE, is the technology development leader for the Thermal Oxidation Systems Group at
the John Zink Co., LLC (Tulsa, Oklahoma). He received his degree in chemical engineering from the University of
North Dakota under a Bureau of Mines research fellowship. He has spent much of his career in research and devel-
opment and has been employed by the Calgon Corporation, Department of Energy, Combustion Engineering Co.,
and several thermal oxidation companies. His qualifcations include process design, new product development,
testing, and project management of governmental and corporate research and development groups. He has fve
patents and has authored numerous technical reports and papers.
© 2011 by Taylor and Francis Group, LLC
Contributors xix
Vit Kermes, PhD, currently works as a lecturer at Brno University of Technology (Czech Republic). He holds his
degree in process engineering from Brno University. His work is directed at applied and industrial research of
reduction of NO
x
emissions in combustion of gaseous fuels and industrial research of nonstandard liquid combus-
tion such as liquid wastes and renewable liquid fuels. He is an author and co-author of about 20 papers related to
combustion modeling presented at international journals and international conferences.
Thomas M. Korb, PhD, PE, is a technical leader in the Process Burner Group at the John Zink Co., LLC (Tulsa,
Oklahoma). He received his degree in mechanical engineering from Arizona State University (Tempe, Arizona).
Dr. Korb has 15 years of experience in combustion and thermal sciences. His work has included design and
testing of combustion equipment for the refning and petrochemical industries as well as failure analysis engi-
neering of accidental fres and explosions. He has also worked in the development of both gas turbine and diesel
engines. He is a registered professional engineer and is a member of Tau Beta Pi National Engineering Honor
Society, American Society of Mechanical Engineers, Society of Automotive Engineering and the Experimental
Aircraft Association. His research focuses on fundamental ignition mechanisms with a particular emphasis on
hot surface ignition of hydrocarbon fuels and the impact of hot surface material and surface oxide structure.
He is a recipient of the Darryl E. Metzger Scholarship and Dean’s Graduate Scholars Award at Arizona State
University.
Richard J. Lacelle, CET, LSO, is an electro-optics technologist working at CanmetENERGY, Natural Resources
CANADA (Ottawa, Ontario). He is a graduate of Algonquin College of Applied Arts and Technologies (Ottawa,
Ontario, Canada) in the feld of electronics engineering technology and electronics engineering techniques. His ini-
tial research project was the commissioning of the coherent anti-Stokes Raman spectroscopy (CARS) system for the
measurement of high temperature combustion fames in industrial burners. He is also the laser safety offcer who
oversees all laser operations at the CanmetENERGY Bells Corners Complex located west of Ottawa. Lacelle has been
a member of the CARS team at CanmetENERGY since its inception. He is responsible for the development, assem-
bly, bench testing, operation, and data acquisition of the CARS system. For the past 26 years he has been working
on research projects like very high speed electronic triggering circuits for the various laser applications, Schlieren
photography, high speed camera imaging, laser sheet visualization, infrared imaging and analysis, laser doppler
velocimetry, Canmet fame identifcation control system, tuneable diode laser absorption spectroscopy, and laser
induced breakdown spectroscopy. He has authored over 20 technical documents, as laser safety manuals based on
ANSI Z136.1 standards for the safe use of lasers, operator’s manual, and standard operating procedures for the above
technologies.
William J. Lang, BS, is vice president and co-owners of Lenox Instrument Co., Inc. (Trevose, Pennsylvania), manufac-
turer of the FireSight high-temperature video camera system, along with a full line of other remote visual inspection
equipment. A graduate of La Salle College, he began his career in the Lenox shop fabricating high-temperature lenses
and optics, and he later pioneered the use of the portable FireSight camera system that is in wide use in fossil-fuel
power plants. He has 42 years of application engineering experience visual inspection and process monitoring. He
has written articles that have appeared in a variety of technical publications. One such article, “Furnace Cameras
Assist in NO
x
Reduction” appeared in Power Engineering magazine in November 2002. His extensive background in
applications include: installing a furnace camera in the combustion chamber of an operating gas turbine, monitor-
ing nuclear waste encapsulation in glass, chemical and biological warfare incineration, and thousands of boiler and
furnace installations around the world. He has experience with all fossil fuels.
Jeffery Lewallen, BSME, PE, is the applications sales manager of the burner division of Callidus Technologies by
Honeywell (Tulsa, Oklahoma). He is a University of Tulsa graduate with a degree in mechanical engineering and
a professional engineer licensed in Oklahoma. He has over 17 years of combustion related experience including
design engineering, production testing, technical feld support, sales, and project management for global projects
in the refning and petrochemical industry. He is a contributing author in the books Industrial Burners Handbook and
the John Zink Combustion Handbook.
Steve Londerville, BSME, is currently director of design engineering at Coen Company (Foster City, California). He
received his degree from San Jose State University (San Jose, California) in 1977. Previous positions, since 1978, at Coen
were chief technical offcer research and development, vice president research and development, director research
and development, and chief engineer. During the last 31 years he has been involved with all aspects of product
© 2011 by Taylor and Francis Group, LLC
xx Contributors
development at Coen. He holds seven patents and has authored 15 publications. He is a member of ASME, ACHIE,
Combustion Institute, Tau Beta PI, and past offcer and board of directors for Institute for Liquid Atomization and
Spray Systems. He was recognized as Engineer of the Year by ASME, Santa Clara Valley section. He also received a
Best Paper Award from ASME.
Adrian J. Majeski, MSc, is a research engineer at Natural Resources Canada’s CanmetENERGY research center in
Ottawa, Canada. He received his degree from the University of Alberta where he participated in the Flare Research
Project. Since joining in 2001, he has worked on computational fuid dynamic models of both pilot- and industrial-
scale combustion systems including utility boilers and equipment related to metal processing and petroleum refn-
ing. His current research includes model development for clean-coal technologies, such as gasifcation and oxyfuel
combustion. Before joining CanmetENERGY, Majeski worked on low-swirl burner technology at Lawrence Berkeley
National Laboratory.
Michele Marrocco, PhD, is a researcher in laser spectroscopy at ENEA (Rome, Italy) (1999 to present). He received
his degree in physics from the University of Rome in 1994. He was employed as a postdoctorate at the Max-Planck
Institute for Quantum Optics (Munich, Germany), as a researcher at the Quantum Optics Labs at the University of
Rome (Rome, Italy), and as an optics researcher by the army. His research activities include: traditional and inno-
vative spectroscopic techniques for diagnosis of combustion and nanoscopic systems studied by means of optical
microscopy. The techniques used include: adsorption, laser induced fuorescence, spontaneous Raman, stimulated
Raman gain, stimulated Raman loss, coherent anti-Stokes Raman, degenerate four wave mixing, polarization spec-
troscopy, laser induced breakdown, laser induced incandescence, laser induced thermal gratings. He has over
30 technical publications.
Ambrogio Milani, DrIng, is a consultant for WS GmbH (Germany). He received a degree from the Politecnico di
Milano (Milan, Italy) in 1965. He has 40 years of experience in combustion technology in energy intensive indus-
trial sectors (steel and power generation). He was the head of the former CSM Experimental Station on combustion
devoted to research and development for products and industrial processes. He works with the International Flame
Research Foundation and the Combustion Institute. He is the manager of ECSC-funded research and development
projects and of educational/training programs and courses. His interests include: combustion research (mainly
iron and steel making), burner development, heat recovery, high effciency, low emissions, and fameless oxidation.
He has a number of technical publications including co-authoring the Handbook of Burner Technology for Industrial
Furnaces.
Susumu Mochida is the director and general manager of Technology & Engineering Division in Nippon Furnace
Co., Ltd. (Yokohama, Japan) where he has been since 1982. He has participated in a number of projects and has been
an active member of the high temperature air combustion (HiTAC) project team. The HiTAC technology has been
widely adapted in industrial furnaces to save energy, reduce size of the equipment, and reduce pollution emission.
He serves as chairman on the Japanese Flame Research Committee and a member of the Executive Committee of
International Flame Research Foundation from 2006. He has authored over 25 technical publications, contributed
to 20 patents, and made several presentations at meetings and conferences. He has been honored with AIAA Best
Paper Award in 1999 and ASME George Westinghouse Silver Medal Award in 2001.
Thangam Parameswaran, PhD, is a research scientist at CanmetENERGY, Natural Resources Canada (Ottawa,
Canada). She obtained her degree from Northwestern University (Evanston, Illinois). Her early research was focused
on the theoretical aspects of the optical properties of organic and transition metal complexes. During later years her
research activities at Carleton University (Ottawa, Canada) involved theoretical and experimental aspects of laser
Raman spectroscopy of transition metal compounds. Subsequently she worked for the National Research Council of
Canada toward the development of coherent anti-Stokes Raman spectroscopy (CARS) for combustion diagnostics.
Dr. Parameswaran has been a member of the CARS teams at NRC, Canada and CanmetENERGY for many years.
During this period she was responsible for developing and applying theoretical calculations and analysis methods
for retrieving information from CARS spectra. For the past 11 years she has been working as a research scientist at
CanmetENERGY, Natural Resources Canada. She has also developed methods for applying fame emission spec-
troscopy for fame performance monitoring in industrial burners. Recently this approach was tested in an industrial
boiler and has the potential to be implemented in a fame advisory system. Other optical methods she has initiated at
CanmetENERGY are tunable diode laser absorption spectroscopy for stack gas measurements in pilot-scale facilities
© 2011 by Taylor and Francis Group, LLC
Contributors xxi
and laser induced breakdown spectroscopy for trace metal detection in combustion emissions. She has authored
over 60 technical documents, as journal publications, conference proceedings, presentations, and contract reports
in the felds of optical spectroscopy and combustion applications of spectroscopic methods. In 1993 she received a
Joint Staff Performance Award as a member of the Advanced Combustion Diagnostics Technique (Coherent anti-
Stokes Raman Spectroscopy) team, of the Institute for Chemical Processes and Environmental Technology, National
Research Council of Canada.
Chendhil Periasamy, PhD, BS, MS, a research scientist at Air Liquide Delaware Research and Technology Center
(Newark, Delaware) since 2007. He received his doctorate in 2007 in mechanical engineering from the University
of Oklahoma with Professor S. R. Gollahalli. He has degrees in mechanical engineering from the Indian Institute
of Technology Madras (India) and Anna University (India). He specializes in developing and testing cleaner and
energy-effcient oxy-combustion burners for glass, nonferrous, and steel industry applications. He has developed
test platforms for evaluating oxy-burner performance and conducted several customer feld trials. His research
interests include oxy-combustion, industrial furnaces, energy systems, burner testing, combustion diagnostics,
combustion in porous media, and oxygen safety. He is the author or co-author of over 25 peer-reviewed journal and
conference publications in combustion and energy related topics. He received the Outstanding Graduate Student
Award in 2007 for his porous media combustion research and undergraduate teaching activities.
Nate Petersen, PE, BS, MS, is currently a process engineer at John Zink Co., LLC (Tulsa, Oklahoma) where he has
been since 2005. He has degrees in chemistry and chemical engineering along with a degree in chemical engineer-
ing from the University of Utah (Salt Lake City, Utah). He has served in various engineering roles in the process
burner group, fare group, and thermal oxidizer group consisting of process and mechanical design and equipment
testing. He is a licensed professional engineer in the state of Oklahoma.
Erwin Platvoet, MSc, is the director of process burner engineering at John Zink Co., LLC (Tulsa, Oklahoma) where
he has been since 2009. He has a degree in chemical engineering from Twente University of Technology (Enschede,
The Netherlands). He was a cracking furnace specialist at Total Petrochemicals (Feluy, Belgium) from 2004 to 2009.
He was at ABB companies in The Netherlands, USA, and Switzerland from 1993 to 2009 and held a variety of
positions including thermal engineer, principal development engineer, and research and development engineer at
a variety of locations around the world. He worked at NRF Thermal Engineering (Uden, The Netherlands) from
1991 to 1993. He has authored a number of technical publications and has eight patents.
Roger L. Poe, BS, is a research associate at John Zink Co., LLC (Tulsa, Oklahoma) where he has worked since 1999. He
received a degree in mechanical engineering from Fairmont State University (Fairmont, West Virginia). Previously,
he served as manager of the Callidus Technologies (Tulsa, Oklahoma) Test and Research Center from 1995 to 1999
where he was responsible for the design and testing of specialty burners, as well as the development of new burner
equipment for the refnery and petrochemical industry. From 1989 to 1995 he managed the facilities and personnel
for the Penn State University Energy and Fuels Research Center (State College, Pennsylvania). He served as a work-
ing manager and researcher for the Donlee Technologies Research and Development Group (York, Pennsylvania)
from 1985 to 1989. During his career he has been involved with low NO
x
boiler burner technologies as they relate
to both liquid and gaseous fuels, coal gasifcation in pilot-scale, fuidized bed reactors, and the development and
testing of fuidized bed combustion units while working with Donlee Technologies. Further work, when on staff at
Penn State University, was done with fuidized bed combustion, coal gasifcation, micronized and pulverized coal
applications, coal slurry formulation and combustion, as well as low NO
x
gas and oil development. His most recent
work has been concentrated in low NO
x
process type burners and large-scale faring equipment while working
with both the John Zink Co. and Callidus Technologies. He has published more than 24 articles and holds numer-
ous patents relating to burners, fares, and pilots. His areas of interest are centered on fuid mechanics, combustion,
thermodynamics, combustion testing, and manufacturing. Over the course of his career he has jointly worked with
the Department of Energy, the Department of Defense, NORAD, the Institute of Gas Technology, the Gas Research
Institute, Sandia National Labs, and the Natick Naval Test labs.
Lee J. Rosen, DSc, is a senior manager of the combustion research and development group for Praxair, Inc.
(Tonawanda, New York) where he has been since 2003. He received his doctorate in mechanical engineering from
Washington University in St. Louis, Missouri. Dr. Rosen has 19 years of basic scientifc research work, industrial
technology research and development, and combustion design engineering. His experience includes oxyfuel
© 2011 by Taylor and Francis Group, LLC
xxii Contributors
combustion, fame stability, pulsed combustion, and fame synthesis of ultrafne particles. He holds three U.S.
patents and has authored several published articles and conference papers.
Silvio Rudy Stella, PhD, is the marketing director of Reway SrL (Possagno, Italy) that manufactures cogeneration
systems where he has been since 2006. He has a degree in electrical engineering from the Ministry of Scientifc
Research and Technology (Rome, Italy). He worked in a variety of roles at Riello Energy Group for Life from 2002
to 2005, at Thermital SpA from 2000 to 2002, at Calortecnica SpA from 1992 to 2000, at Gemmo SpA from 1991 to
1992 and at Padova Ricerche from 1989 to 1990. He has participated in a number of organizations and associations,
held offces in some of those organizations including president of the Italian cogeneration federated ANIMA called
ItalCogen, has given many presentations, has been an expert witness, and is an inventor on fve patents. He won a
prize in memory of Antonio Sarpi (University of Padua) as the best graduate of the Faculty of Engineering in 1989.
Allan M. Runstedtler, MASc, is a research scientist with CanmetENERGY, the energy research and technology
centre of Natural Resources Canada of the Canadian Government (Ottawa, Canada). He received his degree in
mechanical engineering from the University of Waterloo (Waterloo, Canada) in 2000. He has been the technical
lead on the investigation of industrial problems in refnery heaters and metal ore processing. Other work has led
to the development of a simple boiler model for process modeling of utility boilers and the conceptual design of
an ultra-low NO
x
burner for heat recovery from microturbines. He is interested in fundamental physical, chemi-
cal, and mathematical issues related to energy systems and has authored or co-authored journal papers on radiant
properties of combustion gases and diffusion in micro-scale pores. Among his current interests are a fuid dynamic
theory of turbulence and the use of density functional theory to study the relationship of material properties and
reactivity to atomic structure.
Khaled A. Sallam, PhD, is an associate professor at Oklahoma State University (Stillwater, Oklahoma) since 2003.
He received his degree in aerospace engineering from the University of Michigan (Ann Arbor, Michigan) where
he worked in the spray dynamics lab in 2002. From 2003 to 2009 he worked as an assistant professor of mechanical
and aerospace engineering at Oklahoma State University. In the summer of 2009, he was tenured and promoted
to associate professor. In 2008 he was selected as a summer faculty fellow for the Air Force Summer Fellowship
Program at Wright-Patterson Air Force Base. He was awarded the 2007 Halliburton Excellent Young Teacher Award
from Oklahoma State University and the 2006 W.R. Marshall Award from ILASS-Americas—Institute for Liquid
Atomization and Spray Systems, North and South America. He is a member of the American Society of Mechanical
Engineers, the American Institute of Aeronautics and Astronautics (and a member of AIAA Fluid Dynamics
Technical Committee), the American Physical Society, and the Institute for Liquid Atomization and Spray Systems.
He published 12 journal articles and 29 conference papers and supervised two PhD students and seven master
students.
Edwin Schorer, PhD, has been working as a consulting engineer in industrial acoustics for Müller-BBM GmbH
(Munich, Germany) since 1989. He received his degree in electrical engineering and his doctorate in psychoacous-
tics from the Technical University Munich, Germany. He was promoted to managing director in 2006. He works in
industrial acoustics in general, including theoretical and applied acoustics, with special focus on noise predictions
for fare noise and fan noise, fuid mechanics, ship acoustics, and acoustic optimization of postal automation sys-
tems. His research work resulted in 15 publications on psychoacoustics as well as industrial and technical acoustics.
Dr. Schorer is a member of the German Institute for Standardization, the Noise Control and Vibration Engineering
Standards Committee, and the German Acoustical Society. His research focuses on a functional schematic of just
noticeable frequency and amplitude variations. He worked as temporary academic counsel at his university, lec-
turing electroacoustics and technical acoustics. He acted as supervising tutor for the student’s diploma theses and
practical trainings.
Robert E. Schwartz, PE, BS, MS, is a senior technical specialist at John Zink Co., LLC (Tulsa, Oklahoma). He has
received his degrees in mechanical engineering from the University of Missouri. He has worked in the felds of com-
bustion, fares, pressure relieving systems, fuid fow, and heat transfer for more than 40 years including 42 years
with Zink where he has provided technical and business leadership in all product areas and has extensive interna-
tional experience. He has 51 U.S. patents for inventions of apparatus and methods that are in use throughout the John
Zink Co. He is the associate editor of The John Zink Combustion Handbook. His areas of technical expertise include:
development, design, fabrication, and operation of combustion equipment including fares, incinerators, process
© 2011 by Taylor and Francis Group, LLC
Contributors xxiii
burners, boiler burners, and vapor control; reduction of NO
x
and other emissions from combustion processes; fuid
fow and heat transfer in process and combustion equipment; noise elimination and control; vapor emissions con-
trol using recovery processes; hazardous waste site remediation; and permitting and operation of hazardous waste
storage and disposal sites. His professional organizations and awards include: member of the American Society of
Mechanical Engineers, the American Institute of Chemical Engineers and Sigma Xi, the Scientifc Research Society;
Registered Professional Engineer in the state of Oklahoma; recipient of the University of Missouri Honor Award for
Distinguished Service in Engineering and election to the University of Tulsa Engineering Hall of Fame.
Pavel Skryja, MS, currently works as a consultant for industrial power burners and combustion equipment. He
holds a degree in process engineering from Brno University of Technology. He has more than seven years of expe-
rience as a project manager and designer in research and development of industrial power burners designed for
refneries and special installations. He cooperates with Brno University of Technology (Czech Republic) on research
of renewable liquid fuels and liquid waste combustion.
Petr Stehlík is a professor and director of the Institute of Process and Environmental Engineering at Brno University
of Technology in the Czech Republic. He currently holds a position of vice president of the Czech Society of
Chemical Engineers. He has several years of experience in engineering industrial practice before joining the uni-
versity, and at present he is also a director of the research and development team for a certifed engineering and
contracting company with activities focusing on waste and biomass processing. Some of his main activities include:
executive editor of Heat Transfer Engineering and guest editor of international journals, coordinator or contractor of
international research projects, author or co-author of more than 200 papers in journals and proceedings, and ple-
nary or keynote speaker at various international conferences. His research and development as well as application
activities involve waste and biomass processing, waste to energy systems, applied heat transfer, energy saving, and
environmental protection.
Jun Sudo is a consultant for Nippon Furnace Co., Ltd. (Japan). He has 40 years of experience in combustion technol-
ogy in the felds of steel, petro-refnery, boiler, and cement industries. He has taken a leading role for the develop-
ment of regenerative combustion system in Nippon Furnace. He has numerous technical publications and several
patents. He co-received the Technical Award from the Combustion Society of Japan in 1995 for The Regenerative
Combustion System.
Dariusz Szewczyk, PhD, specializes in high temperature combustion and innovative methods of combustion
related to traditional fuels, biofuels, industrial, and waste gases. He received his degree at the Poznan University
of Technology (Poznan, Poland). His particular interests are in oxygen defcient combustion technologies centered
on the idea of lowering fuel consumption and pollutants emission. He worked at the Royal Institute of Technology
(Stockholm, Sweden) where he published, either as author or co-author, a number of papers concerning high tem-
perature air combustion as well as high temperature air gasifcation. From 2004 to 2007, Dr. Szewczyk worked for
VTS AB, a Swedish engineering company working in the feld of industrial combustion systems. Currently he is a
general manager and co-owner of ICS Industrial Combustion System Sp z o.o., Poland, an engineering company
working in the feld of industrial combustion systems. In VTS as well as in ICS he is responsible for projects utiliz-
ing NFK HRS/HTB/HiTAC technology in Europe.
Giuseppe Toniato, DrIng, is business innovation manager of Riello Group (Legnago, Italy) where he has been since
1999. He received his degree in mechanical engineering from the University of Padova (Italy) in 1991. He has been
director of engineering for Riello Burner Division for eight years. Prior to this he was projects manager in Magneti
Marelli Engine Control Division (Fiat Group). He has been engaged in combustion research for more than 14 years.
He holds ten patents and has authored or co- authored over ten publications on metallic mat combustion, catalytic
combustion, and burner controls.
Guido Troiani, PhD, is a researcher in combustion and fuid-mechanics at ENEA (Rome, Italy) where he has
been since 2006 as a postdoctorate. He received his degree in fuid mechanics from the Faculty of Engineering
of the University of Rome “La Sapienza” (2004). He began his postdoctoral research at the University of Rome
“La Sapienza” in cooperation with the Italian Ship Model Basin (INSEAN), performing experiments and theo-
retical analysis on free-surface turbulence and on the transition to turbulence of laminar fows. Successively he
was employed as postdoctorate at the ENEA research center in the feld of turbulent combustion. His main topics
© 2011 by Taylor and Francis Group, LLC
xxiv Contributors
of interest are the interactions between turbulence and combustion, fame chemiluminescence emissions, fractal
aspects of fame fronts and heavy particle dynamics from a theoretical and experimental point of view. He carried
out his experiments by means of intrusive devices such as Pitot and hot-wire probes, acoustic doppler anemom-
eters, and also spectroscopical techniques: LDA, PDA, PIV, and laser induced fuorescence. He is the author of fve
peer-reviewed publications and of 15 conference proceedings.
Joachim G. Wünning, DrIng, is the general manager and co-owner of WS GmbH (Germany). He received his
degree from the RWTH Aachen (Technical University of Aachen, Germany). His special interests include: combus-
tion research, burner development, heat recovery, high effciency, low emissions, fameless oxidation. He has numer-
ous technical publications and several patents.
Weihong Yang, PhD, is an associate professor in the Division of Energy and Furnace Technology, Royal Institute
of Technology, Sweden, where he has been since 2000. He has a degree in thermal energy from Central South
University of Technology (China). His major research areas include: high-temperature air combustion (fameless,
or MILD combustion), fameless oxyfuel combustion, gasifcation with high-temperature air/steam, improvement
of combustion systems in boilers and incinerators and rotary kiln in the process industries. He has published about
40 papers in international journals, and presented over 60 papers at international conferences. He also carried out
and managed many research projects fnanced by the Swedish nation, EU, and industries from Sweden, United
States, Japan, China, France, Poland.
Tsutomu Yasuda, BS, is auditor (Supervisory Board Member) of NFK-Holdings Co., Ltd., and auditor of Nippon
Furnace Co., Ltd (Yokohama, Japan; wholly owned subsidiary of above company). He received a degree of electric
and electronic engineering, from Tokyo Institute of Technology (Tokyo, Japan). He has been engaged as director
in research and development of high temperature air combustion technology since he joined the company in 1993.
He moved in October 1998 to Japan Industrial Furnace Manufactures Association, which is a nonproft organiza-
tion attached to the Japanese Ministry of Economy, Trade and Industries. He assumed director of Japan Industrial
Furnace in charge of HiTAC industrial furnaces and took responsibility of administrating the national project
funded by Japanese Ministry of Economy, Trade and Industries. In March 2000, he returned to Nippon Furnace
Kogyo Co. and assumed director of project promotion mainly related to high temperature air combustion. He has
20 patents and authored 15 papers all regarding HiTAC and high temperature steam gasifcation. He has presented
the papers in Germany, France, Italy, Sweden, Poland, Indonesia, Malaysia, Thailand, Vietnam, Taiwan, and P.R.
China.
Andrea Zambon, PhD, is an associate professor at the University of Padova (Padova, Italy) where he has been a
researcher since 1990 and as an associate professor in the Faculty of Engineering since 1998. He received his degree
in metallurgical engineering from the Polythecnic of Turin (Italy) in 1986. He has taught courses in metallic materi-
als, technology of the metallic materials, science and technology of composite materials, materials science and metal-
lurgy, and of materials selection and design. He is also a professor in the PhD course of mechatronics and industrial
systems of the PhD School in Industrial Engineering by the University of Padova. He is a member of the Steering
Committee at the university. He is a member of the Italian Association of Metallurgy (AIM-Milan) and of both the
Technical Committee of the Center for Physical Metallurgy and Materials Science and of the Technical Committee
of the Center for Welding and Permanent Joints of the Italian Association of Metallurgy. He has authored over 100
scientifc or technical publications on: development of mathematical models for the analysis and the forecast of ther-
mal felds in laser welding and processing of metallic materials; application of nonconventional techniques (plasma
torches, laser, high velocity oxygen-fuel) in the production of coatings and in the modifcation of the surfaces for the
improvement of the wear resistance; methods of mechanical characterization and study of the tribological behavior
of metal matrix composites; and powder metallurgy: processes of production by gas atomization, analysis of the
cooling regimes, production of bulk spray-formed and sintered samples, microstructural and mechanical character-
ization of the spray-formed or sintered products.
© 2011 by Taylor and Francis Group, LLC
I Section
General
© 2011 by Taylor and Francis Group, LLC
3
1
Introduction
Charles E. Baukal, Jr.
CONTENTS
1.1 Introduction ....................................................................................................................................................................... 4
1.2 Industrial Combustion Applications ............................................................................................................................. 8
1.2.1 Metals Production ................................................................................................................................................. 8
1.2.2 Minerals Production ............................................................................................................................................. 9
1.2.3 Chemicals Production ........................................................................................................................................ 10
1.2.4 Thermal Oxidation ............................................................................................................................................. 11
1.2.5 Industrial Boilers and Power Generation ........................................................................................................ 11
1.2.6 Drying .................................................................................................................................................................. 12
1.3 Combustion System Components ................................................................................................................................ 13
1.3.1 Burners ................................................................................................................................................................. 13
1.3.1.1 Burner Design Factors ......................................................................................................................... 13
1.3.1.2 General Burner Types .......................................................................................................................... 17
1.3.1.3 Burner Components ............................................................................................................................. 22
1.3.2 Combustors .......................................................................................................................................................... 23
1.3.2.1 Design Considerations ........................................................................................................................ 23
1.3.2.2 General Classifcations ........................................................................................................................ 23
1.3.3 Heat Load ............................................................................................................................................................. 25
1.3.3.1 Opaque Materials ................................................................................................................................. 26
1.3.3.2 Transparent Materials ......................................................................................................................... 26
1.3.4 Heat Recovery Devices ....................................................................................................................................... 26
1.3.4.1 Recuperators ......................................................................................................................................... 26
1.3.4.2 Regenerators ......................................................................................................................................... 27
1.4 Testing .............................................................................................................................................................................. 27
1.4.1 Purposes ............................................................................................................................................................... 27
1.4.2 Probe Types .......................................................................................................................................................... 29
1.4.3 Confgurations ..................................................................................................................................................... 30
1.4.3.1 Bench Scale Testing .............................................................................................................................. 30
1.4.3.2 Pilot Scale Testing ................................................................................................................................ 31
1.4.3.3 Large-Scale Testing ............................................................................................................................. 31
1.4.3.4 Field Testing .......................................................................................................................................... 31
1.4.4 Important Input Parameters .............................................................................................................................. 31
1.4.4.1 Fuel Composition ................................................................................................................................. 32
1.4.4.2 Air/Fuel Ratio ....................................................................................................................................... 32
1.4.4.3 Geometry ............................................................................................................................................... 32
1.4.4.4 Furnace Temperature and Pressure .................................................................................................. 32
1.5 Experimental Errors ....................................................................................................................................................... 33
1.5.1 Sources of Error ................................................................................................................................................... 33
1.5.2 Minimizing Errors .............................................................................................................................................. 33
1.5.3 Uncertainty Analysis .......................................................................................................................................... 35
1.6 Combustion Testing Resources ..................................................................................................................................... 36
1.6.1 General References ............................................................................................................................................. 36
© 2011 by Taylor and Francis Group, LLC
4 Industrial Combustion Testing
1.1 Introduction
Combustion systems are among the most challenging
technologies to study. There are not only high tempera-
tures, but usually very high temperature gradients rang-
ing from the incoming reactants at ambient temperature
up to fame temperatures. The fuid fow is typically tur-
bulent and may include swirl. The heat transfer includes
conduction, convection, and radiation. The radiation is
further complicated by the spectral nature of gaseous
combustion products. The chemistry is extremely com-
plicated, where the combustion of a relatively simple fuel
like methane can involve hundreds of chemical reac-
tions and dozens of species. Trace combustion products
such as carbon monoxide and nitrogen oxides are critical
because they are typically regulated. For liquid and solid
fuels, multiple phases are present. The fuel composition
can vary widely and may contain multiple components,
waste products, and sometimes multiple phases, depend-
ing on the process. The length scales in industrial com-
bustion processes may vary by orders of magnitude,
ranging from millimeters for fuel injection ports up to
meters for the combustor itself. The combustion system
may also include heat recuperation equipment such as air
preheaters and waste heat boilers, and pollution control
equipment such as scrubbers and catalytic treatment reac-
tors. The materials being heated may be solids, liquids, or
gases and have a wide range of properties. For example,
molten aluminum can be highly refective, molten glass is
spectrally absorptive, and cement is highly absorptive.
The feld of industrial combustion is very broad and
touches, directly or indirectly, nearly all aspects of our
lives. The electronic devices we use are generally pow-
ered by fossil fuel fred power plants. The cars we drive
use internal combustion engines. The planes we fy in
use jet fuel powered turbine engines. Most of the mate-
rials we use have been made through some type of heat-
ing process. While this book is concerned specifcally
with industrial combustion, all of the above combustion
processes have many features in common.
Industrial combustion is complicated by many fac-
tors. First, the science of combustion is still developing
and has a long way to go until we completely under-
stand it so it can be better applied and controlled. While
fre has been with us since the beginning of time, much
remains to be learned about it. As the science of combus-
tion combines heat transfer, thermodynamics, chemical
kinetics, and multiphase turbulent fuid fow to name a
few areas of physics, the study of industrial combustion
is interdisciplinary by necessity.
Combustion has been the foundation of world-
wide industrial development for the past 200 years [1].
Industry relies heavily on the combustion process as
shown in Table 1.1. The major uses for combustion in
industry are shown in Table 1.2. Hewitt et al. (1994) have
listed some of the common heating applications used
in industry, as shown in Table 1.3 [2]. As can be seen in
Figure 1.1, the worldwide demand for energy contin-
ues to increase. Most of the energy (86%) is produced
by the combustion of fossil fuels like petroleum, natu-
ral gas, and coal (see Figure 1.2). According to the U.S.
Department of Energy, the demand in the industrial
sector is projected to increase by 0.8% per year to the
year 2020 [3]. Figure 1.3 shows that the industrial sec-
tor is one of the largest energy consumers in the United
States. Figure 1.4 shows the projected energy source and
end use for the United States in 2008. This again high-
lights the importance of industrial combustion.
The combustion community has identifed a number
of driving forces that will shape industrial combus-
tion for the foreseeable future: markets and economics;
environmental quality and greenhouse gases; process
improvement; policy and politics; fuel/oxidant choices;
energy effciency; enabling technologies; health, safety,
and reliability; and research and education [1]. Many
of these are complex issues that dynamically change
over time. For example, pollution regulations vary by
1.6.2 Test Facilities ........................................................................................................................................................ 37
1.6.2.1 Academic ............................................................................................................................................... 37
1.6.2.2 Institutional ........................................................................................................................................... 38
1.7 Future ............................................................................................................................................................................... 38
References .................................................................................................................................................................................. 38
Table 1.1
The Importance of Combustion to Industry
% Total Energy from (at the
Point of Use)
Industry Steam Heat Combustion
Petroleum refning 29.6 62.6 92.2
Forest products 84.4 6.0 90.4
Steel 22.6 67.0 89.6
Chemicals 49.9 32.7 82.6
Glass 4.8 75.2 80.0
Metal casting 2.4 67.2 69.6
Aluminum 1.3 17.6 18.9
Source: U.S. Department of Energy (DOE). Industrial
Combustion Vision: A Vision by and for the
Industrial Combustion Community. Washington,
DC: U.S. DOE, 1998.
© 2011 by Taylor and Francis Group, LLC
Introduction 5
location and continue to get more rigorous. Technology
continues to improve as the emission requirements get
more stringent.
The U.S. Department of Energy sponsored a workshop
to develop a roadmap for industrial combustion tech-
nology. The resulting report [4] identifed the following
priority research and development (R&D) needs:
Through a system approach, create burner •
designs that effectively transfer heat to the load.
Improve heat recovery processes to capture •
waste gases cost-effectively.
Create innovative new furnace designs through •
better fundamental understanding of combus-
tion and scale-up.
Improve boiler technology and combustion •
cycles through fundamental design innovations
such as gasifcation and fuidized beds.
Determine advanced methods to maintain a •
stable fame and achieve low emissions while
using different fuels.
Develop a burner capable of adjusting operat- •
ing parameters in real time.
Develop real-time sensors and process con- •
trols that are more reliable and robust in harsh
environments.
Develop computational tools that are more accu- •
rate in a wide variety of applications through the
collection of physical data and model validation.
Table 1.2
Major Process Heating Operations
Metal melting
Steel making •
Iron and steel melting •
Nonferrous melting •
Metal heating
Steel soaking, reheat, ladle preheating •
Forging •
Nonferrous heating •
Metal heat treating
Annealing •
Stress relief •
Tempering •
Solution heat treating •
Aging •
Precipitation hardening •
Curing and forming
Glass annealing, tempering, forming •
Plastics fabrication •
Gypsum production •
Fluid heating
Oil and natural gas production •
Chemical/petroleum feedstock •
preheating
Distillation, visbreaking, hydrotreating, •
hydrocracking, delayed coking
Bonding
Sintering, brazing •
Drying
Surface flm drying •
Rubber, plastic, wood, glass products drying •
Coal drying •
Food processing •
Animal food processing •
Calcining
Cement, lime, soda ash •
Alumina, gypsum •
Clay fring
Structural products •
Refractories •
Agglomeration
Iron, lead, zinc •
Smelting
Iron, copper, lead •
Nonmetallic materials melting
Glass •
Other heating
Ore roasting •
Textile manufacturing •
Food production •
Aluminum anode baking •
Source: U.S. Department of Energy (DOE). Industrial Combustion Vision: A Vision by and for the
Industrial Combustion Community. Washington, DC: U.S. DOE, 1998.
Table 1.3
Examples of Processes in the Process Industries Requiring Industrial Combustion
Process Industry Examples of Processes Using Heat
Steel making Smelting of ores, melting, annealing
Chemicals Chemical reactions, pyrolysis, drying
Nonmetallic minerals (bricks, glass, cement, and other refractories) Firing, kilning, drying, calcining, melting, forming
Metal manufacture (iron and steel, and nonferrous metals) Blast furnaces and cupolas, soaking and heat treatment, melting,
sintering, annealing
Paper and printing Drying
Source: Adapted from U.S. Department of Energy, Energy Information Administration. Annual Energy Outlook 2008, report DOE/EIA-0384,
2008, Washington, DC, released June 26, 2009.
© 2011 by Taylor and Francis Group, LLC
6 Industrial Combustion Testing
500
450
400
350
300
250
200
150
100
Q
u
a
d
r
i
l
l
i
o
n

B
t
u
50
0
1
9
8
0
1
9
8
2
1
9
8
4
1
9
8
6
1
9
8
8
1
9
9
0
1
9
9
2
1
9
9
4
1
9
9
6
1
9
9
8
Year
Other
Nuclear electric power
Hydroelectric power
Coal
Natural gas
Petroleum
2
0
0
0
2
0
0
2
2
0
0
4
2
0
0
6
Figure 1.1
Historical and projected (2008) world energy consumption. (Energy Information Administration).
Natural gas, 23%
Coal, 27%
Hydroelectric
power, 6%
Nuclear electric
power, 6%
Other, 2%
Petroleum, 36%
Figure 1.2
2006 world energy consumption. (Energy Information Administration).
120
100
80
Q
u
a
d
r
i
l
l
i
o
n

B
t
u
60
40
20
0
1
9
4
9
1
9
5
2
1
9
5
5
1
9
5
8
1
9
6
1
1
9
6
4
1
9
6
7
1
9
7
0
1
9
7
3
1
9
7
6
1
9
7
9
1
9
8
2
1
9
8
5
1
9
8
8
1
9
9
1
1
9
9
4
1
9
9
7
2
0
0
0
2
0
0
3
2
0
0
6
2
0
0
9
Year
Transportation
Industrial
Commercial
Residential
Figure 1.3
U.S. energy consumption by industry sector. (Energy Information Administration).
© 2011 by Taylor and Francis Group, LLC
Introduction 7
Create a pathway to demonstrate and commer- •
cialize new technologies and enable informa-
tion sharing throughout industry about new
technologies.
Develop robust design tools that are more user- •
friendly and accurate, especially with complex
phenomena such as turbulence and create a
unifed code to allow sharing of information
more easily and speed development.
Improve integration of boiler systems with the •
rest of the plant process with new “smart” con-
trol systems and designs that capture unused
waste heat.
Table 1.4 lists the priority R&D needs that were identi-
fed for burners, boilers, and furnaces. Essentially all of
these needs require some amount of testing. These needs
were generated from the following end-use require-
ments: increased system effciency; reduced NOx, CO,
CO
2
, and particulate emissions; increased fuel fexibility;
more robust and fexible process control and operations;
better safety, reliability, and maintenance; lower capital
and operational costs; faster, low-cost technology devel-
opment; and enhanced system integration. Coupled with
these needs are some barriers to improvement: fnancial
risk, inability to accurately predict the performance of new
systems, lack of industry standards, and the wide gap that
often exists between the research done at a small scale that
needs to be applied to industrial-scale systems. Testing is
often required to address some of these barriers.
As shown in Figure 1.5, three elements are required
to sustain combustion processes: fuel, oxidizer, and an
ignition source usually in the form of heat. Industrial
combustion is defned here as the rapid oxidation
of hydrocarbon fuels to generate large quantities of
energy for use in industrial heating and melting pro-
cesses. Industrial fuels may be solids (e.g., coal), liquids
(e.g., oil), or gases (e.g., natural gas). The fuels are com-
monly oxidized by atmospheric air (which is approx-
imately 21% O
2
by volume) although it is possible in
certain applications to have an oxidizer (sometime
referred to as an “oxidant” or “comburent”) contain-
ing less than 21% O
2
(e.g., turbine exhaust gas [TEG]
[5]) or more than 21% O
2
(e.g., oxy/fuel combustion [6]).
The fuel and oxidizer are typically mixed in a device
referred to as a burner that is discussed in more detail
below. An industrial heating process may have one or
many burners depending on the specifc application
and heating requirements.
Many theoretical books have been written on the
subject of combustion, but they have little if any discus-
sion of industrial combustion processes [7–12]. Edwards
(1974) has a brief chapter on applications including both
stationary (boilers and incinerators primarily) and
mobile sources (primarily internal combustion engines)
[13]. Barnard and Bradley (1985) have a brief chapter on
industrial applications, but have little on pollution from
those processes [14]. A book by Turns (2000), which is
designed for undergraduate and graduate combustion
courses, contains more discussions of practical combus-
tion equipment than most similar books [15].
There have also been many books written on the more
practical aspects of combustion. Griswold’s (1946) book
has a substantial treatment of the theory of combustion,
Coal
23.86
Natural gas
21.15
Renewable energy
3
7.32
Petroleum
4
27.56
Imports
32.84
Other
imports
5

5.28
Domestic
production
73.71
Supply
106.55
Exports 7.06
Coal
22.42
Petroleum
9
37.14
Nuclear electric power 8.46
Industrial
12
31.21
Transportation
27.92
Residential
12
21.64
Commercial
12
18.54
Renewable energy
3
7.30
Fossil
fuels
10
83.44
Consumption
11
99.30
Natural gas
8
23.84
Stock change
and other
6

0.18
Petroleum
3.77
Other exports
7
3.30
Fossil fuels
57.94
NGPL
2
2.41
Nuclear electric power
8.46
Crude oil
1
10.52
Figure 1.4
2008 projected energy fow (quadrillion Btu) by source and end use in the United States. (Energy Information Administration.)
© 2011 by Taylor and Francis Group, LLC
8 Industrial Combustion Testing
but is also very practically oriented and includes chap-
ters on gas burners, oil burners, stokers and pulverized
coal burners, heat transfer, furnace refractories, tube
heaters, process furnaces, and kilns [16]. Stambuleanu’s
(1976) book on industrial combustion has information on
actual furnaces and on aerospace applications, particu-
larly rockets [17]. There is much data in the book on fame
lengths, fame shapes, velocity profles, species concen-
trations, liquid, and solid fuel combustion. A book on
industrial combustion has signifcant discussions on
fame chemistry, although little on pollution from fames
[18]. A book by Borman and Ragland (1998) attempts to
bridge the gap between the theoretical and practical
books on combustion [19]. Deshmukh (2005) discusses
a range of topics of interest in industrial combustion
including fuels, burners, and refractories, where the
emphasis is more on metal treatment [20]. However, the
book has little discussion about the types of industrial
applications considered here. Keating’s (2007) book on
applied combustion is aimed at engines and has no treat-
ment of industrial combustion processes [21]. Mullinger
and Jenkins (2008) have written a very practical book
that discusses a wide range of industrial furnaces and
processes [22]. Surprisingly, many handbooks on com-
bustion applications have little if anything on industrial
combustion systems [23–27].
1.2 Industrial Combustion Applications
Some of the most common industrial applications are
briefy discussed next. Many of them are discussed in
more detail in other chapters in this book.
1.2.1 Metals Production
Metals are used in nearly all aspects of our lives and
play a very important role in society. The use of metals
Table 1.4
Priority Research and Development Needs
Burners
Burners capable of adjusting operating parameters in real time
Sensors and controls for emissions and heat distribution
Fuel and oxidant physical and chemical characteristics
Fuel and oxidant mixing within the burner
Process interactions within the burner system
Robust design tools/unifed code
Fundamental understanding of physical and chemical
science
Burner system design models (using fundamentals)
Performance targets for models
More effective heat transfer to load
Radiative heat transfer from the fame to the load
Heat delivery to the load
Physical modeling
Sensors and controls
Advanced combustion stabilization methods
Use of multiple fuels and their characteristics
New fame stabilization techniques
Boilers
New boiler technologies
Sensors and controls
Alternatives to burners for solid and liquid fuels
Fuel and oxidant fexibility
Boiler system effciency improvement
System integration
Sensors and controls
Application of combined heat and power systems (CHP)
Condensate systems integration
Boiler personnel training
Technology transfer
Test and demonstration programs
Speed of the commercialization process
Furnaces
New furnace designs
Fundamental knowledge of physical and chemical science
Materials of construction
Model validation and technology transfer
End user product quality
Process characterization and development
Advanced sensors and process control
Best practices (from other industries)
Advanced materials of construction
Application of sensors and controls
Cost-effective heat recovery
Design, application, and performance of heat recovery systems
Advanced materials of construction
Flue gas characterization
Integrated computational design tools
Design tool characteristics
Technology transfer
Source: Energetics, Inc. Industrial Combustion Technology Roadmap: A Technology Roadmap by and for the Industrial Combustion
Community, October 2002, www1.eere.energy.gov/industry/combustion/pdfs/combustion_roadmap2002.pdf, 2002.
Fuel Oxygen
Source of ignition
Figure 1.5
Combustion triangle. (From Baukal, C. E., ed., Oxygen-Enhanced
Combustion, Boca Raton, FL: CRC Press LLC, 1998; courtesy of CRC
Press.)
© 2011 by Taylor and Francis Group, LLC
Introduction 9
has been around for thousands of years. There are two
predominant classifcations of metals: ferrous (iron-
bearing) and nonferrous (e.g., aluminum, copper, and
lead). Ferrous metal production is often high tempera-
ture because of higher metal melting points compared
to nonferrous metals. Many metals production processes
are done in batch, compared to most other industrial
combustion processes considered here (e.g., glass pro-
duction) that are typically continuous. Another fairly
unique aspect of metal production is the very high use
of recycled materials. This often lends itself to batch pro-
duction because of the somewhat unknown composition
of the incoming scrap materials that may contain trace
impurities that could be very detrimental to the fnal
product if not removed. The metals are typically melted
in some type of vessel and then sampled to determine
the chemistry so that the appropriate chemicals can be
either added or removed to achieve the desired grade of
material. Another unique aspect of the metals industry
is that transfer vessels are preheated prior to the intro-
duction of molten metal into the vessel to minimize
the thermal shock to the refractory. Figure 1.6 shows
an example of preheating a transfer ladle used to move
molten metal around a plant.
Since metals melt at higher temperatures, higher
intensity burners are often used in these applications.
This includes, for example, oxygen-enhanced combus-
tion (OEC) [28,29] (see Chapter 27) and air preheating
(see Chapter 21) to increase the fame temperatures and
metal melting capability. These higher intensity burners
have the potential to produce high pollutant emissions
so the burner design is important to minimize these
emissions.
Another somewhat unique aspect of metals produc-
tion is that supplemental heating may be required to
reheat the metals for further processing. For example,
ingots may be produced in one location and then trans-
ported to another location to be made into the desired
shape (e.g., wheel castings are often made from remelt-
ing aluminum ingots or sows). While this process may
be economically effcient, it is energy and pollutant inef-
fcient due to the additional heating. Burners are used in
the original melting process as well as in the reheating
process (see Chapter 9). This is something that has begun
to attract more attention in recent years where the entire
life cycle of a product is considered rather than just its
unit cost and initial energy requirements. For example,
aluminum has a low life cycle cost compared to many
other metals because of its high recycle ratio. While the
energy consumption to make aluminum from raw ore is
fairly high, remelting scrap aluminum takes only a frac-
tion of that energy that also means less overall pollution
as well. Burners commonly used in the metals indus-
try include high velocity burners, regenerative burners
(Chapter 21), radiant tube burners (Chapter 24), air-oxy/
fuel burners and oxy/fuel burners (Chapter 27).
1.2.2 Minerals Production
Some common minerals processes include the
production of glass, cement, bricks, refractories, and
ceramics. These are typically high temperature heat-
ing and melting applications that require a signifcant
amount of energy per unit of production. They also tend
to have fairly high pollutant emissions as a result of the
high temperatures and unit energy requirements. Most
of the minerals applications are continuous processes,
but there is a wide range of combustors. Large glass
furnaces are typically rectangularly shaped and have
multiple burners (see Chapter 32). On the other hand,
cement kilns are long refractory-lined rotating cylinders
that are slightly inclined so that the materials fow grad-
ually down hill (see Figure 1.7; see Chapter 31). A typical
cement plant is shown in Figure 1.8.
Many of the minerals applications employ some type
of heat recovery in the form of air preheating to improve
Firewall
Ladle
Ladle seal
Figure 1.6
Ladle preheater. (From Baukal, C. E., ed., Industrial Burners Handbook,
Boca Raton, FL: CRC Press, 2004.)
To off gas
treatment
Feed
materials
Processed
clinker
Burner
Figure 1.7
Countercurrent rotary cement kiln schematic. (From Baukal, C. E.,
ed., Industrial Burners Handbook, Boca Raton, FL: CRC Press, 2004.)
© 2011 by Taylor and Francis Group, LLC
10 Industrial Combustion Testing
energy effciency. However, the heat recovery typically
signifcantly increases NOx emissions. While recycling
of used glass (referred to as cullet) is practiced in some
applications, there is generally must less recycling in the
minerals industry compared to the metals industry.
1.2.3 Chemicals Production
This is a very broad classifcation that encompasses
many different types of production processes that have
been loosely subcategorized into chemicals (organic
and inorganic) and petrochemicals (organic) applica-
tions. A typical refnery is shown in Figure 1.9. There is
some overlap in terms of the types of heating equipment
used where many of the incoming feed materials are in
liquid form (e.g., crude oil) that are processed in heat-
ers with tubes running inside them. These are generally
lower temperature applications (< 2300°F or < 1300°C)
that incorporate heat recovery to preheat the incom-
ing feed materials. Nearly all of the chemicals heating
applications employ multiple burners but in much more
diverse confgurations compared to many other indus-
tries. Burners may be fred horizontally, vertically up,
vertically down, or at some angles in- between depend-
ing on the specifc process (see Chapter 18). There are
numerous confgurations for fred process heaters (a
typical example is shown in Figure 1.10).
There are some aspects that make this industry unique
compared to others. The frst and one of the most impor-
tant is the wide range of fuel compositions used to fre
the heaters. These are mostly gaseous fuels that are by-
products of the production process. These gaseous fuels
often contain signifcant quantities of hydrogen, meth-
ane, and propane and may include large quantities of
inert gases such as nitrogen and carbon dioxide. A given
heater may need to be able to fre on multiple fuels that
may be present during various times in the production
process. Another unique aspect of this industry is that
many of the heaters are fred with natural draft burners
(see Chapter 18) where no blower is used to supply the
Figure 1.8
Cement plant. (From Baukal, C. E., ed., Industrial Burners Handbook, Boca Raton, FL: CRC Press, 2004.)
Figure 1.9
Refnery. (From Baukal, C. E., ed., Industrial Burners Handbook, Boca Raton, FL: CRC Press, 2004.)
© 2011 by Taylor and Francis Group, LLC
Introduction 11
combustion air. These burners are designed differently
than conventional forced draft burners and are more
susceptible to variations in ambient conditions such as
air temperature, humidity, and wind speed. A specifc
type of burner used in petrochemical applications is a
radiant wall burner (see Chapter 18).
1.2.4 Thermal Oxidation
The objective of thermal oxidation (also known as waste
incineration) processes is to reduce or eliminate waste
products that involves combusting those materials
(see Chapter 33). Not only is the thermal oxidizer fred
with burners, but the waste material itself is often part
of the fuel that generates heat in the process. However,
the waste usually has a very low heating value, hence
the need for supplemental fuel (e.g., Figure 1.11).
Thermal oxidation is a more complicated and dynamic
process compared to most other industrial combustion
processes by nature of the variability of the feed mate-
rial. Figure 1.12 shows a schematic of a municipal solid
waste incinerator. The waste may be very wet after a
rain storm that may put a huge extra heat load on the
incinerator. In some locations where waste materials are
separated for recycling, the waste actually fed into the
thermal oxidizer may have a much higher heating value
compared to other incinerators where there is no sepa-
ration of the waste.
A complicating factor with incinerators is that the end
product, for example the noncombustible waste, must
also be disposed of which means that one of the goals
of most incineration processes is to produce minimal
waste output. Because of the waste material variability,
other pollutants may be generated that are not normally
associated with industrial combustion processes. An
example is the burning of plastics that can produce
dioxins and furans. The types of thermal oxidizers can
vary greatly depending upon a variety of factors. In
some cases, waste materials to be destroyed may be fed
through the burners. This is particularly true of waste
hydrocarbon liquids. Some of the burners used in incin-
eration include air-oxy/fuel and oxy/fuel (Chapter 27).
1.2.5 industrial boilers and Power generation
Boilers are used for a variety of purposes in an assort-
ment of applications. Common uses include producing
hot water or steam for heating, producing steam for use
within a plant such as atomizing oil for oil-fred burn-
ers, and producing steam to generate power in large
power plants (see Chapter 34). Applications range from
small single-burner uses in hospitals, schools, and small
businesses up to large multiburner boilers in power
plants. The burners used in boilers are typically regu-
lated because of their proliferation and widespread use
in applications involving the general public. The burn-
ers are normally required to have a full complement of
safety controls to ensure safe operation. These burners
are often highly regulated to minimize pollutant emis-
sions, particularly in large power plants because of the
size of the source. Boiler burners used in larger appli-
cations requiring multiple burners are considered in
Chapter 20.
A special category of burners sometimes used in large
power generating plants with gas turbines are called
duct burners (see Figure 1.13). A schematic of the typical
Stack
Damper
Convection
section
Radiant
section
Burner
Heat
loss
Figure 1.10
Process refnery heater. (From Baukal, C. E., ed., The John Zink
Combustion Handbook, Boca Raton, FL: CRC Press, 2001; courtesy of
John Zink Co. LLC.)
Figure 1.11
Municipal waste incinerator. (From Baukal, C. E., ed., Industrial
Burners Handbook, Boca Raton, FL: CRC Press, 2004.)
© 2011 by Taylor and Francis Group, LLC
12 Industrial Combustion Testing
location of duct burners downstream of the turbine is
shown in Figure 1.14. These burners are unique because
they use the combustion products from a turbine as
their combustion “air.” The TEG is at an elevated tem-
perature and contains signifcant quantities of carbon
dioxide and water that are the products of the upstream
combustion process. The TEG is also at a depleted oxy-
gen level (< 21%) so duct burners are designed to oper-
ate under these conditions. They are treated in detail in
Chapter 26.
1.2.6 Drying
Burners are used in a wide variety of lower tem-
perature drying applications to remove water from
products that was added during the manufacturing
process. These are lower temperature applications
that include paper manufacturing, printing and pub-
lishing, textile manufacturing, and food processing.
Drying is defned as “a process in which a wet solid
is heated or contacted with a hot gas stream, causing
some or all of the liquid wetting the solid to evapo-
rate” [30]. Kudra and Mujumdar (2002) have written a
book [31] on advanced drying technologies that cover a
wide range of industries. One example is the drying of
paper produced from a wet slurry. A typical paper mill
is shown in Figure 1.15.
In many drying processes, moisture is removed from
webs that may be traveling at high speeds. Radiant
heating is often used to supplement steam-heated
Feed
hopper
Moving
grate
Ash and
metal
Combustion air
To steam generator
and bag house
Figure 1.12
Schematic of municipal solid waste incinerator. (From Baukal, C. E., ed., Industrial Burners Handbook, Boca Raton, FL: CRC Press, 2004.)
Figure 1.13
(See color insert following page 424.) Duct burner fame. (From
Baukal, C. E., ed., The John Zink Combustion Handbook, Boca Raton, FL:
CRC Press, 2001; courtesy of John Zink Co. LLC.)
Steam generator
Stack
Steam drums
Duct burner
Gas turbine
Figure 1.14
Duct burner process schematic. (From Baukal, C. E., ed., The John Zink
Combustion Handbook, Boca Raton, FL: CRC Press, 2001; courtesy of
John Zink Co. LLC.)
© 2011 by Taylor and Francis Group, LLC
Introduction 13
cylinders or high velocity hot air dryers [32]. The
radiant heaters are either electric or fred with a fuel
gas such as natural gas. Thermal radiation burners
used in many of these applications are discussed in
Chapter 25.
1.3 Combustion System Components
There are six components that may be important in
industrial combustion processes (see Figure 1.16). One
component is the burner that combusts the fuel with an
oxidizer to release heat. Another component is the load
itself that can greatly affect how the heat is transferred
from the fame. In most cases, the fame and the load are
located inside of a combustor, which may be a furnace,
heater, dryer, or kiln that is the third component in the
system. In some cases, there may be some type of heat
recovery device to increase the thermal effciency of the
overall combustion system, which is the fourth com-
ponent of the system. The ffth component is the fow
control system used to meter the fuel and the oxidant to
the burners. The sixth and last component is the air pol-
lution control system used to minimize the pollutants
emitted from the exhaust stack into the atmosphere. The
frst four system components are considered next.
1.3.1 burners
1.3.1.1 Burner Design Factors
The burner is the device that is used to combust the fuel
with an oxidizer to convert the chemical energy in the
fuel into thermal energy. A given combustion system
may have a single burner or many burners, depending
Figure 1.15
Paper mill. (From Baukal, C. E., ed., Industrial Burners Handbook, Boca
Raton, FL: CRC Press, 2004.)
Load
Furnace
Burner
Fuel
Flow control
system
Combustion
air blower
Pollution control
system
To atmosphere
Heat exchanger
Flue gases
Combustion air
Exhaust
fan
Figure 1.16
Schematic of an industrial combustion process. (From Baukal, C. E., ed., Industrial Burners Handbook, Boca Raton, FL: CRC Press, 2004.)
© 2011 by Taylor and Francis Group, LLC
14 Industrial Combustion Testing
on the size and type of application. For example, in a
rotary kiln a single burner is located in the center of
the wall on one end of a cylindrically shaped furnace
(see Figure 1.17). The heat from the burner radiates in
all directions and is effciently absorbed by the load.
However, the cylindrical geometry has some limitations
concerning size and load type that makes its use limited
to certain applications such as melting scrap aluminum
or producing cement clinker (see Chapter 31). A more
common combustion system has multiple burners in a
rectangular geometry (see Figure 1.18). This type of sys-
tem is generally more diffcult to simulate with testing
because of the multiplicity of heat sources and because
of the interactions between the fames and their associ-
ated products of combustion.
There are many factors that go into the design of a
burner. This section briefy considers some of the
important factors that are taken into account for a par-
ticular type of burner. These factors affect things like
heat transfer [32] and pollutant emissions [46]. There
have many changes in the traditional designs that
have been used in burners, primarily because of the
continued interest in reducing pollutant emissions. In
the past, the burner designer was primarily concerned
with effciently combusting the fuel and transferring
the energy to a heat load. New and increasingly more
stringent environmental regulations have added the
requirement to consider the pollutant emissions pro-
duced by the burner. In many cases, reducing pollutant
emissions and maximizing combustion effciency are
at odds with each other. For example, a well-accepted
technique for reducing NOx emissions is known as stag-
ing (discussed below), where the primary fame zone is
defcient of either fuel or oxidizer [33]. The balance of
the fuel or oxidizer may be injected into the burner in
a secondary fame zone or, in a more extreme case, may
be injected somewhere else in the combustion chamber.
Staging reduces the peak temperatures in the primary
fame zone and also alters the chemistry in a way that
reduces NOx emissions because fuel rich or fuel lean
zones are less conducive to NOx formation than near
stoichiometric zones. Figure 1.19 shows how the NOx
emissions are affected by the exhaust product tempera-
ture. Since thermal NOx is exponentially dependent on
the gas temperature, even small reductions in the peak
fame temperature can dramatically reduce NOx emis-
sions. However, lower fame temperatures often reduce
the radiant heat transfer from the fame since radiation
is dependent on the fourth power of the absolute tem-
perature of the gases. Another potential problem with
staging is that it may increase CO emissions, which is
an indication of incomplete combustion and reduced
combustion effciency. However, it is also possible that
staged combustion may produce soot in the fame that
Rotary
furnace
Air
Burner Fuel
Figure 1.17
Rotary kiln with single burner. (From Baukal, C. E., Heat Transfer in
Industrial Combustion, Boca Raton, FL: CRC Press, 2000; courtesy of
CRC Press.)
Burner firing Burner firing Burner firing
Batch
Burner off Burner off Burner off
Glass
Flue
Figure 1.18
Plan view of multiple burners in a glass furnace. (From Baukal, C.
E., Heat Transfer in Industrial Combustion, Boca Raton, FL: CRC Press,
2000; courtesy of CRC Press.)
500
0 400 800 1200 1600
Gas temperature (°F)
Gas temperature (K)
S
p
e
c
i
e
s

c
o
n
c
e
n
t
r
a
t
i
o
n

(
p
p
m
v
w
)
S
p
e
c
i
e
s

c
o
n
c
e
n
t
r
a
t
i
o
n

(
p
p
m
v
w
)
2000 2400 2800 3200 3600
1000 1500 2000
200
180
160
140
120
100
80
60
40
20
0
200
180
160
140
120
100
80
60
40
20
0
Figure 1.19
NOx as a function of gas temperature. (From Baukal, C. E., Heat
Transfer in Industrial Combustion, Boca Raton, FL: CRC Press, 2000.
courtesy of CRC Press.)
© 2011 by Taylor and Francis Group, LLC
Introduction 15
can increase fame radiation. The actual impact of stag-
ing on the heat transfer from the fame is highly depen-
dent on the actual burner design.
In the past, the challenge for the burner designer was
often to maximize the mixing between the fuel and the
oxidizer to ensure complete combustion, especially if the
fuel was diffcult to burn, as in the case of low heating
value fuels such as waste liquid fuels or process gases
from chemicals production. Now the burner designer
must balance the mixing of the fuel and the oxidizer to
maximize combustion effciency while simultaneously
minimizing all types of pollutant emissions. This is no
easy task as, for example, NOx and CO emissions often
go in opposite directions as shown in Figure 1.20. When
CO is low, NOx may be high and vice versa. Modern
burners must be environmentally friendly, while simul-
taneously effciently transferring heat to the load.
There are many types of burner designs that exist due
to the wide variety of fuels, oxidizers, combustion cham-
ber geometries, environmental regulations, thermal
input sizes, and heat transfer requirements that includes
things like fame temperature, fame momentum, and
heat distribution. Some of these design factors are briefy
considered here. Other important design factors such as
heat fux (Chapter 6) and noise (Chapter 8) are discussed
elsewhere in the book. Some of the tools used to optimize
a burner design include computational fuid dynamic
(CFD) modeling (Chapter 11), testing (the subject of this
book), and physical modeling (Chapter 10).
1.3.1.1.1 Fuel
Depending upon many factors, certain types of fuels
may be preferred for certain geographic locations due
to cost and availability considerations. Gaseous fuels,
particularly natural gas, are commonly used in most
industrial heating applications in the United States and
Europe. In Asia and South America, heavy fuel oils are
generally preferred although the use of gaseous fuels is
on the rise. Fuels also vary depending on the applica-
tion. For example, in incineration processes, waste fuels
are commonly used either by themselves or with other
supplemental fuels like natural gas. In the petrochemi-
cal industry, fuel gases often consist of a blend of several
components, including gases like hydrogen, methane,
propane, butane, and propylene.
The fuel choice has an important infuence on the heat
transfer from a fame. In general, solid fuels like coal
and liquid fuels like oil produce very luminous fames
that contain soot particles that radiate like blackbodies
to the heat load. Gaseous fuels like natural gas often pro-
duce nonluminous fames because they burn so cleanly
and completely without producing soot particles. A
fuel like hydrogen is completely nonluminous as there
is no carbon available to produce soot. In cases where
highly radiant fames are required, a luminous fame
is preferred. In cases where convection heat transfer is
preferred, a nonluminous fame may be preferred to
minimize the possibility of contaminating the heat load
with soot particles from a luminous fame. Where natu-
ral gas is the preferred fuel and highly radiant fames
are desired, new technologies are being developed to
produce more luminous fames. These include things
like pyrolyzing the fuel in a partial oxidation process
[34], using a plasma to produce soot in the fuel [35], and
generally controlling the mixing of the fuel and oxi-
dizer to produce fuel rich fame zones that generate soot
particles [36]. Therefore, the fuel itself has a signifcant
impact on the heat transfer mechanisms between the
fame and the load. In most cases, the fuel choice is dic-
tated by the customer as part of the specifcations for the
system and is not chosen by the burner designer. The
designer must make the best of whatever fuel has been
selected. In most cases, the burner design is optimized
based on the choice for the fuel.
The fuel also has a large impact on the pollutant emis-
sions. For example, gaseous fuels generally contain
little or no sulfur so SOx emissions are usually small.
However, heavy oils often contain signifcant quantities
of sulfur and therefore SOx emissions are of concern
and need to be controlled. Another example is particu-
late emissions. Gaseous fuels generally burn very cleanly
and produce negligible particulates. However, heavy
liquid oil fuels can generate high levels of particulate
emissions. Therefore, the burner design is important in
minimizing pollutant emissions depending on the fuel.
In some cases, the burner may have more than one
type of fuel. An example of a duel fuel burner is shown
in Figure 1.21 [37]. Dual-fuel burners are designed
to operate typically on either gaseous or liquid fuels.
Equivalence ratio
0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
0.5
13
12
11
10
9
8
7
6
5
4
3
2
1
0
3500
3000
2500
2000
1500
1000
500
0
1.0 1.5 2.0 2.5 3.0 3.5
CO (vol. %)
NO (ppmvw)
4.0
Equivalence ratio
N
O

c
o
n
c
e
n
t
r
a
t
i
o
n

(
p
p
m
v
w
)
C
O

c
o
n
c
e
n
t
r
a
t
i
o
n

(
v
o
l
.

%
)
Figure 1.20
NOx and CO as a function of stoichiometry. (From Baukal, C. E., Heat
Transfer in Industrial Combustion, Boca Raton, FL: CRC Press, 2000;
courtesy of CRC Press.)
© 2011 by Taylor and Francis Group, LLC
16 Industrial Combustion Testing
These burners are used where the customer may need
to switch between a gaseous fuel like natural gas and a
liquid fuel like oil, usually for economic reasons. These
burners normally operate on one fuel or the other, and
occasionally on both fuels simultaneously. Another
application where multiple fuels may be used is in waste
incineration. One method of disposing of waste liquids
contaminated with hydrocarbons is to combust them by
direct injection through a burner. The waste liquids are
fed through the burner that is powered by a traditional
fuel such as natural gas or oil. The waste liquids often
have very low heating values and are diffcult to com-
bust without auxiliary fuel. This further complicates the
burner design where the waste liquid must be vaporized
and combusted concurrently with the normal fuel used
in the burner.
1.3.1.1.2 Oxidizer
The predominant oxidizer used in most industrial heat-
ing processes is atmospheric air. This can present chal-
lenges in some applications where highly accurate control
is required due to the daily variations in the barometric
pressure and humidity of ambient air. The combustion
air is sometimes preheated and sometimes blended with
some of the products of combustion, which is usually
referred to as fue gas recirculation (FlGR). In certain
cases, preheated air is used to increase the overall thermal
effciency of a process (see Chapter 21). The FlGR is often
used to both increase thermal effciency and to reduce
NOx emissions. The thermal effciency is increased by
capturing some of the energy in the exhaust gases that are
used to preheat the incoming combustion oxidizer. The
NOx emissions may also be reduced because the peak
fame temperatures are reduced, which can reduce the
NOx emissions that are highly temperature dependent.
There are also many high temperature combustion
processes that use an oxidizer that contains a higher
proportion of oxygen than the 21% (by volume) that is
found in normal atmospheric air. This is referred to as
OEC and has many benefts, which include increased
productivity and thermal effciency while reducing the
exhaust gas volume and pollutant emissions [6]. A sim-
plifed global chemical reaction for the stoichiometric
combustion of methane with air is given as follows:
CH
4
 + 2O
2
 + 7.52N
2
→ CO
2
  + 2H
2
O + 7.52N
2
, trace species.
(1.1)
This compares to the same reaction where the oxidizer
is pure O
2
instead of air:
CH
4
 + 2O
2
→ CO
2
 + 2H
2
O + trace species. (1.2)
The volume of exhaust gases is signifcantly reduced
by the elimination of N
2
. In general, a stoichiometric
oxygen-enhanced methane combustion process may be
represented by:
CH
4
 + 2O
2
 + xN
2
→ CO
2
 + 2H
2
O + xN
2
 + trace species,
(1.3)
where 0 ≤ x ≤ 7.52, depending on the oxidizer. The N
2

contained in air acts as ballast that may inhibit the
combustion process and have negative consequences.
The benefts of using OEC must be weighed against the
added cost of the oxidizer, which in the case of air is
essentially free except for the minor cost of the air han-
dling equipment and power for the blower. The use of
a higher purity oxidizer has many consequences with
regard to heat transfer from the fame and pollutant
emissions generated. Oxygen-enhanced combustion
testing is considered in more detail in Chapter 27.
1.3.1.1.3 Gas Recirculation
A common technique used in combustion systems is to
design the burner to induce furnace gases to be drawn
into the burner to dilute the fame, usually referred to
as furnace gas recirculation (FuGR). Even though the
furnace gases are hot, they are still much cooler than
the fame itself. This dilution may accomplish sev-
eral purposes. One is to minimize NOx emissions by
reducing the peak temperatures in the fame, as in FlGR
(see Figure 1.22). However, FuGR may be preferred to
FlGR because no external high temperature ductwork
or fans are needed to bring the product gases into the
fame zone. Another reason to use FuGR may be to
increase the convective heating from the fame because
Staged air is mixed
with the combustion
products from the
primary zone,
which lowers the
peak flame
temperature.
Sub-stoichiometric
conditions in primary
zone increase the
amount of reducing
agents (H
2
and CO).
Primary air
Oil gun
Tertiary air
Secondary air
Figure 1.21
Dual fuel burner. (From Baukal, C. E., ed., The John Zink Combustion
Handbook, Boca Raton, FL: CRC Press, 2001; courtesy of John Zink
Co. LLC.)
© 2011 by Taylor and Francis Group, LLC
Introduction 17
of the added gas volume and momentum. An example of
FlGR recirculation into the burner is given in Fioravanti,
Zelson, and Baukal [38]. A specifc type of burner incor-
porating FuGR are so-called regenerative burners (see
Chapter 21).
1.3.1.2 General Burner Types
There are numerous ways that burners can be classifed.
Some of the common methods are discussed below, with
a brief consideration as to how the burner performance
is impacted.
1.3.1.2.1 Mixing Type
One common method for classifying burners is accord-
ing to how the fuel and the oxidizer are mixed. In pre-
mixed burners, shown in a cartoon in Figure 1.23 and in
a drawing in Figure 1.24, the fuel and the oxidizer are
completely mixed before combustion begins. Thermal
radiation burners (Chapter 25) usually are of the pre-
mixed type. Premixed burners often produce shorter
and more intense fames, compared to diffusion fames.
This can produce high temperature regions in the fame
leading to nonuniform heating of the load and higher
NOx emissions, although this is very dependent on the
specifc design. However, in fame impingement heating
(see Chapter 9), premixed burners are useful because the
higher temperatures and shorter fames can enhance the
heating rates.
In diffusion-mixed burners, shown schematically in
Figure 1.25 and in a drawing in Figure 1.26, the fuel
and the oxidizer are separated and unmixed prior to
combustion, which begins where the oxidizer/fuel mix-
ture is within the fammability range (assuming the
temperature is high enough for ignition). Oxygen/fuel
burners (see Chapter 27) are usually diffusion burn-
ers, primarily for safety reasons, to prevent fashback.
Diffusion gas burners are sometimes referred to as “raw
gas” or “nozzle mix” burners as the fuel gas exits the
burner essentially intact with no oxidant mixed with it
until it reaches the end of the burner nozzle. Diffusion
burners typically have longer fames than premixed
burners, have lower peak fame temperatures, and usu-
ally have a more uniform temperature and heat fux
distribution. They may also have lower NOx emissions
although again this is design dependent.
ID fan
To atmosphere
Recirculated
Combustion products
Combustor
Burner
A
i
r
F
u
e
l
Figure 1.22
Schematic of fue gas recirculation. (From Baukal, C. E., Heat Transfer
in Industrial Combustion, Boca Raton, FL: CRC Press, 2000; courtesy of
CRC Press.)
Air
Fuel
Figure 1.23
Schematic of a premixed burner. (From Baukal, C. E., Heat Transfer in
Industrial Combustion, Boca Raton, FL: CRC Press, 2000; courtesy of
CRC Press.)
Figure 1.24
Drawing of a premixed burner. (Courtesy of John Zink Co. LLC.)
Air
Fuel
Air
Figure 1.25
Schematic of a diffusion-mixed burner. (From Baukal, C. E., Heat
Transfer in Industrial Combustion, Boca Raton, FL: CRC Press, 2000;
courtesy of CRC Press.)
© 2011 by Taylor and Francis Group, LLC
18 Industrial Combustion Testing
It is also possible to have partially premixed burners,
shown schematically in Figure 1.27 and in a drawing in
Figure 1.28, where a portion of the fuel is mixed with
the oxidizer prior to reaching the end of the burner.
This is often done for stability and safety reasons where
the partial premixing helps anchor the fame, while not
fully premixing lessons the chance for fashback. This
type of burner often has a fame length and temperature
and heat fux distribution that is in-between the fully
premixed and diffusion fames.
Another burner classifcation based on mixing is
known as staging: staged air and staged fuel. A staged
air burner is shown schematically in Figure 1.29 and
in a drawing in Figure 1.30. A staged fuel burner is
shown schematically in Figure 1.31 and in a drawing in
Figure 1.32. Secondary and sometimes tertiary injectors
in the burner are used to inject a portion of the fuel and/
or the oxidizer into the fame, downstream of the root
of the fame. Staging is often done to reduce NOx emis-
sions and produce longer fames. These longer fames
typically have a lower peak fame temperature and more
uniform heat fux distribution than nonstaged fames.
However, an additional challenge is that multiple lon-
ger fames may interact with each other and produce
unpredictable consequences compared to single fames,
including increasing NOx rather than reducing it.
1.3.1.2.2 Fuel Type
Burners may also be classifed according to the fuel
type. Gaseous fuel burners are the predominant type
used in most of the applications considered here. In
general, natural gas is the predominant gaseous fuel
used because of its low cost and availability. However,
a wide range of gaseous fuels are used, for example, in
Figure 1.26
Drawing of a diffusion-mixed burner. (Courtesy of John Zink Co.
LLC.)
Air
Fuel
Air
Fuel
Air
Figure 1.27
Schematic of a partially premixed burner. (From Baukal, C. E., Heat
Transfer in Industrial Combustion, Boca Raton, FL: CRC Press, 2000;
courtesy of CRC Press.)
Figure 1.28
Drawing of a partially premixed burner. (Courtesy of John Zink Co.
LLC.)
Air
Fuel
Air
Air
Fuel
Figure 1.29
Schematic of a staged-air burner. (From Baukal, C. E., Heat Transfer
in Industrial Combustion, Boca Raton, FL: CRC Press, 2000; courtesy
of CRC Press.)
© 2011 by Taylor and Francis Group, LLC
Introduction 19
the chemicals industry [39]. These fuels contain multiple
components such as methane, hydrogen, propane, nitro-
gen, and carbon dioxide and are sometimes referred to
as refnery fuel gases. Figure 1.33 shows an example of a
typical nonluminous gaseous fame from a burner used
in the petrochemical industry. Gaseous fuels are among
the easiest to control because no fuel vaporization is
required as in liquid and solid fuels. They are also often
simpler to control to minimize pollution emissions
because they are more easily staged compared to liquid
and solid fuels.
Liquid fuel burners are used in some limited appli-
cations. One of the specifc challenges of using oils is
vaporizing the liquid into small enough droplets to
burn completely. Improper atomization produces high
unburned hydrocarbon emissions and reduces fuel eff-
ciency. Steam and compressed air are commonly used
to atomize liquid fuels. The atomization requirements
often reduce the options for modifying the burner design
to reduce pollutant emissions. Another challenge is that
liquid fuel oils often contain impurities like nitrogen
and sulfur that can increase pollution emissions. In the
Regen tile
Oil gun
Gas risers (for
combination firing)
Gas pilot
Tertiary air control
Air inlet
plenum
Primary
air control
Secondary air control
Gas riser manifold
(for combination firing)
Figure 1.30
Drawing of a staged-air process burner. (From Baukal, C. E., ed., The
John Zink Combustion Handbook, Boca Raton, FL: CRC Press, 2001;
courtesy of John Zink Co. LLC.)
Fuel
Fuel
Fuel
Air
Air
Figure 1.31
Schematic of a staged fuel burner. (From Baukal, C. E., Heat Transfer
in Industrial Combustion, Boca Raton, FL: CRC Press, 2000; courtesy of
CRC Press.)
Secondary
gas nozzle
Primary gas nozzle
Flame holder
Burner tile
Air register
handle
Air
register
Air plenum
Figure 1.32
Drawing of a staged fuel burner. (From Baukal, C. E., ed., The John
Zink Combustion Handbook, Boca Raton, FL: CRC Press, 2001; courtesy
of CRC Press.)
Figure 1.33
Typical nonluminous fame. (From Baukal, C. E., ed., Heat Transfer in
Industrial Combustion, Boca Raton, FL: CRC Press, 2000; courtesy of
John Zink Co. LLC.)
© 2011 by Taylor and Francis Group, LLC
20 Industrial Combustion Testing
case of fuel bound nitrogen, so-called fuel NOx emis-
sions increase. In the case of sulfur, essentially all of the
sulfur in the fuel converts to SOx emissions.
Solid fuels are not commonly used in most industrial
combustion applications except in utility boilers (see
Chapter 34). The most common solid fuels are coal and
coke. Coal is used in power generation and coke is used
in some primary metals production processes. Another
type of pseudosolid fuel is sludge that is processed
in incinerators. Solid fuels also often contain impuri-
ties such as nitrogen and sulfur that can signifcantly
increase pollutant emissions. Some solid fuels may also
contain hazardous chemicals that can produce carcino-
genic pollution emissions. Because solid fuels are not
used frequently in the applications considered, they are
only discussed in those specifc cases.
There are some applications that require the burner to
be able to fre on a gaseous fuel-like natural gas, a liquid
fuel-like fuel oil, or both simultaneously. This is gener-
ally due to the economics of the fuel costs. In some loca-
tions, a more favorable fuel cost rate can be obtained, for
example on natural gas, if the supply may be interrupted
with suffcient notice. The back-up fuel is typically fuel
oil. These dual-fuel burners (see Figure 1.21) have spe-
cial challenges because of the signifcant differences in
designs for gaseous and liquid burners.
1.3.1.2.3 Oxidizer Type
Burners and fames are often classifed according to the
type of oxidizer that is used. The majority of industrial
burners use air for combustion. In many of the higher
temperature heating and melting applications, such as
glass production, the oxidizer is pure oxygen. In other
applications, the oxidizer is a combination of air and
oxygen, often referred to as oxygen-enriched air com-
bustion. Both of these burner types are discussed in
Chapter 27.
Figure 1.34 shows a schematic of an air/fuel burner
that is the most commonly used type in industrial
combustion applications. In most cases the combus-
tion is supplied by a fan or blower although there are
many applications in the petrochemical industry where
natural draft burners are commonly used. There are
numerous variations of air/fuel burners that are dis-
cussed throughout the book.
Figure 1.35 shows a method of using OEC com-
monly referred to as an oxy/fuel burner. In nearly
all cases, the fuel and the oxygen remain separated
inside the burner and do not mix until reaching the out-
let of the burner. This is commonly referred to as a noz-
zle-mix burner that produces a diffusion fame. There is
no premixing of the gases for safety reasons. Due to the
extremely high reactivity of pure O
2
, there is the poten-
tial for a fashback if the gases are premixed. In this
method, high purity oxygen ( > 90% O
2
by volume) is
used to combust the fuel. There are several ways of gen-
erating the O
2
. In an oxy/fuel system, the actual purity
of the oxidizer will depend on what method has been
chosen to generate the O
2
. Oxy/fuel combustion has the
greatest potential for improving a process compared to
air/fuel, but it also may have the highest operating cost.
Figure 1.36 shows an air/fuel process where the air is
enriched with O
2
. This may be referred to as low-level O
2

enrichment or premix enrichment. Many conventional
air/fuel burners can be adapted for this technology [40].
The O
2
is injected into the incoming combustion air
supply, usually through a diffuser to ensure adequate
mixing. This is usually an inexpensive retroft that can
provide substantial benefts. Typically, the added O
2
will
shorten and intensify the fame. However, there may
Air
Fuel
Figure 1.34
Schematic of an air/fuel burner. (From Baukal, C. E., ed., Oxygen-
Enhanced Combustion, Boca Raton, FL: CRC Press, 1998.)
Fuel
Oxygen
Figure 1.35
Schematic of an oxy/fuel burner. (From Baukal, C. E., ed., Oxygen-
Enhanced Combustion, Boca Raton, FL: CRC Press LLC, 1998; courtesy
of CRC Press.)
Air
Fuel
Oxygen
Figure 1.36
Schematic of an air-oxy/fuel burner. (From Baukal, C. E., ed., Oxygen-
Enhanced Combustion, Boca Raton, FL: CRC Press LLC, 1998; courtesy
of CRC Press.)
© 2011 by Taylor and Francis Group, LLC
Introduction 21
be some concerns if too much O
2
is added to a burner
designed for air/fuel. The fame shape may become
unacceptably short. The higher fame temperature may
damage the burner or burner block. The air piping may
need to be modifed for safety reasons to handle higher
levels of O
2
.
1.3.1.2.4 Draft Type
Most industrial burners are known as forced-draft
burners. This means that the oxidizer is supplied to the
burner under pressure. For example, in a forced-draft
air burner, the air used for combustion is supplied to
the burner by a blower. In natural-draft burners (see
Chapter 18), the air used for combustion is induced into
the burner by the negative draft produced in the com-
bustor and by the motive force of the incoming fuel that
may be at a signifcant pressure. A schematic is shown
in Figure 1.37 and an example is shown in Figure 1.38.
In this type of burner, the pressure drop and combustor
stack height are critical in producing enough suction to
induce enough combustion air into the burners. This
type of burner is commonly used in the chemical and
petrochemical industries in fuid heaters. The main
consequence of the draft type on burner performance is
that the natural-draft fames are usually longer than the
forced-draft fames so that the heat fux from the fame
is distributed over a longer distance and the peak tem-
perature in the fame is often lower.
1.3.1.2.5 Heating Type
Burners are often classifed as to whether they are direct
(see Figure 1.39) or indirect heating (see Figure 1.40). In
direct heating, there is no intermediate heat exchange
surface between the fame and the load. In indirect
heating, such as radiant tube burners (see Chapter 24),
there is an intermediate surface between the fame and
the load. This is usually done because the combustion
Secondary
air
Primary
air
Gas Pilot gas
Air
Air
Figure 1.37
Schematic of a natural draft burner. (Courtesy of John Zink Co.
LLC.)
Figure 1.38
Photograph of a natural draft burner. (From Baukal, C. E., Heat
Transfer in Industrial Combustion, Boca Raton, FL: CRC Press, 2000;
courtesy of CRC Press.)
Burner
Load
Figure 1.39
Direct fred process. (From Baukal, C. E., ed., Industrial Burners
Handbook, Boca Raton, FL: CRC Press, 2004.)
Muffle
Load
Burner
Figure 1.40
Indirect fred process. (From Baukal, C. E., ed., Industrial Burners
Handbook, Boca Raton, FL: CRC Press, 2004.)
© 2011 by Taylor and Francis Group, LLC
22 Industrial Combustion Testing
products cannot come in contact with the load because
of possible contamination.
Radiation heat transfer from the fame to the product
is the primary mode used in many industrial combus-
tion systems. There are a variety of burner designs that
rely primarily on this mechanism. Radiant wall burners
are discussed in Chapter 18. Radiant tube burners are
discussed in Chapter 24. Thermal radiation burners are
discussed in Chapter 25.
Forced convection is the other predominant mecha-
nism for transferring heat from fames to a load. For
example, high velocity burners are particularly useful in
applications where primarily radiant heating may over-
heat the surface with much less energy getting inside
the load. An example would be heating a pile of scrap
metal. Highly radiant heating could melt the outside of
the pile and cause excessive oxidation leading to high
metal yield losses. Convective heating can penetrate
inside the load to cause more uniform heating. In certain
applications, high velocity burners may not be preferred
because the materials being heated may contain fne
particles that can easily become airborne. An example is
glass manufacturing where the incoming batch materi-
als contain fne powders.
1.3.1.2.6 Geometry
There are two primary shapes for the outlet nozzle of
industrial burners: round or rectangular. Figure 1.41
shows identical heaters with the same number of burn-
ers, but with different burner shapes: round fame and
fat fame. Round fames are the predominant shape
used in industry. Most of the burners discussed in this
book are predominantly round. This is often due to the
lower cost of making round shapes compared to making
rectangular shapes. It is also often due to the burner tile
where round shapes generally require less maintenance
compared to rectangular tiles that have corners more
susceptible to cracking. Another reason may be due to
more preferred fow patterns inside round burner ple-
nums compared to rectangular shapes.
Rectangular shapes are sometimes preferred in cer-
tain applications depending upon the geometry of the
combustor and the load. Burners with a fairly high
aspect ratio (length to width) are sometimes referred to
as “fat” fame burners because the fame shape appears
to be fat. One example is in ethylene cracking furnaces
where fat-shaped burners fre up along a refractory wall
to heat the wall to radiate to tubes opposite that wall.
Another example is in glass furnaces where fat-shaped
fames fre over the molten glass where these fat shapes
often give better fame coverage, more uniform heating,
and better thermal effciency.
1.3.1.3 Burner Components
There are several important components that are briefy
considered here that impact the burner design. The igni-
tion system is an important component in the burner
system to ensure safe and reliable operation. The ignition
system is often built into the burner, but in some cases it
may be separate from the burner. The system may be fully
automatic or completely manual. Different types of ignit-
ors are available. In many cases a pilot is used to ignite
the main fame. This may be continuous or interruptible
depending on the system design. The pilot may be perma-
nent or removable. The pilot may be ignited by something
like a spark-ignitor or by an external torch. Pilots require a
separate fuel supply and are typically premixed.
Plenums are used to homogenize the incoming gas
fows to evenly distribute them to the outlet of the
burner. This is important to ensure proper operation of
the burner over the entire range of operating conditions,
especially at turndown. These gases may include com-
bustion air, premixed fuel and air, or partially premixed
fuel and air. If the plenum is too large, then the fows
may be unevenly distributed across the burner nozzle
outlet. If the plenum is too small, then the pressure drop
through the plenum may be excessive.
The burner tile, sometimes referred to as a block or
quarl, is an important component because it helps shape
the fame and protects the internal parts from overheat-
ing. In the majority of designs, the burner tile is made of
some type of ceramic that often contains alumina and sil-
ica depending upon the temperature requirements. The
burner tile may also play an important role in the igni-
tion and fuid dynamics of the combustion process. The
tile may have bluff body components that enhance fame
stability. There may be holes through the tile to enhance
mixing of furnace gases with the gases fed into the burner.
Advances in ceramics and manufacturing processes have
led to increasingly more complicated burner tiles.
Burner
(a)
Burner
Burner
(b)
Burner
Burner
Burner
Figure 1.41
(a) Round and (b) rectangular burner shapes in identical combustors.
(From Baukal, C. E., ed., Industrial Burners Handbook, Boca Raton, FL:
CRC Press, 2004.)
© 2011 by Taylor and Francis Group, LLC
Introduction 23
There may be controls on the burner. For example,
there is often a damper built into natural draft burn-
ers (see Chapter 18) to control the incoming air fow and
the furnace draft. Other controls on a burner may be
for adjusting the distribution of fuels or air throughout
the burner. For example, if a burner has multiple fuel
injectors particularly for fuel staging, controls on the
burner may be used to control how much fuel goes to
each injector.
The fame safety system is critical to the safe opera-
tion of the combustion system. This may include some
type of fame scanner or fame rod to ensure either the
burner or the pilot is operating. These are connected
to the fuel supply system so that the fuel fow will
be stopped if the fame goes out to prevent a possible
explosion for unreacted fuel gases contacting a hot sur-
face somewhere in the combustor.
1.3.2 Combustors
This section briefy introduces the combustors that
are commonly used in industrial heating and melting
applications.
1.3.2.1 Design Considerations
There are many important factors that must be consid-
ered when designing a combustor. This section only
briefy considers a few of the more important factors.
1.3.2.1.1 Load Handling
A primary consideration for any combustor is the type
of material that will be processed. The various types of
loads are considered later in this section. One obvious
factor of importance in handling the load and trans-
porting it through the combustor is its physical state,
whether it is a solid, liquid, or gas. Another factor is
the transport properties of the load. For example, the
solid may be granular or it might be in the form of a
sheet (web). Related to that is how the solid will be fed
into the combustor. A granular solid could be fed into
a combustor continuously with a screw conveyor or it
could be fed in with discrete charges from a front-end
loader. The shape of the furnace will vary according to
how the material will be transported through it. For
example, limestone is fed continuously into a rotat-
ing and slightly downwardly inclined cylinder (see
Figure 1.17).
1.3.2.1.2 Temperature
Industrial heating applications may be divided into
two broad categories: higher and lower temperatures.
The division between the two is somewhat arbitrary
but mainly concerns the different types of applications
used in each. For example, most of the metal and glass
melting applications fall into the higher temperature
categories as the furnace temperatures are often well
over 2000°F (1400 K). They use technologies like air
preheating and oxygen-enrichment (see Chapter 27) to
achieve those higher temperatures. Lower temperature
applications include dryers, process heaters, and heat
treating and are typically below about 2000°F (1400 K).
Although many of these processes may use air preheat-
ing, it is primarily to improve the thermal effciency and
not to get higher fame temperatures. Those processes
rarely use oxygen-enrichment, which are usually only
economical for higher temperature processes. Both the
combustors and the burners are designed differently
for higher and lower temperature processes. The heat
transfer and pollutant generation mechanisms are often
different as well. In higher temperature processes, the
primary mode is often radiation, while in lower temper-
ature applications convection often plays a signifcant
role. The NOx emissions, for example, are much more
signifcant at higher temperatures compared to lower
temperatures.
1.3.2.1.3 Heat Recovery
When heat recovery is used in an industrial combus-
tion process, it is an integral part of the system. The two
most popular methods are regenerative and recupera-
tive that are discussed briefy below. The heat recovery
system is important in the design of the combustor as
it determines the thermal effciency of the process and
the fame temperatures in the system. It also infuences
the heat transfer modes as it may increase both the
radiation and convection because of higher fame tem-
peratures. Another type of heat recovery that is used in
some processes is furnace or FlGR where the exhaust
products are recirculated back through the fame. This
also infuences the heat transfer and furnace design
as it can moderate the fame temperature but increase
the volume fow of gases through the combustion
chamber.
1.3.2.2 General Classifcations
There are several ways that a combustor can be classi-
fed that are briefy discussed in this section.
1.3.2.2.1 Load Processing Method
Furnaces are often classifed as to whether they are batch
or continuous. In a batch furnace, the load is charged into
the furnace at discrete intervals. There may be multiple
load charges, depending on the application. Normally,
the fring rate of the burners is reduced or ceased during
the charging cycle. On some furnaces, a door may also
need to be opened during charging. These signifcantly
impact the thermal effciency in the system as the
heat losses during the charge cycle are very large. The
© 2011 by Taylor and Francis Group, LLC
24 Industrial Combustion Testing
radiation losses through open doors are high and the
reduced fring rate may not be enough to maintain the
furnace temperature. In some cases, the temperature on
the inside of the refractory wall, closest to the load, may
actually be lower than the temperature of the refractory
at some distance from the inside, due to the heat losses
during charging. The heating process and heat trans-
fer are dynamic and constantly changing as a result of
the cyclical nature of the load charging. The burners
may cycle on and off or between high fre and low fre.
This makes testing of these systems more complicated
because of the transient nature of the process.
In a continuous furnace, the load is fed into and out
of the combustor constantly. The feed rate may change
sometimes due to conditions upstream or downstream
of the combustor or due to the production needs of the
plant, but the process is nearly steady state. This makes
continuous processes simpler to test as there is no need
to include time as a variable in the analysis. It is often
easier to make meaningful measurements in continu-
ous processes due to their steady-state nature.
There are some furnaces that are semicontinuous
where the load may be charged in a nearly continuous
fashion, but the fnished product may be removed from
the furnace at discrete intervals. An example is an alu-
minum reverberatory furnace that is charged using an
automatic side well feed mechanism (see Figure 1.42).
In that process, shredded scrap is added in nearly
continuous increments to a circulating bath of molten
aluminum. When the correct alloy composition has
been reached and the furnace has a full load, some or all
of that load is then tapped out of the furnace. The effect
on heat transfer is somewhere between that for batch
and continuous furnaces.
1.3.2.2.2 Heating Type
As described above for burners, combustors are
often classifed as direct (see Figure 1.39) or indirect
(see Figure 1.40) heating. In indirect heating, there
is some type of intermediate heat transfer medium
between the fames and the load that keeps the
combustion products separate from the load. One
example is a muffe furnace where there is a high tem-
perature ceramic muffe between the fames and the
load. The fames transfer their heat to the muffe, which
then radiates to the load that is usually some type of
metal. The limitation of indirect heating processes is
the temperature threshold of the intermediate mate-
rial. Although ceramic materials have fairly high tem-
perature limits, other issues such as structural integrity
over long distance spans and thermal cycling can still
reduce the recommended operating temperatures.
Another example of indirect heating is in process heat-
ers where fuids are transported through metal tubes
that are heated by hot gases. Indirect heating pro-
cesses often have fairly uniform heat fux distributions
because the heat exchange medium tends to homoge-
nize the energy distribution from the fames to the load.
The heat transfer from the heat exchange surface to the
load is often fairly simple and straightforward to com-
pute because of the absence of chemical reactions in-
between. However, the heat transfer from the fames to
the heat exchange surface and the subsequent thermal
conduction through that surface are as complicated as
if the fame was radiating directly to the load. Indirect
heating may also have an advantage for reducing pol-
lutant emissions when contact of the high temperature
exhaust gases with the load could generate pollutants.
As a result of the temperature limits of the heat
exchange materials, most higher temperature processes
are of the direct heating type where the fames can
directly heat the load.
1.3.2.2.3 Geometry
Another common way of classifying combustors
is according to their geometry that includes their shape
and orientation. The two most common shapes are rect-
angular and cylindrical. The two most common orien-
tations are horizontal and vertical, although inclined
furnaces are commonly used in certain applications
such as rotary cement furnaces. An example of using the
shape and orientation of the furnace as a means of classi-
fcation would be a vertical cylindrical heater (sometimes
referred to as a VC) used to heat fuids in the petrochemi-
cal industry. Both the furnace shape and orientation
have important effects on the heat transfer in the system.
They also determine the type of test that will be done.
Pump well
Charge well
Figure 1.42
Aluminum reverberatory furnace. (From Baukal, C. E., Heat Transfer
in Industrial Combustion, Boca Raton, FL: CRC Press, 2000; courtesy of
CRC Press.)
© 2011 by Taylor and Francis Group, LLC
Introduction 25
For example, in a VC heater it is often possible to model
only a slice of the heater due to its angular symmetry, in
which case cylindrical coordinates would be used. On
the other hand, it is usually not reasonable to model a
horizontal rectangular furnace using cylindrical coordi-
nates, especially if buoyancy effects are important.
Some furnaces are classifed by what they look like.
One example is a shaft furnace used to make iron. The
raw materials are loaded into the top of a tall thin verti-
cally oriented cylinder. Hot combustion gases generated
at the bottom through the combustion of coke fow up
through the raw materials that get heated. The melted
fnal product is tapped out of the bottom. The furnace
looks and acts almost like a shaft because of the way
the raw materials are fed in through the top and exit at
the bottom. A transfer chamber used to move molten
metal around in a steel mill is often referred to as a ladle
because of its function and appearance. These ladles are
preheated using burners (see Figure 1.6) before the mol-
ten metal is poured into them to prevent the refractory-
lined vessels from thermally shocking.
Another aspect of the geometry that is important in
some applications is whether the furnace is moving or
not. For example, in a rotary furnace for melting scrap
aluminum, the furnace rotates to enhance mixing and
heat transfer distribution. This again affects the type of
testing that would be appropriate for that system and
can add some complexity to the computations.
The burner orientation with respect to the combustor
is also sometimes used to classify the combustor. For
example, a wall-fred furnace has burners located in and
fring along the wall.
1.3.2.2.4 Heat Recuperation
In many heat processing systems, energy recuperation
is an integral part of the combustion system. Sometimes
the heat recuperation equipment is a separate component
of the system and sometimes it is part of the burners.
Depending on the method used to recover the energy,
the combustors are commonly referred to as either
recuperative or regenerative (see discussion below). The
heat transfer in these systems is a function of the energy
recovery system. For example, the higher the combustion
air preheat temperature, the hotter the fame and the
more radiant heat that can be produced by that fame. The
convective heat transfer may also be increased due to
the higher gas temperature and also due to the higher
thermal expansion of the gases that increases the average
gas velocity through the combustor.
1.3.3 Heat load
This section is a brief introduction to some of the
important issues concerning the heat load in a furnace
or combustor. In petrochemical production processes,
process heaters are used to heat petroleum products for
further processing. The fuids are transported through
the process heaters in process tubes. These heaters often
have a radiant section and a convection section. In the
radiant section, the process tubes are heated primarily
by radiation. In the convection section, the process tubes
are heated primarily by convection. The design of the
radiant section is especially important as fame impinge-
ment on the tubes can cause premature failure of the
tubes or cause the hydrocarbon fuids to coke inside the
tubes that reduces the heat transfer to the fuids.
In some applications, heaters and burners are used to
heat or dry moving substrates or webs. An example is
shown in Figure 1.43. One common application is the use
of gas-fred infrared burners (see Chapter 25) to remove
moisture from paper during the forming process [41].
These paper webs can travel at speeds over 300 m/s (1000
ft/s) and are normally dried by traveling over and con-
tacting steam-heated cylinders. The IR (infrared) heat-
ers are often used to selectively dry certain portions of
the web that may be wetter than others. For example, if
the target moisture content for the paper is 5%, then the
entire width of the paper must have no more than 5%
moisture. Streaks of higher moisture areas often occur
in sections along the width of the paper. Without selec-
tively drying those areas, those streaks would be dried
to the target moisture level, which means that the rest of
the sheet would be dried to even lower moisture levels.
This creates at least three important problems. The frst
is lost revenue because paper is usually sold on a weight
basis. Any water unnecessarily removed from the paper
decreases its weight and therefore results in lost income.
The second problem is a reduction in the quality of the
paper. If areas of the paper are too dry, they do not
handle as well in devices like copiers and printers and
are not nearly as desirable as paper of uniform mois-
ture content. The third problem is increased operating
costs where more energy is consumed than necessary
to dry sections of the paper that are already below the
target moisture content. Therefore, selective drying of
the paper only removes the minimum amount of water
from the substrate. The challenge in this application is
IR burner
Dry web
IR burner
Roller
Wet web
Figure 1.43
Elevation view of infrared burner heating a moving web. (From
Baukal, C. E., Heat Transfer in Industrial Combustion, Boca Raton, FL:
CRC Press, 2000; courtesy of CRC Press.)
© 2011 by Taylor and Francis Group, LLC
26 Industrial Combustion Testing
to measure the moisture content profle across the width
of a sheet that may be several meters wide and mov-
ing at high speeds. The moisture content information
must then be fed to the control system for the IR heaters,
which then must be able to react almost instantaneously.
This is possible today because of advances in measure-
ment and controls systems.
Another example of a moving substrate application is
using IR burners to remove water during the produc-
tion of fabrics in textile manufacturing [42]. Moving
substrates present unique challenges for burners. Often
the material being heated can easily be set on fre if there
is a line stoppage and the burners are not turned off
quickly enough. This means that the burner control sys-
tem must be interlocked with the web handling equip-
ment so that the burners can be turned off immediately
in the event of a line stoppage. If the burners have sub-
stantial thermal mass, then the burners may need to be
retracted away from the substrate during a stoppage or
heat shields may need to be inserted between the burn-
ers and the substrate to prevent overheating.
Convection dryers are also used to heat and dry sub-
strates. Typically, high velocity heated air is blown at
the substrate from both sides so that the substrate is ele-
vated between the nozzles. In many cases, the heated air
is used for both heat and mass transfer, to volatilize any
liquids on or in the substrate such as water, and then
carry the vapor away from the substrate.
An important aspect of heating webs is how the
energy is transferred into the material. For example, dry
paper is known to be a good insulator. When steam cyl-
inders are used to heat and dry paper, they become less
and less effective as the paper becomes drier because
the heat from the cylinder does not conduct through
the paper as well as when it is moist since the thermal
conductivity of the paper increases with moisture con-
tent. Infrared burners are effective for drying paper
because the radiant energy transfers into the paper
and is absorbed by the water. The radiant penetration
into the paper actually increases as the paper becomes
drier, unlike steam cylinders that become less effective.
Another important aspect related to burner design is the
ability to cool down rapidly to prevent causing a fre, for
example, when a paper web stops unexpectedly during
processing due to some type of problem. An alternative
is to shield the burner from the web in the event of a
sudden line stoppage.
1.3.3.1 Opaque Materials
This type of load encompasses a wide range of materials
including granular solids like limestone and liquids like
molten metal. For this type of load, the heat transfers
to the surface of the load and must conduct down into
the material. This process can be enhanced by proper
mixing of the materials so that new, cooler material is
constantly exposed to the surface as in rotary kilns or
in aluminum reverberatory furnaces that have molten
metal pumps to continuously recirculate the metal
through the heating zone. The potential problems with
this method include overheating the surface materials
or having lower thermal effciencies by limiting the heat
transfer to the surface to prevent overheating.
1.3.3.2 Transparent Materials
The primary example of this type of load is glass that
has selective radiant transmission properties. In glass
melting processes, the primary mode of heat transfer is
by radiation. Flames have specifc types of radiant out-
puts that vary as a function of wavelength (see Chapter
6). If the fame is nonluminous, the fame usually has
higher radiant outputs in the preferred wavelengths for
water and carbon dioxide bands. If the fame is lumi-
nous, it has a broader, more graybody-type of spectral
radiant profle. Luminous fames are preferred in melt-
ing glass because of the selective transmission proper-
ties of molten glass. This allows a signifcant portion of
the radiation received at the surface of the glass to pen-
etrate into the glass, which enhances heat transfer rates
and reduces the chances of overheating the surface that
would reduce product quality.
1.3.4 Heat recovery Devices
Heat recovery devices are often used to improve the
effciency of combustion systems. Some of these devices
are incorporated into the burners, but more com-
monly they are another component in the combustion
system, separate from the burners. These heat recov-
ery devices incorporate some type of heat exchanger,
depending on the application. The two most common
types have been recuperators and regenerators that are
briefy discussed next. Reed (1987) predicts an increas-
ing importance for heat recovery devices in industrial
combustion systems for increasing heat transfer and
thermal effciencies [43].
1.3.4.1 Recuperators
A recuperator is a low- to medium-temperature (up to
about 1300°F or 700°C), continuous heat exchanger that
uses the sensible energy from hot combustion prod-
ucts to preheat the incoming combustion air. These
heat exchangers are commonly counterfow where the
highest temperatures for both the combustion products
and the combustion air are at one end of the exchanger
with the coldest temperatures at the other end. Lower
temperature recuperators are normally made of metal,
while higher temperature recuperators may be made of
© 2011 by Taylor and Francis Group, LLC
Introduction 27
ceramics. Recuperators are typically used in lower tem-
perature applications because of the limitations of the
metals used to construct these heat exchangers.
1.3.4.2 Regenerators
A regenerator is a higher temperature, transient heat
exchanger that is used to improve the energy effciency
of high temperature heating and melting processes, par-
ticularly in the high temperature processing industries
like glass production. In a regenerator, energy from the
hot combustion products is temporarily stored in a unit
constructed of frebricks. This energy is then used to
heat the incoming combustion air during a given part
of the fring cycle up to temperatures in excess of 2000°F
(1000°C).
Regenerators are normally operated in pairs. During
one part of the cycle, the hot combustion gases are fow-
ing through one of the regenerators and heating up
the refractory bricks, while the combustion air is fow-
ing through and cooling down the refractory bricks in
the second regenerator. Both the exhaust gases and the
combustion air directly contact the bricks in the regen-
erators, although not both at the same time since they
are in a different regenerator at any given time. After a
suffcient amount of time (usually from 5 to 30 min.), the
cycle is reversed so that the cooler bricks in the second
regenerator are then reheated while the hotter bricks in
the frst regenerator exchange their heat with the incom-
ing combustion air. A reversing valve is used to change
the fow from one gas to another in each regenerator.
The burners used in these systems must be capable of
not only handling the high temperature preheated air,
but also the constant thermal cycling.
1.4 Testing
This section discusses some of the common purposes
for testing, the general classifcation of probes used in
testing, a brief discussion of the testing confgurations
in industrial combustion processes, and some of the
important parameters in typical tests.
1.4.1 Purposes
There are numerous reasons why tests are conducted
in industrial combustion processes. Initial new product
development is sometimes done at a reduced scale to
minimize costs and risks. When the product is close to
commercialization, large-scale (preferably full-scale)
tests are required to validate the technology. End
users are not likely to use a new technology until it
has been validated at or near full scale, although they
may be willing to do validation testing if the risk is low
enough and the reward is large enough. One way this
is sometimes done, for example, is to replace a single
old burner with a new burner design in a combustor
with many burners. While the data collected from
such a test is of limited value, the new technology can
be demonstrated to be feasible. This may be done, for
example, when a new low NOx burner design is being
validated. The key is that there are no operational
issues identifed during this testing that could cause
an unscheduled process shutdown or unexpectedly
reduce throughput.
Testing may be done to generate data that can be
used to validate computer models. There is sometimes
the tendency to believe anything generated by a com-
puter. However, the adage “garbage in = garbage out”
is appropriate because the model is only as good as
the physics in the model and the accuracy of the input
data. Nearly all models involve some level of simpli-
fcation because it is normally prohibitive in terms of
time and cost to model the exact physics and because
the exact physics are not always known. Computational
fuid dynamic (CFD) modeling is a tool that has become
popular for simulating all types of combustion prob-
lems [44] (see Chapter 11). While it is often possible to
model almost anything, this does not mean all predic-
tions are equally valid. Experimental validation must
be done to show how well models match actual mea-
surements. However, this means the variability in the
experimental measurements must be accurately quan-
tifed, so the comparisons with modeling results can be
fairly judged. It also means that enough experimental
data must be collected to compare the model results
over a range of conditions. Failure to do so may limit
the reliable range of the model. It may also mean that
defciencies in the model may not be identifed because
the range of conditions experimentally tested was too
narrow. This is one reason why an adequate design of
experiments should be completed prior to starting test-
ing (see Chapter 3).
Testing can be used to validate new experimental
techniques. For example, the optical techniques for mea-
suring temperature discussed in Chapter 5 have been
tested against more traditional and well- established
techniques such as bare wire thermocouples and suction
pyrometers. Some of these optical techniques (e.g., tun-
able diode lasers discussed in Chapter 14) used more
traditional methods such as bare wire thermocouples
for calibration. In either case, testing is required to
validate new experimental techniques. In many cases,
a well known and fully tested confguration with suf-
fcient measurements using proven techniques will be
used to demonstrate the validity of a new measurement
technique.
© 2011 by Taylor and Francis Group, LLC
28 Industrial Combustion Testing
Testing is done to determine equipment performance.
Performance is a broad term that can encompass many
different elements. For burners, some important mea-
surements may include pollution emissions (e.g., NOx,
CO), turndown range (minimum and maximum fr-
ing rates), fuel variability (e.g., hydrogen content),
noise, fame shape (length and cross section), heat fux
profle, waste destruction capability for waste fuels,
and fame stability (a subjective parameter). For fares,
some important measurements include the smoke-
less capacity, the hydraulic capacity, thermal radiation,
and noise (see Chapter 28). All of these are determined
experimentally to ensure that the equipment meets any
guarantees and warrantees that may have been made.
For example, Figure 1.44 shows a process burner with a
very large muffer. The noise specifcation imposed by
the end user was no more than 64 dBA, which had to be
tested to demonstrate the burner met that very stringent
performance criterion.
Testing may be done to identify and solve problems.
For example, process burners (see Chapter 18) are often
tested under a range of conditions that rarely exist in
actual operation. The tests are done to determine how
the burner would behave under those conditions if they
ever should arise. One test condition is referred to as
“CO breakthrough” where the excess air in the burner
is continuously reduced until there is a spike in CO
emissions. In general, the CO emissions begin to rise at
a faster pace as the excess O
2
approaches zero. The pur-
pose of this test condition is to make sure the burner
remains stable even at low excess O
2
levels. While an
actual production process heater is not normally run at
those levels, it is possible they could be seen due to an
upset or an unexpected transient such as a rapid increase
in the fuel heating value. Related to that, process burn-
ers are tested under a wide range of fuel compositions
and fring rates to ensure stability. If the burner becomes
unstable under any of the test conditions, the burner
vendor modifes the design to correct the problem so
the burner safely operates under the range of condi-
tions that may be encountered. This does not mean the
burner will optimally perform (e.g., low emissions) over
the entire range of conditions tested, but it must be safe
over the entire range. There is normally a design case
where optimal performance is guaranteed.
Testing may be done to measure the variability in a
process. This is particularly important in industrial pro-
duction processes where the material being heated may
have tight tolerances and specifc performance require-
ments. If there is considerable variation in the combus-
tion process, this could adversely impact the product
quality. This may mean that appropriate control equip-
ment must be used to ensure any combustion variability
is within acceptable tolerances. An example will illus-
trate this aspect of testing. Changes in the ambient air
can signifcantly impact the pollution emissions from a
combustion process [45]. These normal variations in
air temperature and humidity, atmospheric pressure,
and wind speed and direction must be considered to
properly adjust the operating conditions. Either more
frequent manual adjustments or automatic controls can
be used to combat these environmental fuctuations.
Another example is the variation that occurs in the fuel
gas composition in, for example, most process heaters
[39]. Testing can be used to determine if variations in the
combustion process adversely affect the product, and if
they do, how effective are the corrective measures.
Testing is done to generate performance data that
could be related to the burners, the combustor, or to the
load. One example of performance data for a burner is
the heat fux profle (see Chapter 6) it produces under
specifc operating conditions. For example, ethylene
cracking furnaces are designed to operate with burn-
ers producing a specifc heat fux profle. Flux rates are
typically highest near the middle of the furnace and
lower near the foor and ceiling. However, the variation
between the highest and lowest fux rates is typically
specifed by the furnace designer. If there is too much
variation, then either production rates may have to be
reduced or the run length, before performing mainte-
nance, may have to be reduced. In either case, this reduc-
tion in performance is very costly to the end user so it
is very important the burners produce the desired pro-
fle. This is usually determined through testing in pilot-
scale furnaces (see Chapter 18) that have only a fraction
of the number of burners that a production furnace has,
but the burners are normally full scale. The fuel com-
position, draft conditions, and furnace temperature are
matched to that of the production furnace to ensure the
testing is realistic.
Testing is sometimes done to identify opportunities
for increasing process effciency that often reduces fuel
consumption. For example, it is well known that the
excess O
2
has a big impact on the thermal effciency of
a combustion process [45]. If excess O
2
is too high, then
the extra air absorbs energy, which it carries out the
Burner
Muffler
4 ft
(1.2 m)
8 ft
(2.4 m)
Figure 1.44
Process burner with very large muffer.
© 2011 by Taylor and Francis Group, LLC
Introduction 29
exhaust, thus reducing effciency. If excess O
2
is too low,
then high CO emissions may result, which also reduces
effciency because CO is a fuel with heating value that is
not being utilized. Testing may be done in a test furnace
or in the feld in an actual production furnace to deter-
mine how low the excess O
2
can be before signifcant
levels of CO are generated. For manually controlled fur-
naces, some margin for error is often built into the mini-
mum O
2
level to account for variations that occur such
as changes in air temperature, humidity, and baromet-
ric conditions as well as other weather-related factors
such as high winds and rain. Other examples of testing
to improve process effciency might be using a different
type of burner, changing the fring rate profle in the
furnace, or moving burners to different locations in the
combustor to try to improve the heating effciency.
Testing may be necessary for environmental compli-
ance. There continues to be great interest in reducing
pollution emissions from combustion equipment [46].
Many industrial combustion processes have permitted
limits on how much pollution they are allowed to emit.
Testing is nearly always required to prove the equip-
ment meets the permitted levels. There may be periodic
testing (e.g., annually) or there may be continuous test-
ing such as with continuous emissions monitoring sys-
tems or CEMS (see Chapter 7). Environmental testing is
usually done in the actual production equipment, but in
some cases it may be done in a test facility, depending
on the requirements of the permitting agency.
Testing may be used to enhance combustion safety
including determining equipment operational limits.
There are safety devices such as ultraviolet fame scanners,
fame rods (ionization scanners), and thermocouples that
are used to detect the presence (and by default the absence)
of a fame. These are not traditional testing devices as
defned here. However, there are tests that are conducted
to determine the operating limits of combustion equip-
ment such as burners and fares. Nearly all burners have
a lower and upper fring rate range [47]. Premix burners
can fashback if the fring rate is too low. Flame liftoff can
occur in all types of burners if the maximum design fr-
ing rate is exceeded. Although in some limited cases it
may be possible to determine operating limits with com-
puter modeling, in most cases these limits are determined
through testing.
Testing is sometimes used to explain complex phe-
nomenon. While this is often done with CFD modeling,
it is also done experimentally. In some cases, it is not yet
possible to model all of the physics for a complete indus-
trial combustion system. Better understanding of the
physics behind a process can lead to both improvements
of existing technologies as well as leading to new devel-
opments. For example, understanding how and where
NOx is formed in a fame can help in the development
of lower NOx technologies. Besides the importance to
new product development, test data is often used in the
patent process to demonstrate why a particular technol-
ogy is in fact new and novel and should be granted a
patent.
Testing is sometimes used to develop empirical models
that can be used to predict equipment performance.
Testing and CFD modeling can both be expensive and
time-consuming. For existing combustion equipment, it
is often more cost and time effective to have empirical
models to predict performance. For example, a given
technology may come in a variety of sizes. Rather
than testing every size, a common approach is to test
at least the extremes and possibly one or more sizes in-
between depending on how big the range is between
those extremes. Then, the performance for any size can
be extrapolated from the empirical data, without the
need to test or model every possible size.
1.4.2 Probe Types
Becker (1993) developed a classifcation system for phys-
ical probes based on the duty conditions as shown in
Table 1.5 [48]. While optical probes continue to increase
Table 1.5
Classifcation of Duty Conditions for Physical Probes
Duty Class Typical Case
Radiative
Boundary Gradients Probe Issues
A Lean premixed gas fame Cold Very high Miniaturization important
B Turbulent gas diffusion fame in
open air
Cold Medium Modest if any cooling
requirements
C Turbulent gas diffusion fame in a
combustor
Hot Medium Modest if any cooling
requirements
D Turbulent pulverized-fuel fame in
a combustor
Hot Medium High particulate loading
E Fluidized-bed combustion Hot Small High particular loading but
modest cooling requirements
Source: Adapted from Becker, H. A. “Physical Probes.” In Instrumentation for Flows with Combustion, edited by
A.M.K.P. Taylor, 53–112. London, UK: Academic Press, 1993.
© 2011 by Taylor and Francis Group, LLC
30 Industrial Combustion Testing
in popularity, there are several technical limitations that
need to be considered (Becker 1993) [48]:
1. Fairly short reach due to the reduction of radi-
ant density fux with distance.
2. Radiant signal degradation by passage through
variable refractive index felds.
3. Inadequate signal-to-noise ratio in felds with
high particle loading.
4. Practical implementation problems in large-
scale systems such as industrial combustion.
Optical probes have become very popular for measuring
a wide range of quantities in laboratory fames. They have
several important advantages over physical probes. One is
that they are nonintrusive so they do not disturb the fows.
This can be very important in regions of high gradients
such as inside fames, where a physical probe could affect
the measurement being made. Another important advan-
tage is that optical probes often have very fast response
times so many measurements can be made in a short time
period. This can be critical in transient environments such
as turbulent fows. Optical probes often are not subject to
temperature limits because the sending and receiving
optics can be located outside of high temperature regions.
While it is usually possible to provide adequate cooling
for physical probes so they can be located in high tem-
perature environment, the increased probe size further
exacerbates the problem of intrusiveness. Some optical
probes can also generate instantaneous two-dimensional
profles of a given parameter such as temperature or spe-
cies concentration. Not only would these profles take
much longer with a physical probe that makes point mea-
surements, but there is also the problem that the compila-
tion of point measurements is only an average over some
period of time. In highly turbulent fows, these average
measurements may not adequately represent the actual
transient nature of the fuid dynamics.
1.4.3 Configurations
There are a range of confgurations that are used
in industrial combustion testing. The Gas Research
Institute (GRI) sponsored a program called Scaling 400
where measurements were made on a range of combus-
tors at multiple locations where the highest fring rate
tested was 400 times larger than the smallest rate [49].
These measurements ranged from laboratory to large
scale furnaces, although no feld measurements were
included. For convenience here, the range of typical test
confgurations has been divided into four general cate-
gories that are briefy discussed. While these categories
are somewhat arbitrary and not precisely defned, they
represent the types of combustion tests that have been
reported in the literature.
1.4.3.1 Bench Scale Testing
This refers to small size tests. While the majority of tests
done in academic institutions are generally at this scale,
this small size can also be useful for certain aspects of
industrial combustion testing. One common example of
this testing category is usually referred to as physical
modeling (see Chapter 10). Scale models may be made for
combustion systems such as a large scale power boiler.
Plexiglas construction permits easy visualization of the
combustion air fow to ensure the fow to each burner
is uniform and that there are no undesirable fow recir-
culations. Figure 1.45 shows an example of a Plexiglas
bench scale model. These small scale models are often
very detailed to capture the proper fuid dynamics.
Bench scale testing may also be used when testing a par-
ticular component in the system. For example, testing
a burner pilot or a new type of refractory can often be
done effectively at a very small scale. Testing with bench
scale models is much more cost effective than testing
full-scale systems. Fewer people are usually required to
Figure 1.45
Plexiglas bench scale model of a corner-fred utility boiler for fow
visualization. (Courtesy of John Zink Co. LLC.)
© 2011 by Taylor and Francis Group, LLC
Introduction 31
run the tests. The initial setup is generally much faster
and can usually be relatively quickly and easily modi-
fed. However, there may be scaling issues depending
on the system. To deal with that problem, physical mod-
eling may be done in conjunction with virtual model-
ing (see Chapter 11) to ensure that any scaling issues are
properly handled. Tests may be conducted over a range
of different size combustors [49] to determine what scal-
ing factors should be used when testing in a bench scale
unit. Another important limitation of bench scale testing
is that industrial end users are unlikely to adopt new
technology based solely on it. They normally want to see
technology demonstrated at or near full scale.
1.4.3.2 Pilot Scale Testing
These are somewhat larger than bench scale tests, but
not as large as full-scale tests. Pilot scale may refer to
all of the equipment scaled to some fraction of the full
size, or it may refer to a full-scale setup but of only part
of the actual equipment. In the former case, it may be
preferred to test equipment at, for example, half of the
full size because of the associated time and cost savings,
where scaling up to full size is well defned based on
past experience. In the latter case, a pilot scale furnace
may have full size burners, but fewer of them compared
to an actual production furnace. For example, it is com-
mon to test single full size process burners that will be
used in a production furnace that may have dozens of
those burners in it.
1.4.3.3 Large-Scale Testing
This refers to full size or nearly full size testing in some
type of test facility where the operating conditions can
be easily changed and controlled. An example would be
testing full size fares (see Chapter 28). These are very
diffcult to test in a production plant because it is almost
impossible to vary the waste gas fow rates and compo-
sitions going to the fare. Due to the critical safety nature
of fares, they need to be tested prior to installation in a
plant because they usually need to operate continually
until the next plant turn around that could be as long as
fve years or more. Premature failure of the fare could
mean shutting down an entire unit or possibly even the
whole plant, which would be very expensive and dis-
ruptive. In a fare test facility, the fare tip is often at or
near full size, but the fare stack is often much shorter
than in an actual plant. If the proper measurements are
made, such as fare radiation (see Chapter 30), then the
fare stack can be properly designed using experimen-
tal data from a much shorter stack. Full-scale tests are
generally the closest to an actual production environ-
ment that helps minimize or eliminate scaling issues.
They have an important advantage over feld testing in
a production plant because the conditions can be bet-
ter controlled, varied, and measured. Conditions that
may rarely occur in a plant can be easily simulated in
a test facility to ensure proper performance. However,
large-scale testing can be very expensive, especially if
a new apparatus must be built or signifcant modifca-
tions made to an existing test rig, to simulate the wide
range of equipment designs that exist in the feld. In
most practical applications, the large-scale test rig will
be close to, but probably not exactly like the actual pro-
duction equipment in the feld.
1.4.3.4 Field Testing
This is often the ultimate goal, but usually not until a
technology has been satisfactorily demonstrated in a
test facility. Few production plants are willing to test
unproven technologies because of the associated risk of
lost product. Field testing may be done incrementally.
For example, a new burner design could be tested by
replacing a small proportion of the burners in a heater
frst before making a complete replacement. This greatly
reduces the risk, although it also may mask the benefts
as well if too few burners are replaced to see any mea-
surable improvement. There are many potential prob-
lems related to feld testing. It is generally very diffcult
to vary the conditions enough to satisfy the researcher
who may only be able to test under a very limited set
of normal conditions. It may not be practical to simu-
late anomalous conditions that rarely occur in the plant.
Field combustion systems typically have much less
instrumentation than test furnaces so it may only be
possible to collect a limited amount of data. Field equip-
ment may not be calibrated as often as test equipment so
data accuracy may be an issue. Field equipment may be
much larger than test equipment, which means instru-
mentation access may be a problem. For example, get-
ting a gas temperature profle across the top of a large
furnace may prove challenging because of both the high
elevation and the long distance across the furnace. Field
experiments usually must be conducted while minimiz-
ing the impact on production, which means staying out
of the way of the operators. Despite the limitations of
feld testing, it offers the ultimate performance measure
for a technology since that is where the technology will
ultimate be used. A technology that works in the lab but
not in the feld may be of academic interest but will not
be of industrial interest.
1.4.4 important input Parameters
There are many input parameters that may be of inter-
est in industrial combustion testing, depending on the
nature of the technology. However, there are a relatively
small number of parameters that are typically important
© 2011 by Taylor and Francis Group, LLC
32 Industrial Combustion Testing
in most tests. Those parameters are briefy discussed
here. The output parameters of interest, for example
temperature and heat fux, are discussed in detail in
other chapters in this book and are not discussed here.
1.4.4.1 Fuel Composition
This is an important input parameter in certain tests,
while in other cases it varies very little and is therefore
not an important input parameter. For example, the fuel
composition can vary signifcantly for process burners
(see Chapter 18) and thermal oxidizers (see Chapter 33).
Process burners normally use the waste gases generated
in the plant as their fuel source. These waste gases may
have a dozen or more components in them that may
include both hydrocarbons (e.g., methane, propane, and
butane) and nonhydrocarbons (e.g., hydrogen, carbon
monoxide, nitrogen, and carbon dioxide). This can be
challenging to simulate in a test facility if the exact
blend composition must be matched. Fortunately, calcu-
lations and experience have shown that a small num-
ber of components can reasonably accurately simulate a
blend with many components, many of which are often
in very small quantities and can be safely ignored or
combined with other species. One of the key aspects of
process burner testing is demonstrating performance
over the range of fuels the burner is expected to use.
1.4.4.2 Air/Fuel Ratio
This is sometimes referred to as stoichiometry or equiv-
alence ratio and is often an important input test parame-
ter. While a combustion system will likely operate over a
relatively narrow range of air/fuel ratios, it is important
to demonstrate the combustion system’s capability over
a wider range of ratios. The air/fuel ratio is related to the
excess O
2
in the combustion products (see Chapter 7).
If the air/fuel ratio is increasing above stoichiometric
levels, then excess O
2
increases. If the air/fuel ratio is
decreasing to or below stoichiometric levels, then excess
O
2
decreases to levels approaching zero. As previously
discussed, one common test for a process burner (see
Chapter 18) is called CO breakthrough, where the excess
O
2
is continuously reduced until high levels of CO are
generated (which normally occurs at excess O
2
levels
close to zero). This test demonstrates safe performance
if the burners should run out of air. While that condition
is normally unlikely to occur, if it does occur the burner
needs to safely operate, although the fame may not be
pretty and the emissions may be high.
1.4.4.3 Geometry
There are several aspects of geometry that may be
important in industrial combustion testing. One impor-
tant geometrical factor is the burner design. Burner
vendors have extensive experimental facilities to test a
wide range of burner concepts. Numerous patents are
generated each year regarding new burner designs. A
primary focus for some time has been on reducing pol-
lution emissions, so today’s burners produce much less
pollution than the designs of older burners. Some of these
burners have components that can be adjusted, depend-
ing on the needs of the application. The various geom-
etries need to be tested to determine the optimal design
for a given set of operating conditions. Furnace design
is sometimes an important geometrical parameter. Some
burner manufacturers have a variety of test furnaces to
simulate the kinds of geometries their burners will see in
the feld. If an end user is witnessing a performance test
of a burner they have purchased, they will want to see it
tested in a furnace geometry similar to what they have
in their plant. Furnace designers may test various geom-
etries to try to improve performance (e.g., better thermal
effciency or increased production). Geometrical similar-
ity between an industrial combustion test and the actual
production system may be important. This means hav-
ing the right scale burner fring in a test chamber that is
representative of the combustor in the feld.
1.4.4.4 Furnace Temperature and Pressure
These are often important input parameters in combus-
tion testing. Furnace temperature has a direct effect on a
number of output parameters such as material process-
ing throughput rate, NOx emissions [50], and heat fux.
Realistic tests then normally need to match the furnace
temperature(s) that will be seen in the feld. This means
there needs to be some method for adjusting the temper-
ature in the test furnace. This is typically done by vary-
ing the heat load. One method that is commonly used
is water-cooled tubes or bayonets. Figure 1.46 shows an
Figure 1.46
(See color insert following page 424.) Water-cooled bayonets in-
serted into foor of a test furnace. (Adapted from Bussman, W.,
Baukal, C., and French, K., “Variable Test Furnace Cooling,” pre-
sented at the 2005 Summer Heat Transfer Conference, paper HT2005-
72012, San Francisco, CA, July 17–22, 2005.)
© 2011 by Taylor and Francis Group, LLC
Introduction 33
example of four water-cooled bayonets inserted into the
foor of a test furnace [51]. There are several variables that
can be adjusted to change the heat load. The insertion
depth into the furnace, the number of bayonets inserted,
and the water-cooling fow rate to each bayonet can all
be varied to change the furnace temperature. Another
method used to control test furnace temperature is to
vary the amount of insulation in front of a water-cooled
wall or tube. Adding insulation increases the furnace
temperature and removing insulation decreases it.
1.5 Experimental Errors
Experimental error is the difference between the true
value and the measured value. For example, if the true
temperature is 1000 K (1340°F) and the measured value is
950 K (1250°F), then the experimental error would be
50 K (90°F). This section considers the possible sources
for experimental errors, minimizing those errors, and
estimating the uncertainty in measurements.
1.5.1 Sources of error
It is important to quantify the uncertainty in measure-
ments made during testing so the data validity is under-
stood. Errors are associated with all testing, regardless
of the care taken by the experimenter [52]. One type of
error is due to some obvious mistakes by the experi-
menter. For example, bad data might be generated by fail-
ing to calibrate an instrument, such as a NOx analyzer,
prior to the start of the experiment, so this data should
be discarded. This does not mean that any data can be
thrown out just because they do not support the experi-
menter’s theory. There must be adequate justifcation for
eliminating outliers or other suspicious data.
There may be certain fxed errors, also called system-
atic errors, which cause repeated measurements to be in
error by approximately the same amount. For example,
if an instrument is incorrectly calibrated, then every
measurement may be off by approximately the same
amount. Other examples of systematic errors include
different experimenters taking measurements differ-
ently or some change to the environment such as a valve
leaking or left open that should have been closed.
There are some measurement techniques that have
an inherent error in the measured value compared to
the actual value. This is sometimes called a fxed error,
although this is somewhat of a misnomer as the amount
of the error usually varies with the reading. An example
of this type of error is using a thermocouple inserted
through the wall of a furnace to measure the gas tem-
perature inside the furnace. This produces a reading
that is generally lower than the actual value. The raw
measurement needs to be corrected using an appro-
priate calculation to account for the convection, con-
duction, and radiation to and from the thermocouple
junction inside the hot chamber. A suction pyrometer
is another instrument used to measure gas temperature
inside a furnace that automatically corrects for these
heat transfer effects. These are discussed in more detail
in Chapter 5.
Another type of error is referred to as random error. If
repeated measurements are made under the same condi-
tions, the measured values will vary somewhat. In gen-
eral, the variations will be normally distributed about
some mean value [53]. If the random error is high, then
the standard deviation of these repeated measurements
will be relatively large. If the random error is low, then
the standard deviation will be small with relatively little
variability. This random error is more diffcult to iden-
tify. If the experimenter knew the error, then it could be
corrected and would not be random. Therefore, the term
“uncertainty” is used because the sources of these ran-
dom errors are unknown. However, as will be shown
later in this section, this uncertainty can be estimated.
1.5.2 Minimizing errors
There are several ways that can be used to minimize
experimental error. The most important is simply to
be very deliberate and careful in the design and setup
of the experiment. While this may seem obvious, it is
sometimes overlooked due to time pressures and bud-
get constraints. For example, if a Gardon gauge is being
used to measure heat fux (see Chapter 6), then the gauge
must either be supplied with a calibration curve from
the supplier that shows the heat fux as a function of the
electrical signal, or it must be calibrated by the experi-
menter. Even if the gauge came with a calibration curve,
it may be prudent to have it recalibrated if the gauge has
not been used in some time.
It is also recommended that the experimenter thor-
oughly analyze what is being collected to ensure the
necessary data will be available for the analysis that
will be done. For example, if the thermal effciency of
a combustion process will be calculated, the experi-
menter needs to measure both the composition and
the temperature of the exhaust products, among other
variables such as the fuel fow rate and composition, the
combustion air fow rate, the furnace pressure, and the
furnace skin temperature. If the actual water content in
the exhaust products is not measured, which is often the
case, it can be calculated knowing the other components
in the stream and the fuel composition and fow rate.
The furnace air leakage can be calculated based on the
fue gas composition and combustion air fow rate. The
point is that the experimenter should carefully check
© 2011 by Taylor and Francis Group, LLC
34 Industrial Combustion Testing
the analysis that will be done to ensure that the neces-
sary data will be collected.
Another fairly obvious method that should be used to
minimize experimental error is frequent calibration. For
example, if an analyzer is used to measure the exhaust
gas composition, the analyzer should be calibrated at
least daily with a suitable span gas cylinder. It would
even be better to calibrate the analyzer more than once
a day. For example, it could be calibrated at the begin-
ning of the day, in the middle of the day, and at the end
of the day. If these multiple calibrations consistently
show negligible drift, then one calibration per day may
be justifed.
Related to the frequent calibration is using the appro-
priate measurement scale. Again using the example of a
gas analyzer, many of these instruments have multiple
scales for measuring gas composition. For example, an
oxygen analyzer might have multiple scales such as
0–1%, 0–5%, 0–10%, 0–25% and 0–100%. The appropri-
ate scale should be used when making measurements
to minimize the error that is often a percent of the full-
scale reading. In most typical industrial combustion
tests, the excess O
2
in the exhaust gases is generally
2–4%, depending on the type of fuel, the burner design,
and several other parameters. Then, the appropriate
scale to use would be 0–5%. If the 0–100% scale were
used instead and the error in the reading is 2% of the full
scale, then the error in using the 0–100% scale would be
2%. This means that a 3% excess O
2
reading on a 0–100%
scale with a 2% error could actually be anywhere from
1 to 5%, which is unacceptable accuracy. However, if the
0–5% scale were used, the error would be 2 of 5%, which
is 0.1%. This means that a 3% excess O
2
reading with a
2% error on a 0–5% scale could actually be anywhere
from 2.9 to 3.1%, which is much better accuracy than
1–5% for the 0–100% scale.
It is important that a suitable calibration standard be
used for the measurement technique. As an example, for
gas analyzer calibration, it is generally recommended
that the calibration gas be 80–90% of the full-scale
reading. For example, if the 0–5% scale will be used
to measure O
2
, then the calibration gas should have
approximately 80–90% of the full-scale reading of 5%,
which means the calibration gas should have 4.0–4.5%
O
2
. One reason to specify slightly less than the full scale
is that the span gas supplier may be slightly off in their
gas blending and normally has some allowable toler-
ance in the gas composition they supply. For example,
if the supplier has a 5% tolerance in the actual blend
they supply and the experimenter requests a blend with
5% O
2
, the actual span gas supplied could be as high as
5.25% O
2
, which is above the range of the 0–5% scale and
therefore could not be used to span that scale.
Wherever possible, instruments should be calibrated
in place when they are connected to the measurement
system. For example, if a thermocouple is being used
to measure temperature, ideally it should be calibrated
with a portable calibrator while it is connected to the
data acquisition system. If necessary, the thermocouple
can be temporarily removed from its measurement loca-
tion to insert it into the calibrator. The important factor is
that it should be electrically connected to the data acqui-
sition system. This will help identify obvious mistakes
such as mistakenly setting up an actual Type R thermo-
couple as a Type K thermocouple in the data acquisition
system. If the thermocouple was disconnected, removed
from the system and calibrated somewhere else, this
mistake may not be immediately recognized, which
would result in bad data. This is a systematic error that
could be corrected after the fact with a proper conver-
sion. Some other errors may not be as easy to correct
after the experiments are completed.
Another way to minimize experimental error is to
use multiple methods to measure the same variable.
Suppose the experimenter is trying to measure the draft
(pressure) in a process heater. There are several devices
commonly used to measure draft. Figure 1.47 shows an
inclined manometer, which is typically flled with red
gauge oil that is manually read off a scale. Figure 1.48
shows a mechanical gauge where a needle defects and
the reading is manually read off a gauge. Figure 1.49
shows an electronic manometer where a transducer
sends an electronic signal that can be displayed on a
computer screen. The experimenter might use two or
three of these devices to measure the draft, or maybe
only one of them that is periodically compared against
another one to ensure accuracy. For example, the elec-
tronic manometer might be used for ease of data collec-
tion, but manual readings might be taken periodically
to corroborate the electronic readings.
When possible, it is good practice to make repeated
readings of the same experimental conditions. For
Figure 1.47
Inclined manometer.
© 2011 by Taylor and Francis Group, LLC
Introduction 35
example, assume that test condition #1 is the base case.
If only one day’s worth of testing will be done, then the
base case conditions should be tested at least twice but
not consecutively. Maybe the frst and last tests of the
day are the base case. More replications would be even
better but time and budget constraints may preclude
more than one replication. If the testing will take place
over multiple days, then the base case may be replicated
on several of those days. This data will give the exper-
imenter an idea of the repeatability of the tests. If the
same test done at two different times on the same day
gives wildly different results, then the experimenter
will have little confdence in the data and should check
for sources of experimental error that can be fxed or
appropriate corrections can be made if the actual test
conditions are not exactly the same. For example, it has
been shown that ambient air humidity can signifcantly
impact NOx measurements [54]. If the base case condi-
tions were tested in the morning when the humidity
was low and then again late in the afternoon when the
humidity was high, then the NOx readings are likely
to be somewhat different. Therefore, the experimenter
should measure the humidity to account for this effect
on NOx. While the experimenter cannot control the
humidity, the measured humidity can be used to com-
pare NOx measurements made under different humid-
ity conditions. Chapter 3 discusses general principles
of experimental design that can be used to effciently
collect data with a minimum number of tests, with
suffcient replications to validate the data measurement
and collection techniques.
1.5.3 uncertainty analysis
Kline and McClintock (1953) developed a method to esti-
mate the uncertainty in experimental data [55]. In some
cases, some primary measurements like temperature,
pressure, and fow are combined to calculate another
result:
R R x x x x
n
( )
1 2 3
, , , , … , (1.4)
where R = the result of interest and x
i
are the variables
involved in the calculation. Then, the uncertainty in the
result R (δR) is calculated using:
δ δ δ δ R
R
x
x
R
x
x
R
x
n



|
.

`
}

+


|
.

`
}

+ +


1
1
2
2
2
2
… xx
n
|
.

`
}

]
]
]
]
2
1 2
(1.5)
where ∂R/∂x
i
is the partial derivative of the equation for
R with respect to the variable x
i
and δx
i
is the uncer-
tainty in the parameter x
i
.
An example will be used to demonstrate the tech-
nique. In the study by Baukal (2009), the heat fux from
an impinging fame was measured using a disk calo-
rimeter segmented into six concentric sections as shown
Figure 1.48
Capsuhelic gauge. (From Baukal, C. E., Industrial Combustion Pollution
and Control, New York: Marcel Dekker, 2004.)
Figure 1.49
Electronic pressure gauge. (From Baukal, C. E., Industrial Combustion
Pollution and Control, New York: Marcel Dekker, 2004.)
© 2011 by Taylor and Francis Group, LLC
36 Industrial Combustion Testing
in Figure 1.50 [56]. The total heat fux (q
i
”) to each calo-
rimeter section, i, was determined as follows:
′′

q
mc t t
A
i
i p
i
i i

( )
,
out in
(1.6)
where

m
i
was the cooling water mass fow rate, c
p
was
the specifc heat of the cooling water, t
out
i
was the out-
let temperature of the cooling water, t
in
i
was the inlet
temperature of the cooling water, and A
i
was the ring
impingement surface area. The total heat fux was cal-
culated using Equation 1.6. The uncertainty was calcu-
lated using:

δ δ δ ′′
∂ ′′

×
|
.

`
}

+
∂ ′′

×
|
.

`
}

q
q
m
m
q
c
c
p
p


2
2

+
∂ ′′

×
|
.

`
}

+
∂ ′′

×
|
.
q
t
t
q
t
t
in
in
out
out
δ δ
2

`
}

+
∂ ′′

×
|
.

`
}

]
]
]
]
2 2
1 2
q
A
A δ ,

(1.7)
where δ

m = 0.3%, δc
p
 = ±0.1%, δt
in
and δt
out
were deter-
mined from calibrations, and δA was determined
using:
δ


δ


δ A
A
d
d
A
d
d
|
.

`
}

+
inner
inner
outer
out
× ×
2
eer
|
.

`
}

]
]
]
]
2
1 2
, (1.8)
where d
inner
and d
outer
were the inner and outer diameters
of each calorimeter ring, respectively. The maximum
uncertainty in the heat fux was 58%. This occurred at
the lowest fring rate (q
f
 = 5 kW) and oxidizer composi-
tion (O
2
fraction in the oxidizer = Ω = 0.21 for air), the
maximum axial spacing (nondimensional distance L = 6)
and radial spacing (nondimensional radius R
eff
 = 0.59.
There, a high water cooling fow rate was used to reduce
the high surface temperatures as much as possible. The
uncertainty was so high because the temperature dif-
ference between the inlet and outlet cooling water tem-
peratures was only 0.58 K. For the same fring rate and
radial location, only a slightly higher oxidizer compo-
sition (Ω = 0.35) and a closer axial spacing (L = 0.5), the
uncertainty dropped to 15%. The maximum uncertainty
for all other calorimeters was 12%. Again that occurred
at the lowest fring rate, where the temperature differ-
ences were smaller. For the most commonly used fring
rate (q
f
 = 15 kW), the maximum uncertainty for any calo-
rimeter was 9.2%. Typical uncertainties were 6–8%.
1.6 Combustion Testing Resources
1.6.1 general references
Specifc references are provided with the various tech-
niques discussed in other chapters in this book. Those
references give more details for the interested reader.
Several excellent general references have been pub-
lished concerning combustion diagnostics. Surugue
(1961) edited a manual (in both English and French)
for the Advisory Group for Aeronautical Research and
Development (part of NATO) that contains papers on
making experimental measurements in combustion
[57]. The book by Beér and Chigier (1972) gives detailed
descriptions and diagrams of different probes (e.g., sur-
face and suction pyrometers, ellipsoidal radiometer,
gas sampling, pitot tube) used to make measurements
in industrial-scale fames [58]. This book is particu-
larly helpful for those planning to use physical probes
as the authors describe in detail the design of various
probes and how to make specifc measurements includ-
ing calibration and error correction if appropriate. Zinn
(1977) edited a comprehensive book on experimental
diagnostics in gas phase combustion systems, which
includes both probe and optical techniques [59]. Gupta
and Lilley (1985) describe how to make fow feld mea-
surements, including fow visualization, in combustion
[60]. The report by Okoh and Brown (1988) [61] reviews
and compares many techniques for the measurement
of a wide range of variables, in combustion systems. It
also includes equipment specifcations and suppliers.
The books edited by Durão et al. (1992) [62] and Taylor
(1993) [63] contain chapters on the use of both physi-
cal probes and optical techniques. Fristrom (1995) has
2 1 3 4 5 6
Figure 1.50
Disk calorimeter segmented into six concentric sections. (From
Baukal, C. E., Heat Transfer from Flame Impingement Normal to a Plane
Surface, Saarbrücken, Germany: VDM Verlagsservicegesellschaft
mbH, 2009.)
© 2011 by Taylor and Francis Group, LLC
Introduction 37
several chapters devoted to probe and optical mea-
surements in fames [64]. The papers and chapters by
Fristrom (1976) [65], Bowman (1977) [66], Gouldin (1980)
[67], and Becker (1993) [48] review probe measurements
in fames. Newbold et al. (1996) gave a good example of
making probe measurements of velocity, species, radia-
tion, and gas temperature in an industrial, gas-fred,
fat-glass furnace [68]. Mullinger and Jenkins (2008) [22]
describe many types of physical probes and present rep-
resentative results of measurements made in industrial
processes.
There are many advanced, mostly laser-based, tech-
niques that have been successfully used in labora-
tory fames. These techniques include, for example,
Raman and Rayleigh scattering, coherent antistokes
Raman scattering (CARS), degenerate four wave mix-
ing (DFWM), Mie scattering, laser-induced fuorescence
(LIF), laser-induced incandescence (LII), resonantly
enhanced multiphoton ionization (REMPI), tunable
diode laser (TDL) absorption, and laser Doppler veloci-
metry (LDV). Solomon et al. (1986) reviewed Fourier
Transform Infrared (FT-IR) optical techniques for mea-
suring particles, species concentrations, and tempera-
ture in combustion processes [69]. The books by McCay
and Roux (1984) [70], Eckbreth (1988) [71], Chigier (1991)
[72], and Kohse-Höinghaus and Jeffries (2002) [73] focus
specifcally on optical techniques for diagnosing fames.
Papers are available on the use of laser diagnostics in
combustion systems [74–79]. Advanced optical tech-
niques permit spatially, temporally, and spectrally
resolved measurements that are diffcult if not impos-
sible to make with traditional probes. Many of these
advanced techniques have proven to be a challenge in
industrial-scale fames and are generally beyond prac-
tical implementation by most industrial combustion
researchers. The cost of implementing these techniques
on a large scale is often prohibitive. The environmental
conditions such as high temperatures and vibration also
make implementation more diffcult in large-scale com-
bustors. Therefore, most of the techniques described in
this book concern physical probes that have been proven
in large-scale combustion systems, although there are
some chapters dedicated to optical techniques as well.
1.6.2 Test Facilities
There are relatively few industrial-scale combustion
test facilities outside of equipment manufacturers. A
report prepared from a workshop sponsored by the
U.S. Department of Energy states, “For the most part,
the size and type of laboratory test equipment avail-
able are inadequate, and the costs are prohibitive” ([4],
p. 11). This section is intended to provide a sample of
some of those industrial-scale test facilities, although
there are others as well. They are broadly categorized
as academic and institutional. While this is somewhat
arbitrary, academic refers to universities (web addresses
generally have an “edu” suffx) and institutional refers
to organizations that are not universities (although
some are affliated with universities) and are typically
not-for-proft (web addresses generally have an “org”
suffx). Some of the authors in this book work for some
of these organizations that are listed in alphabetical
order in each category.
1.6.2.1 Academic
There are not many universities that conduct industri-
al-scale research, mostly because of the cost of build-
ing and operating such facilities. Adelaide University
(Adelaide, Australia) (www.adelaide.edu.au) has done
some research on large-scale fares in its Turbulence,
Energy & Combustion (TEC) group. The TEC consists of
members from the mechanical and chemical engineer-
ing schools.
Brigham Young University (BYU) in Provo, Utah
(www.byu.edu) and the University of Utah (U of U) in
Salt Lake City, Utah (www.utah.edu) formed a coop-
erative consortium called the Advanced Combustion
Engineering Research Center (ACERC; www-acerc.byu.
edu). The mission of ACERC is “to develop, apply, and
transfer advanced combustion technology to industry
through fundamental engineering research and educa-
tional programs aimed at the solution of critical com-
bustion problems.” The ACERC has an international
reputation for conducting applied research in a wide
range of areas such as coal, biomass, black liquor, for-
est fres, and gas turbines. The University of Utah has a
number of industrial-scale combustors used for making
a wide range of measurements.
Cranfeld Institute of Technology (Cranfeld, United
Kingdom; www.cranfeld.ac.uk) has a gas turbine engi-
neering lab in an off-campus site with multiple test cells
for conducting large-scale research as part of its energy
program.
Osaka Gas in Osaka, Japan (www.osakagas.co.jp) and
Tokyo Gas in Tokyo, Japan (www.tokyo-gas.co.jp) both
conduct a wide range of research for industrial combus-
tion applications. Some examples include cogeneration,
advanced fuel-air mixing, gas turbines, oxy-natural gas
combustion, and biogas.
Università di Pisa in Pisa, Italy (www.unipi.it) is asso-
ciated with the IFRF that is discussed below in the next
section.
University of California at Irvine in Irvine, California
(www.uci.edu) has done research on industrial burners
using natural gas in their UCI Combustion Laboratory
(UCICL; www.ucicl.uci.edu). The UCICL has also done
research on active control, emissions, gas turbine com-
bustion, and liquid fuels and sprays.
© 2011 by Taylor and Francis Group, LLC
38 Industrial Combustion Testing
1.6.2.2 Institutional
There are a number of well-known institutions that
conduct industrial-scale combustion testing. CANMET
Energy Technology Centre is located in Ottawa, Ontario
(Canada; canmetenergy-canmetenergie.nrcan- rncan.
gc.ca) and is part of Natural Resources Canada. CANMET
is a leader in clean energy research and technology devel-
opment and develops technologies in many industrial
areas including, for example, furnaces, boilers, oxy-fuel
combustion, process integration, and NOx reduction.
Chapters 11 and 13 discuss some of the modeling and
experimental data generated by CANMET.
The ENEA headquartered in Rome, Italy (www.enea.
it) is the Italian National Agency for New Technologies,
Energy and the Environment. ENEA has labs located in
multiple locations in Italy and conducts a wide range
of industrial combustion ranging in many application
areas with particular emphasis on the environment, cli-
mate change, and sustainability. Chapter 12 discusses
advanced laser measurements made at ENEA.
Gas Technology Institute (GTI) located in Des Plaines,
Illinois (www.gastechnology.org) has been a leader in
the development of technology solutions in the natural
gas industry for more than 65 years. The institute has a
wide range of industrial-scale test capabilities and has
done testing in essentially all major application areas.
They have developed numerous technologies such as a
forced internal recirculation burner, oxygen-enriched
air staging, methane de NOx®, a high-luminosity oxy-
gas burner, and the super boiler. Some experimental
results from GTI are discussed in Chapter 19.
The International Flame Research Foundation (IFRF;
www.ifrf.net) in Livorno, Italy has long been recognized
as one of the premier organizations for industrial com-
bustion research. The IFRF has published numerous
research studies over the many decades of its existence.
The foundation has conducted research in its own facili-
ties and in conjunction with many of its members in a
wide range of industries including, for example, ferrous
and nonferrous metals, cement, lime, biomass, and boil-
ers. The IFRF has an online handbook that may be of
interest to researchers.
1.7 Future
Industrial combustion is expected to continue to play a
prominent role in the future. It is directly related to the
vast majority of energy consumption in the world and
indirectly related to many of the products people use on a
daily basis. The increasing attention on the environment
means that industrial combustion pollution emissions
are expected to continue to reduce over time. Energy eff-
ciency is also expected to be of interest as well.
A wide range of testing confgurations will continue
to be used to simulate realistic operating conditions. It is
expected that feld testing in particular, with all of its lim-
itations, will continue to be an important component of a
combustion technology development program, especially
since most end users will be unwilling to use a technol-
ogy that has not been verifed on a large scale, preferably
in an operating plant. Laboratory testing will also con-
tinue to be important because of its convenience, cost-
effectiveness, controllability, and accuracy. Initial testing
can be done more quickly, less expensively, with better
control, and with more accuracy than feld testing. New
technology development typically needs to be proven in
the laboratory frst before proceeding to feld testing.
The use of optical instrumentation is expected to
increase in the future because it has some important
advantages compared to physical probes. The main bar-
rier to using many of the optical techniques has been
adapting them for use in an industrial environment.
More experience continues to be gained using these
sophisticated techniques in the harsh environments
often present in industry.
Industrial combustion testing should continue to be
important well into the foreseeable future. While com-
puter modeling continues to grow in popularity, cur-
rent models can not replicate the complete physics of
an industrial combustion system because of computer
hardware limitations and because not all of the physics
are completely known. In nearly all models, simplifca-
tions need to be made to get results in a reasonable
amount of time. In addition, those models still need
experimental data for validation. Testing is needed
because of the complexity of combustion problems. It is
expected that computer modeling and industrial scale
testing will be used together as complementary tools to
further improve industrial combustion technologies.
References
1. U.S. Department of Energy (DOE). Industrial Combustion
Vision: A Vision by and for the Industrial Combustion
Community. Washington, DC: U.S. DOE, 1998.
2. Hewitt, G. F., Shires, G. L., and Bott, T. R. Process Heat
Transfer. Boca Raton, FL: CRC Press LLC, 1994.
3. U.S. Department of Energy, Energy Information
Administration. Annual Energy Outlook 2008, report DOE/
EIA-0384 (2008), Washington, DC, released June 26, 2009.
4. Energetics, Inc. Industrial Combustion Technology Roadmap:
A Technology Roadmap by and for the Industrial Combustion
Community, October 2002, www1.eere.energy.gov/
industry/combustion/pdfs/combustion_roadmap2002.
pdf (accessed July 29, 2009).
© 2011 by Taylor and Francis Group, LLC
Introduction 39
5. Barry, P., and Somers, S. “Duct Burners.” In The John Zink
Combustion Handbook, edited by C. E. Baukal, 523–44.
Boca Raton, FL: CRC Press, LLC, 2001.
6. Baukal, C. E., ed. Oxygen-Enhanced Combustion. Boca
Raton, FL: CRC Press LLC, 1998.
7. Strehlow, R. A. Fundamentals of Combustion. Scranton, PA:
Inter. Textbook Co., 1968.
8. Williams, F. A. Combustion Theory. Menlo Park, CA:
Benjamin/Cummings Publishing, 1985.
9. Lewis, B., and von Elbe, G. Combustion, Flames and Explo-
sions of Gases. 3rd ed. New York: Academic Press, 1987.
10. Bartok, W., and Sarofm, A. F., eds. Fossil Fuel Combustion.
New York: John Wiley & Sons, Inc., 1991.
11. Fristrom, R. M. Flame Structure and Processes. New York:
Oxford University Press, 1995.
12. Glassman, I. Combustion. 3rd ed. New York: Academic
Press, 1996.
13. Edwards, J. B. Combustion: The Formation and Emission
of Trace Species. Ann Arbor, MI: Ann Arbor Science
Publishers, 1974.
14. Barnard, J. B., and Bradley, J. N. Flame and Combustion.
2nd ed. London: Chapman and Hall, 1985.
15. Turns, S. R. An Introduction to Combustion. 2nd ed. New
York: McGraw-Hill, 2000.
16. Griswold, J. Fuels, Combustion and Furnaces. New York:
McGraw-Hill, 1946.
17. Stambuleanu, A. Flame Combustion Processes in Industry.
Tunbridge Wells, UK: Abacus Press, 1976.
18. Perthuis, E. La Combustion Industrielle. Paris: Éditions
Technip, 1983.
19. Borman, G., and Ragland, K. Combustion Engineering.
New York: McGraw-Hill, 1998.
20. Deshmukh, Y. V. Industrial Heating. Boca Raton, FL: CRC
Press, 2005.
21. Keating, E. L. Applied Combustion. Boca Raton, FL: CRC
Press, 2007.
22. Mullinger, P., and Jenkins, B. Industrial and Process
Furnaces. Oxford, UK: Butterworth-Heinemann, 2008.
23. Segeler, C. G., ed. Gas Engineers Handbook. New York:
Industrial Press, 1965.
24. Reed, R. D. Furnace Operations. 3rd ed. Houston, TX: Gulf
Publishing, 1981.
25. Pritchard, R., Guy, J. J., and Connor, N. E. Handbook
of Industrial Gas Utilization New York: Van Nostrand
Reinhold, 1977.
26. Reed, R. J. North American Combustion Handbook. Vol. I,
3rd ed. Cleveland, OH: North American Mfg. Co., 1986.
27. IHEA. Combustion Technology Manual. 5th ed. Arlington,
VA: Industrial Heating Equipment Assoc., 1994.
28. Kistler, M. D., and Becker, J. S. “Ferrous Metals.” In
Oxygen-Enhanced Combustion, edited by C. E. Baukal.
Boca Raton, FL: CRC Press, 1998.
29. Saha, D., and Baukal, C. E. “Non-Ferrous Metals.” In
Oxygen-Enhanced Combustion, edited by C. E. Baukal.
Boca Raton, FL: CRC Press, 1998.
30. Fedler, R. M., and Rousseau, R. W. Elementary Principles
of Chemical Process. 3rd ed. New York: John Wiley & Sons,
Inc., 2000.
31. Kudra, T., and Mujumdar, A. S. Advanced Drying
Technologies. New York: Marcel Dekker, 2002.
32. Baukal, C. E. Heat Transfer in Industrial Combustion. Boca
Raton, FL: CRC Press, 2000.
33. Reese, J. L., Moilanen, G. L., Borkowicz, R., Baukal, C.,
Czerniak, D., and Batten, R. “State-Of-The-Art of NOx
Emission Control Technology.” ASME paper 94-JPGC-
EC-15, Proceedings of International Joint Power Generation
Conference, Phoenix, AZ, October 3–5, 1994.
34. Joshi, M. L., Tester, M. E., Neff, G. C., and Panahi, S. K.
“Flame Particle Seeding with Oxygen Enrichment for
NOx Reduction and Increased Effciency.” Glass 68, no. 6
(1990): 212–13.
35. Ruiz, R., and Hilliard, J. C. “Luminosity Enhancement
of Natural Gas Flames.” Proceedings of 1989 International
Gas Research Conference, edited by T. L. Cramer, 1345–53.
Rockville, MD: Govt. Institutes, 1990.
36. Slavejkov, A. G., Gosling, T. M., and Knorr, R. E. “Low-
NOx staged combustion device for controlled radia-
tive heating in high temperature furnaces.” U.S. patent
5,611,682, March 18, 1997.
37. API Publication 535. Burner for Fired Heaters in General
Refnery Services. 1st ed. Washington, DC: American
Petroleum Institute, July 1995.
38. Fioravanti, K. J., Zelson, L. S., and Baukal, C. E. “Flame
Stabilized Oxy-Fuel Recirculating Burner.” U.S. Patent
4,954,076, September 4, 1990.
39. Baukal, C. E., ed. The John Zink Combustion Handbook.
Boca Raton, FL: CRC Press, 2001.
40. Joshi, S. V., Becker, J. S., and Lytle, G. C. “Effects of
Oxygen Enrichment on the Performance of Air-Fuel
Burners.: In Industrial Combustion Technologies, edited by
M. A. Lukasiewicz, 165. Materials Park, OH: American
Society of Metals, 1986.
41. Longacre, S. “Using Infrared to Dry Paper and Its
Coatings.” Process Heating 4, no. 2 (1997): 45–49.
42. Smith, T. M., and Baukal, C. E. “Space-Age Refractory
Fibers Improve Gas-Fired Infrared Generators for Heat
Processing Textile Webs.” Journal of Coated Fabrics 12, no.
3 (January 1983): 160–73.
43. Reed, R. J. “Future Consequences of Compact, Highly
Effective Heat Recovery Devices.” In Heat Transfer in
Furnaces, edited by C. Presser and D. G. Lilley, 23–28, Vol.
74. New York: ASME HTD, 1987.
44. Baukal, C. E., Gershtein, V. Y., and Li, X. M., eds.
Computational Fluid Dynamics in Industrial Combustion.
Boca Raton, FL: CRC Press, 2001.
45. Bussman, W., and Baukal, C. “Ambient Condition
Effects on Process Heater Emissions.” Proceedings of the
International Mechanical Engineering Congress & Exhibition,
Paper IMECE2008-68284, Boston, MA, November 2008.
46. Baukal, C. E. Industrial Combustion Pollution and Control.
New York: Marcel Dekker, 2004.
47. Baukal, C. E., ed. Industrial Burners Handbook. Boca Raton,
FL: CRC Press, 2004.
48. Becker, H. A. “Physical Probes.” In Instrumentation for
Flows with Combustion, edited by A.M.K.P. Taylor, 53–112.
London, UK: Academic Press, 1993.
49. Weber, R. “Scaling Characteristics of Aerodynamics, Heat
Transfer, and Pollutant Emissions in Industrial Flames.”
26th Symposium (International) on Combustion, The
Combustion Institute, Pittsburgh, PA, pp. 3343–54, 1996.
© 2011 by Taylor and Francis Group, LLC
40 Industrial Combustion Testing
50. Bussman, W., and Baukal, C. “The Effect of Firebox
Temperature on NOx Emissions.” Presented at 2004
Air & Waste Management Association annual meeting,
paper #04-A-664-AWMA, Indianapolis, IN, June 2004.
51. Bussman, W., Baukal, C., and French, K. “Variable Test
Furnace Cooling.” Presented at the 2005 Summer Heat
Transfer Conference, paper HT2005-72012, San Francisco,
CA, July 17–22, 2005.
52. Holman, J. P. Experimental Methods for Engineers. 5th ed.
New York: McGraw-Hill, 1989.
53. Harrison, S. J., and Dieck, R. H. “Measurement Errors
and Uncertainty.” In The Engineering Handbook, edited
by R. C. Dorf, 2nd ed. Boca Raton, FL: CRC Press,
2005.
54. Bussman, W., and Baukal, C. “Ambient Conditions
Impact CO and NO
x
Emissions: Part I.” Petroleum
Technology Quarterly 14, no.3 (2009): 93–99.
55. Kline, S. J., and McClintock, F. A. “Describing
Uncertainties in Single-Sample Experiments.” Mechanical
Engineering (January 1953): 3–8.
56. Baukal, C. E. Heat Transfer from Flame Impingement
Normal to a Plane Surface. Saarbrücken, Germany: VDM
Verlagsservicegesellschaft mbH, 2009.
57. Surugue, J., ed. Experimental Methods in Combustion
Research: A Manual. New York: Pergamon Press, 1961.
58. Beér, J. M., and Chigier, N. A. Combustion Aerodynamics.
London: Applied Science Publishers, 1972.
59. Zinn, B. T., ed. Experimental Diagnostics in Gas Phase
Combustion Systems. Washington, DC: American Institute
of Aeronautics and Astronautics, 1977.
60. Gupta, A. K., and Lilley, D. G. Flowfeld Modeling and
Diagnostics. Tunbridge Wells, UK: Abacus Press, 1985.
61. Okoh, C. I., and Brown, R. A. Combustion Experimentation
Handbook. Chicago, IL: Gas Research Institute report
GRI-88/0143, 1988.
62. Durão, D. F. G., Heitor, M. V., Whitelaw, J. H., and
Witze, P. O. Combusting Flow Diagnostics. Dordrecht, The
Netherlands: Kluwer Academic Publishers, 1992.
63. Taylor, A.M.K.P. Instrumentation for Flows with Combustion.
London: Academic Press, 1993.
64. Fristrom, R. M. Flame Structure and Processes. New York:
Oxford University Press, 1995.
65. Fristrom, R. M. “Probe Measurements in Laminar
Combustion Systems.” In Combustion Measurements:
Modern Techniques and Instrumentation, edited by R.
Goulard. New York: Academic Press, 1976.
66. Bowman, C. T. “Probe Measurements in Flames.” Progress
in Astronautics and Aeronautics 53, (1977): 1–24.
67. Gouldin, F. C. “Probe Measurements in Multi-Dimensional
Reacting Flows.” In Testing and Measurement Techniques in
Heat Transfer and Combustion, AGARD CP-281, paper #4,
1980.
68. Newbold, J., McQuay, M. Q., Webb, B. W., and Huber,
A. M. “The Experimental Characterization of the
Combustion Process in an Industrial, Gas-Fired, Flat-
Glass Furnace.” 29th International ISATA Conference:
Automotive Technology & Automation, Florence,
Italy, Automotive Association Ltd., Vol. 2, 967–76, June
1996.
69. Solomon, P. R., Best, P. E., Carangelo, R. M., Markham,
J. R., Chien, P-L, Santoro R. J., and Semerjian, H. J.
“FT-IR Emission/Transmission Spectroscopy for
In Situ Combustion Diagnostics.” 21st Symposium
(International) on Combustion, The Combustion
Institute, Pittsburgh, PA, 1763–71, 1986.
70. McCay, T. D., and Roux, J. A. Combustion Diagnostics by
Nonintrusive Methods. New York: American Institute of
Aeronautics and Astronautics, 1984.
71. Eckbreth, A. C. Laser Diagnostics for Combustion
Temperature and Species. Cambridge, MA: Abacus Press,
1988.
72. Chigier, N. Combustion Measurements. New York:
Hemisphere Publishing Corp., 1991.
73. Kohse-Höinghaus, K., and Jeffries, J. B., eds. Applied
Combustion Diagnostics. New York: Taylor & Francis,
2002.
74. Eckbreth, A. C. “Recent Advances in Laser Diagnostics
for Temperature and Species Concentrations in
Combustion.” 18th Symposium (International) on
Combustion, The Combustion Institute, Pittsburgh, PA,
1471–88, 1980.
75. Penner, S. S., Wang, C. P., and Bahadori, M. Y. “Laser
Diagnostics Applied to Combustion Systems.” 20th
Symposium (International) on Combustion, The
Combustion Institute, Pittsburgh, PA, 1149–76, 1984.
76. Hanson, R. K. “Combustion Diagnostics: Planar Imaging
Techniques.” 21st Symposium (International) on
Combustion, The Combustion Institute, Pittsburgh, PA,
1677–91, 1986.
77. Kohse-Höinghaus, K. “Laser Techniques for the
Quantitative Detection of Reactive Intermediates in
Combustion Systems.” Progress in Energy and Combustion
Science 20 (1994): 203–79.
78. Fornaciari, N. R., Schefer, R. W., Walsh, P. M., and
Claytor, L. E. “Application of Laser-Based Diagnostics to
Industrial Scale Burners.” Proceedings of 1995 International
Gas Research Conference, edited by D. A. Dolenc, 2398–
2405. Rockville, MD: Govt. Institutes, 1996.
79. Perrin, M., Imbach, J., Albert, S., Mariasine, J.,
and Quinqueneau, A. “Application of Advanced
Instantaneous In-Flame Measurements Techniques in
an Industrial Flame with Preheated Air.” Proceedings of
1995 International Gas Research Conference, edited by D. A.
Dolenc, 2406–15. Rockville, MD: Govt. Institutes, 1996.
© 2011 by Taylor and Francis Group, LLC
41
2.1 Introduction
Industrial combustion testing can be dangerous for
many reasons. As stated in Chapter 1, one of the rea-
sons for testing is to try new concepts as part of research
and development. This often means that there is some
uncertainty in the result, hence the need for testing. In
some cases, this uncertainty could mean potentially
dangerous conditions. For example, fameless combus-
tion (see Chapter 23) could be dangerous because the
combustor must be above the autoignition temperature
of the air–fuel mixture or an explosion could occur. It is
also potentially dangerous because fameless combus-
tion is usually diffcult to detect with traditional optical
fame detectors. This means an alternative method, such
as furnace temperature, may be needed to shut off the
fuel supply to the furnace if the fame goes out.
Fires and explosions are a major concern in industrial
combustion processes and account for as much as 95%
of the losses in accidents in the process industries [1].
Figure 2.1 shows the bulging walls in a test heater that
resulted from an overpressure event caused by a rapid
defagration. The consequences of a fre or an explosion
in a chemical or petrochemical plant, for example, can be
very severe and very public because of the high volume
of fammable liquids and gases handled in those plants
[2,3]. Burning large quantities of fuel means appropriate
precautions must be taken to prevent equipment dam-
age and personnel injury.
There are many factors that can contribute to an
accident [4]: human error [5,6], equipment malfunc-
tion, upset conditions, fre or explosion near the
apparatus [7], improper procedures, and severe weather
conditions. There are also many potential dangers
caused particularly by fres and explosions such as fy-
ing shrapnel, pressure waves from a blast, high heat
loads from fame radiation [8–10], and high tempera-
tures. All of these can have severe consequences to both
people and equipment and may need to be considered
in minimizing the potential impact of an incident.
Kletz (1998) [11] lists four circumstances that are
frequent causes of accidents or dangerous conditions
including performing or preparing for maintenance,
making modifcations to furnace design, human error,
and labeling errors or omissions. When preparing for
maintenance, it is important to remove hazards from the
maintenance area, isolate the area and/or equipment
2
Testing Safety
Charles E. Baukal, Jr.
CONTENTS
2.1 Introduction ..................................................................................................................................................................... 41
2.2 Safety Review .................................................................................................................................................................. 42
2.3 Hazards ............................................................................................................................................................................ 44
2.3.1 Excessive Temperature ....................................................................................................................................... 44
2.3.2 Thermal Radiation .............................................................................................................................................. 45
2.3.3 Noise ..................................................................................................................................................................... 46
2.3.4 High Pressure ...................................................................................................................................................... 51
2.3.5 Explosion .............................................................................................................................................................. 52
2.3.6 Flame Instability ................................................................................................................................................. 53
2.3.7 Environmental ..................................................................................................................................................... 54
2.4 Accident Prevention ....................................................................................................................................................... 55
2.4.1 Ignition Control ................................................................................................................................................... 55
2.4.2 General ................................................................................................................................................................. 56
2.5 Accident Mitigation ........................................................................................................................................................ 58
2.6 Recommendations .......................................................................................................................................................... 60
References .................................................................................................................................................................................. 60
© 2011 by Taylor and Francis Group, LLC
42 Industrial Combustion Testing
from operational equipment, and follow maintenance
procedures carefully. When modifying furnace design,
even when the modifcation seems minor, the proposed
modifcation should go through design procedures
similar to the procedures used for the original installa-
tion of the equipment. Without appropriate design, it is
not always possible to anticipate how a change in one
piece of equipment will affect other equipment.
Human error is sometimes caused by inattention or
poor training, but is frequently caused by a deliber-
ate attempt to shortcut a cumbersome procedure or to
make an inconvenient piece of equipment more conve-
nient to use. Accidents caused by labeling are frequently
the result of out-of-date labeling, incorrect labeling, or
no labeling at all resulting in the incorrect operation of
equipment.
Safety documentation and operator training provide
the backbone of a strong safety program, and are abso-
lutely essential to maintain a safe combustion working
environment. Safety documentation for combustion
related processes includes design information, process
hazard analysis (PHA) reports, standard operating
procedures, and training documentation. Feedback
from each of these documentation elements are linked
together as part of a plant’s overall process safety
program.
This chapter is not intended to be exhaustive, but
is designed to highlight some of the more common
safety concerns for industrial combustion processes.
Additional references are available on combustion
safety for the interested reader [12–18]. The National Fire
Protection Association (NFPA) is a good source of infor-
mation including producing some important standards
related to industrial combustion safety (e.g., [19–22]).
Other examples of sources include the European
Committee for Standardization (CEN) EN746-1 standard
“Industrial Thermoprocessing Equipment—Common
Safety Requirements for Industrial Thermoprocessing
Equipment,” the Canadian Standards Association (CSA)
B139-00 standard “Installation Code for Oil Burning
Equipment,” and the Japanese Standards Association
(JSA) JIS B 8415 standard “General Safety Code for
Industrial Combustion Furnaces.” This chapter contains
the following sections: safety review, hazards, accident
prevention and mitigation, and recommendations.
2.2 Safety Review
A safety review is recommended before any type of com-
bustion experiment, no matter how routine. The extent
of the review will depend on how common the test is
and the potential for damage or injury. For tests that are
commonly done with relatively little chance for dam-
age or injury, a short informal review may be all that is
required. For less common tests with a high potential
for damage or injury, a more extensive formal review is
recommended that could take a signifcant amount of
time to complete. Some type of checklist should be used
to ensure that all appropriate factors have been consid-
ered in the review.
There are various types of analyses that are used
for a process hazard analysis (PHA) of the equipment
design and test procedures, including the effects of
human error. Qualitative methods include checklists,
What-If, and Hazard and Operability (HAZOP) studies.
Quantitative methods include Event Trees, Fault Trees,
and Failure Modes and Effect Analysis (FMEA). All
of these methods require rigorous documentation
and implementation to ensure that all potential safety
problems are identifed and the associated recommen-
dations are addressed. The review should also consider
what personal protective equipment (PPE) is needed to
protect workers from injuries.
Figure 2.1
Test heater that has been overpressured.
© 2011 by Taylor and Francis Group, LLC
Testing Safety 43
The review team should consist of those with appro-
priate experience in all of the relevant areas involved
in the test. The engineer or scientist conducting the
experiment should be part of the team, although not
necessarily the team leader. At least one experienced
technician or operator, preferably one who will be
involved in the test, should participate in the review.
Many organizations have someone responsible for
safety who is normally part of such a review team and
may lead the team as appropriate. If a formal review
will be conducted, someone trained in the particular
review process being used will often lead the process.
Some specialists may be needed on the team. For exam-
ple, if new advanced sensors or controls will be used in
the test, then someone familiar with those technologies
should be included. Often, someone involved in the
management of the test facility will be involved in these
safety reviews. Depending on the extent of the review
and the potential for damage or injury, it is generally
recommended that at least one person on the review
team should not be part of the test. This person should
be independent of the budgeting and scheduling so
they can focus on safety and not be unduly infuenced
by other factors.
Prior to conducting the review, the team leader should
gather all appropriate documentation such as drawings
and procedures and make sure they are up-to-date.
Some new documentation may need to be created if it
does not already exist. Each component should have a
unique label that can be used in the review to identify
various scenarios. For example, a fuel supply ball valve
might have a label such as BV101. The review can then
use this label to refer to conditions where BV101 might
be open or closed. The documentation should also
include performance specifcations for the equipment
such as the maximum fow rate, temperature, pressure,
and so forth. If appropriate, vendor literature may also
be needed to determine the limits of operation for a
given piece of equipment.
Formal reviews normally require appropriate docu-
mentation. For example, Excel spreadsheets or Word
templates are available for conducting safety reviews
to ensure that a specifc process has been used in the
analysis and to document the results of the review.
The documentation should include the team mem-
bers, the dates and times of the review, the functional
area of each team member, the purpose of the review,
the equipment and processes being analyzed, key
specifcations for the equipment (e.g., maximum fow
rate), what codes and standards apply to the review
if any, and a discussion of past incidents that may
be relevant to what is being reviewed. The review
should consider the equipment and materials in the
vicinity of what is being reviewed. For example, fuel
storage tanks or high voltage electrical equipment in
the vicinity should be considered. All potential devia-
tions (e.g., high fow or no fow when there should be
fow) should be analyzed including possible causes
for the deviation (e.g., human error or equipment fail-
ure), the likelihood (e.g., frequency) the event might
occur, the consequences if it did occur, and the safe-
guards to help mitigate the consequences or prevent
the event from occurring. This often means educated
estimates as many of the potential events may never
have occurred before. If necessary, if the current
equipment or procedures are deemed inadequate,
then appropriate recommendations should be made.
In some cases, a recommendation might need further
analysis to study a particular issue. Each recommen-
dation should be assigned to someone, not necessarily
a team member, with a completion date. The recom-
mendations should be tracked to ensure they have
been completed on a timely basis. The review team
may recommend that certain recommendations be
completed prior to conducting the experiment, while
other recommendations may not need to be completed
before testing can begin.
Safety reviews should be scheduled far enough in
advance that any recommendations from the reviews
can be implemented in time so they do not adversely
affect the testing schedule, especially if any equip-
ment needs to be ordered or any new apparatus needs
to be built. Waiting until the last minute for the review
may unnecessarily pressure the review team into mak-
ing hasty decisions or cause them to overlook possible
incidents that could occur. This may mean that alter-
native team members may need to be used if primary
team members are unavailable in the time frame of the
review.
Another aspect of a safety review that may need to
be considered (depending on the nature of the test) is
environmental permitting. For example, if a new type
of fuel will be used that the test facility is not currently
permitted to use, then the appropriate approvals will
need to be obtained before testing can begin. This is
often the case if a fuel contains hazardous components,
such as benzene or chlorine-containing compounds. A
related scenario is when the exhaust products may con-
tain components that the facility is not currently per-
mitted to emit. For example, if furans or dioxins may
be potential effuents, the facility environmental permit
may need to be checked to see if these are allowed and if
so at what levels. Getting a new permit or a variance of
an existing permit can be a lengthy process, so be sure
to allow enough time for this. If additional treatment
equipment is required to reduce pollution emissions,
this can further lengthen the process. For example,
obtaining equipment to remove furans or dioxins from
the exhaust gases can be both expensive and lengthy, so
plan accordingly.
© 2011 by Taylor and Francis Group, LLC
44 Industrial Combustion Testing
2.3 Hazards
There are many potential hazards that may need to be
considered in industrial combustion testing. Some of
the common ones are briefy considered next.
2.3.1 excessive Temperature
Combustion typically involves high temperatures that
can cause equipment damage and personnel injuries
if not properly handled. For production equipment in
operating plants, there is often a maximum surface
temperature limit dictated by government regulations.
For example, the U.S. Occupational Safety and Health
Administration (OSHA) guidelines limit the maximum
external temperature of a surface exposed to person-
nel to 140°F (60°C). If the temperature of equipment
in the vicinity of personnel exceeds that temperature,
then some type of shielding (see Figure 2.2) is usually
required to prevent people from accidentally touching
hot surfaces that could burn their skin. An alternative is
to add more insulation to the hot surface to reduce the
skin temperature (see Figure 2.3). For any experiments
that might have high external surface temperatures that
could be contacted by personnel, appropriate precau-
tions must be taken to prevent burn injuries.
Equipment could be damaged by excessive tempera-
tures. This includes equipment outside the combustion
chamber as well as inside. For example, external equip-
ment such as valves for controlling fuel and air fows
may have Tefon
®
seals that could be damaged by high
temperatures. Test setups should be analyzed to ensure
the equipment can either handle the temperatures that
might be encountered or that shields are in place to pro-
tect any equipment that could be adversely affected by
the heat. Equipment inside the combustion chamber
must be capable of handling the temperatures that may
be generated. For example, metal furnace walls are nor-
mally protected by insulation rated to a suitable tem-
perature. Any uninsulated metal that will be directly
exposed to the heat may need to be cooled. The tubes
in a process heater are cooled by the hydrocarbon fuids
fowing through them. The tubes in a boiler are cooled
Figure 2.2
Metal mesh shielding personnel from hot exhaust stack.
Figure 2.3
Insulated temporary ductwork.
© 2011 by Taylor and Francis Group, LLC
Testing Safety 45
by water fowing through them. Instrumentation such
as metal gas sampling probes is often water cooled to
protect the metal and to quench the gas sample as part
of the measurement process (see Chapter 7).
In some cases, the temperatures may not be known a
priori in the case of experimental conditions that have
not been previously tested. While calculations may give
an estimate of the expected temperatures, it is generally
advisable to include some instrumentation, such as ther-
mocouples and infrared detectors, to measure the actual
temperatures. If there are any areas that are a particular
concern for high temperatures, warnings and alarms could
be added to alert operators. Automatic shutdowns could
even be added to prevent equipment damage.
2.3.2 Thermal radiation
Thermal radiation is related to excessive temperature,
but no direct contact is necessary for damage or injury to
occur that distinguishes it here from the hazard caused
by contacting high temperature surfaces. Thermal
radiation is a necessary and important phenomenon
in nearly all industrial combustion processes [23].
Figure 2.4 shows thermal radiation coming from the
viewport of a hot furnace. It is typically the predominant
form of heat transfer from the fame to the combustor
walls and heat load. Most combustion engineers design
systems to optimize the radiation heat transfer to get the
proper temperature distribution in the combustor. Some
burners are deliberately designed to maximize the radi-
ation from the fame by creating luminous fames [24].
Excessive external thermal radiation can cause equip-
ment damage and injure personnel. Electronic equip-
ment in particular is susceptible to damage from high
radiant loadings. External radiation can melt plastic
parts such as valve seals, dry out lubricated parts such
as motors, and make equipment operation more diffcult
where, for example, operators may need to wear gloves
to open or close valves. Figure 2.5 shows a corrugated
stainless steel fence designed to shield combustion fow
control equipment, especially the electronics, from ther-
mal radiation during fare testing (see Chapter 30). An
even more serious concern is premature ignition caused
by thermal radiation of the air–fuel mixture prior to
reaching the burner outlet, where the fuel and oxidant
are premixed upstream of the burner outlet.
Personnel injury from excessive thermal radiation
should be considered in most industrial combustion
applications. Figure 2.6 shows an instrument for mea-
suring thermal radiation that can be used to determine
potentially dangerous levels. Heat stress from high
thermal radiation loading can cause illnesses ranging
from behavioral disorders to heat stroke and even death
[25]. Heat stress is the mildest form and is temporary in
nature. It could include heat rashes sometimes referred to
as “prickly heat” and dehydration. High humidity ambi-
ent conditions can exacerbate the problem for personnel
in the vicinity of high thermal radiation conditions. Skin
can be damaged from excessive heat. Even lower thermal
Figure 2.4
Radiation from a viewport. (From Baukal, C.E., Industrial Combustion
Pollution and Control, New York: Marcel Dekker, 2004.)
Figure 2.5
Stainless steel fence shielding fow control equipment from thermal radiation during fare testing.
© 2011 by Taylor and Francis Group, LLC
46 Industrial Combustion Testing
radiation loadings can cause worker fatigue and reduce
performance. Older, overweight, or out-of-shape workers
are particularly at risk from high thermal radiation loads.
There are two general abatement strategies used to
mitigate and control external thermal radiation loading:
reducing the source of the radiation or shielding person-
nel and equipment from the source. Source reduction
involves reducing the external radiation source. One way
is to insulate the radiation source, for example, a furnace
wall, to reduce the external surface temperature and
hence the radiation. Another way is to reduce the energy
source heating the radiating surface. The easiest way
to do that is to reduce the fring rate; however, this will
also reduce the production rate, which is normally not
desirable. Another way to reduce the radiation source is
to cool it, for example, by water cooling the external fur-
nace walls. Viewports should have some type of shutter
to minimize the heat escaping through them when they
are not in use (see Figure 2.7). Furnace leaks should be
repaired, not only to reduce external thermal radiation,
but also to improve the thermal effciency.
The second abatement technique is shielding and
cooling, which involves shielding personnel and equip-
ment from the external radiation source. This can be
done with some type of physical barrier such as a wall
(e.g., Figure 2.5). It can be done with screens around the
radiation source to prevent personnel from getting too
close. Individual pieces of equipment can be shielded
with insulation, a refective surface, or some type of
solid material. Personnel can be shielded by wearing
appropriate protective clothing designed for high heat
environments (see Figure 2.8).
2.3.3 Noise
Noise is often defned as unwanted sound. There are
many possible sources of noise in industrial combustion
testing (see Chapter 8). Some of these include combus-
tion roar, jet noise for fow through orifces, fow noise
for fuids fowing through piping, and equipment noise
such as from the combustion air fan. Acoustic resonance
can exacerbate the problem by magnifying the noise.
There are several strategies that can be used to reduce
combustion noise (see Bussman and Jayakaran [26] and
Baukal [27] for more details). One strategy is either to
move the source of the undesirable sound away from
the people or to move the people away from the sound.
However, this may not be practical for many industrial
applications. Another strategy is to put some type of
Figure 2.6
Radiometer for measuring thermal radiation. (From Baukal, C.E.,
Industrial Combustion Pollution and Control, New York: Marcel Dekker,
2004.)
Figure 2.7
Viewport with shutter. (From Baukal, C.E., Industrial Combustion
Pollution and Control, New York: Marcel Dekker, 2004.)
Figure 2.8
Heat resistant suit. (From Baukal, C.E., Industrial Combustion Pollution
and Control, New York: Marcel Dekker, 2004.)
© 2011 by Taylor and Francis Group, LLC
Testing Safety 47
sound barrier between the noise and the people. The
barrier can be either refective or absorptive to mini-
mize the noise. The noise source might be surrounded
by an enclosure or the operators may be located inside
a soundproofed enclosure. In some cases, it may be pos-
sible to use a silencer, which would act as a barrier, to
reduce the noise. For example, the exhaust from a car is
reduced by the muffer, which acts like a silencer. The
barrier could also be in the form of earplugs, ear phones,
or some other sound-reducing safety device worn by
people in the vicinity of the noise. Another technique is
to reduce the exposure time to the noise since noise has
a cumulative effect on human hearing. In some cases, it
may be possible to replace noisy equipment with new
equipment that has been specifcally designed to pro-
duce less noise, or to retroft existing equipment to pro-
duce less noise. For example, old combustion air blowers
and fans could be replaced by new, quieter blowers and
fans. Another way to reduce noise is to increase the
pipe size and reduce the number of bends in the pipe
to reduce the fuid noise caused by the fuids fowing
through the pipe. Resonance and instabilities can usu-
ally be designed out of a system if they are a problem.
Noisy burners can be replaced by quieter burners. The
burner noise is a function of the burner design, fuel, fr-
ing rate, air–fuel stoichiometry, combustion intensity,
and aerodynamics of the combustion chamber. These
strategies are discussed in some detail next.
The primary source of noise in most industrial com-
bustion systems is from the burner(s). The design of the
burner nozzle and the burner tile or quarl are important
factors in noise generation in combustion processes. One
method to reduce noise from a burner is to use larger
exit ports that produce lower gas velocities. However,
there are limits to how low gas velocities can be, depend-
ing on the fuels used and the burner type. For example,
the exit gas velocities in a premix burner must be greater
than the fame speed of the oxidizer/fuel mixture or else
fashback may result. Fuels containing high concentra-
tions of hydrogen will necessarily require higher exit
velocities because of the high fame speed of hydrogen.
Another method for reducing the sound from a com-
bustion process is to add a pipe or tube, often referred to
as a quarter wave tube, to the resonance system to cancel
out the harmonics causing the noise. In this technique, a
specially designed tube is attached to the chamber where
sound of a predominant frequency is causing resonance.
The quarter wave tube is designed to cancel out this res-
onance and therefore reduce the noise levels. Figure 2.9
shows a quarter wave tube installed on the side of an
exhaust stack for a gas-fred, propylene vaporizer that
previously exhibited a low frequency rumble during
operation, producing excessive noise levels. The quarter
wave tube was built with some adjustability to determine
the best geometry to maximize noise reduction. The
total noise level for the vaporizer without the tube was
95.3 dBA. The total noise level with the tube dropped to
83.8 dBA for a noise reduction of more than 10 dBA. This
brought the noise levels within acceptable limits. Note
that adding a quarter wave tube is not always a practical
option because, depending on the combustion chamber
geometry and the frequency of the harmonic, the tube
diameter and length may be excessive.
Other techniques to mitigate noise caused by combus-
tion instability include modifying the [28]:
Furnace stack height •
Internal volume of the furnace •
Acoustical properties of the furnace lining •
Pressure drop through the burner by varying •
the damper positions
Location of the pilot •
Flame stabilization techniques •
Figure 2.10 shows some of the common silencers used
to reduce the source of noise from industrial processes
[29]. Figure 2.11 shows a natural draft burner without an
air inlet muffer. These burners are often quiet enough
that no muffer is needed. Figure 2.12 shows a common
type of muffer used on a natural draft burner used in the
foor of refnery heaters where the air inlet has a baffe
lined with sound-deadening insulation. Figure 2.13 shows
a comparable standard muffer for a radiant wall burner
1

4
Wave tube
Exhaust
stack
Figure 2.9
Quarter wave tube on a propylene vaporizer.
© 2011 by Taylor and Francis Group, LLC
48 Industrial Combustion Testing
used in the side of ethylene cracking furnaces. Figure 2.14
shows a nonstandard muffer used on a foor-fred natural
draft burner where the muffer is larger than the burner
because of the very low noise requirements for the partic-
ular application. Figure 2.15 shows a nonstandard muffer
for low noise requirements on wall-fred burners.
Sound transmission mitigation is a technique that
involves mitigating the transmission of the sound from
the source to the receiver. This can be done in a vari-
ety of ways. One rather simple but not always practi-
cal method is simply to increase the distance between
the source and the receiver, which reduces the sound
levels at the receiver. Another strategy is to put some
type of barrier between the source and the receiver.
For example, a concrete wall could be built around the
source. People commonly plant trees and shrubs on
their property to mitigate some of the sound from their
neighbors and from road traffc noise. Figure 2.16 shows
enclosed fares with walls around the enclosures to help
mitigate the sound produced by the fares. Some type of
sound-absorptive material could be placed between the
source and the receiver. Different materials have differ-
ent sound-absorbing characteristics, depending on both
the composition and confguration [30]. Another way
could be to use a medium that is less transmissive for
sound. For example, water is less transmissive than air.
Noise generated by the burners in a combustion system
may be greatly mitigated by the combustion chamber,
Absorptive silencer. This silencer is
the most common type and takes the
form of a duct lined on the interior
with a sound-absorptive material.
Reactive expansion chamber. This
type reflects sound energy back
toward the source to cancel some of
the oncoming sound energy.
Reactive resonator. This type func-
tions in approximately the same way
as the reactive expansion chamber
type.
Plenum chamber. This device allows
the sound to enter a small opening in
the chamber: that sound which has
not been absorbed by the chamber’s
acoustical lining leaves by a second
small opening, generally at the
opposite end of the chamber.
Lined bend. Sound energy flowing
down a passage is forced to turn a
corner, the walls of which are lined
with acoustical material. The sound
energy is thus forced to impinge
directly on a sound-absorbing surface
as it reflects its way around the corner;
each successive impingement takes
sound energy from the traveling wave.
Diffuser. This device does not actually
reduce noise. In effect, it prevents the
generation of noise by disrupting
high-velocity gas streams.
Absorptive silencer
Reactive expansion chamber
Reactive resonator
Plenum chamber
Lined bend
Diffuser
Figure 2.10
Silencers. (From Liu, D. H. F., “6.7 Noise Control in the Transmission
Path,” in Environmental Engineers’ Handbook, edited by D. H. F. Liu and
B. G. Lipták, 2nd ed., Boca Raton, FL: Lewis Publishers, 1997.)
Figure 2.11
Natural draft burner with no air inlet muffer. (From Baukal, C.E.,
Industrial Combustion Pollution and Control, New York: Marcel Dekker,
2004.)
Figure 2.12
Natural draft burner with an air inlet damper. (From Baukal, C.E.,
Industrial Combustion Pollution and Control, New York: Marcel Dekker,
2004.)
© 2011 by Taylor and Francis Group, LLC
Testing Safety 49
which is usually a furnace of some type. The refractory
linings in most furnaces generally signifcantly reduce
the noise emitted from the burners. Noise is not com-
monly considered in many industrial heating applica-
tions for a variety of reasons. This is evidenced by the
general lack of information available on the subject. It
is diffcult to predict the noise levels before installing
the equipment due to the wide variety of factors that
infuence noise. Often, there are many other pieces of
machinery that are much noisier than the combustion
system so that the workers are already required to wear
hearing protection. In the future, noise reduction may
become more important. In combustion testing, noise
may be less of a concern because of the temporary
nature of many tests.
Muffers are commonly used on burners to reduce noise
levels (see Figure 2.17). Figure 2.18 shows how effective a
muffer is at mitigating the noise produced by a burner.
These muffers typically go on the combustion air inlet to
the burner and usually have some type of noise-reducing
insulation on one or more sides. Reed gives an example
of a natural draft burner producing 107 dB of noise before
any mitigation [31]. The addition of a primary muffer
reduced the noise to 93 dB and the addition of a second-
ary muffer further reduced the noise to 84 dB.
Figure 2.19 shows a custom-made cylindrical muf-
fer on the air outlet of a velocity thermocouple, often
called a suction pyrometer (see Chapter 5), used to mea-
sure higher temperatures. This type of temperature
Muffler
Insulation
Figure 2.13
Typical muffer for a radiant wall-fred natural draft burner. (From
Baukal, C.E., Industrial Combustion Pollution and Control, New York:
Marcel Dekker, 2004.)
Figure 2.14
Large muffers on two natural draft burners. (From Baukal, C.E.,
Industrial Combustion Pollution and Control, New York: Marcel Dekker,
2004.)
Figure 2.15
Large muffers on two radiant wall burners. (From Baukal, C.E.,
Industrial Combustion Pollution and Control, New York: Marcel Dekker,
2004.)
Wall at bottom of
enclosure helps
reduce noise
Enclosed flare helps
reduce noise
Figure 2.16
Two enclosed fares. (From Baukal, C.E., Industrial Combustion Pollution
and Control, New York: Marcel Dekker, 2004.)
© 2011 by Taylor and Francis Group, LLC
50 Industrial Combustion Testing
measuring device relies on high air fow rates across
a venturi to induce furnace gases to fow through the
pyrometer and across a shielded thermocouple. This
minimizes the effects of thermal radiation on the ther-
mocouple and produces a more accurate measure of the
true temperature. Measurements using bare wire ther-
mocouples can be as much as 200°F or more too low.
Compressed air is often used as the motive gas to induce
furnace gas inspiration. The high exit gas fow rates into
the atmosphere can produce relatively high noise levels
that are a concern for workers in the vicinity. The cylin-
drical muffer shown in Figure 2.19 reduced noise levels
by more than 10 dBA down to an acceptable level.
There are two levels of protection commonly used by
industrial workers to reduce noise levels [32]: plugs (see
Figure 2.20) and muffs (see Figure 2.21). One or both
types may be used, depending on the noise levels. A third
type of protection device is a helmet, commonly used by
motorcycle drivers, which provides relatively little hear-
ing protection and is rarely used in industry for hearing
protection. Therefore, this is not considered further here.
Plugs can reduce noise levels by 5–45 dB, depending
on the plug type and sound frequency. They come in
a variety of forms including disposable and reusable.
Disposable plugs are typically made of some type of
moldable material (e.g., foam) that can be inserted into a
variety of ear sizes and shapes. These are very inexpen-
sive and are typically bought in large quantities. They
are especially convenient for visitors who do not have
their own ear plugs. Reusable ear plugs can be cleaned
and used multiple times. Custom molded ear plugs are
available that are made to ft exactly in a specifc per-
son’s ears and are designed to be reusable.
Ear muffs are designed to cover the entire ear and typ-
ically reduce levels by 5–50 dB, depending on the muff
type and sound frequency. These can be used in lieu of
or in addition to ear plugs. When both plugs and muffs
are worn, noise protection is greater than either individ-
ually but is not additive. Muffs may be more comfort-
able to some compared to ear plugs. However, they may
also interfere with other PPE such as hard hats. Special
Figure 2.17
Typical muffer used on a natural draft burner. (From Baukal, C.E., Industrial Combustion Pollution and Control, New York: Marcel Dekker, 2004.)
40
45
50
55
60
65
70
75
80
85
6
3
1
2
5
2
5
0
5
0
0
1
0
0
0
2
0
0
0
4
0
0
0
8
0
0
0
1
6
0
0
0
Frequency, Hz
S
o
u
n
d

p
r
e
s
s
u
r
e

l
e
v
e
l
,

d
B
A
Unmuffled A-weighted
burner noise
Muffled A-weighted
burner noise
3
1
.
5
Figure 2.18
Sound pressure versus frequency with and without a muffer. (From
Baukal, C.E., ed., The John Zink Combustion Handbook, Boca Raton, FL:
CRC Press, 2001.)
Figure 2.19
Cylindrical muffer on a suction pyrometer. (From Baukal, C.E.,
Industrial Combustion Pollution and Control, New York: Marcel Dekker,
2004.)
© 2011 by Taylor and Francis Group, LLC
Testing Safety 51
muffs are made to attach to certain types of hard hats
where the muffs can be folded down or up as needed
(see Figure 2.22).
Convenience and comfort are important factors when
choosing appropriate noise protection for a given envi-
ronment. If too much effort is required to use or maintain
the hearing protection devices, or if they are uncomfort-
able, then they are less likely to be used. Proper training
and education is essential to maximize the effectiveness
of any hearing conservation program [33].
While not possible in some cases, a simple way to pro-
tect workers is either to increase their distance from the
sound source or to put them in a soundproofed enclo-
sure such as a control room or building. However, it is
almost impossible to keep all workers away from high
noise sources all the time, so hearing protection will
probably be necessary.
2.3.4 High Pressure
There are multiple ways that equipment and person-
nel might be exposed to high pressures in combus-
tion testing. High pressure gases are sometimes used,
for example, as span gases to calibrate gas analyz-
ers. Specialty fuels may sometimes be supplied from
high pressure cylinders. Proper handling techniques
should be used when moving these cylinders around.
The cylinders must also be properly stored. For exam-
ple, oxidizers and fammables are not normally stored
together. High pressure cylinders should be stored
away from high temperature sources. If possible,
cylinders should be stored outside, so that if a leak
should occur it would not be confned in a building
(see Figure 2.23).
Some combustion experiments may have the potential
to produce high pressures. This can happen if liquids or
gases are being heated, such as water in a boiler or hydro-
carbon fuids in a process heater. The high pressures may
be caused by high temperatures, by fuids being blocked
in, by some type of blockage in the system that causes
a restriction, or possibly by fuid vaporization from a
liquid to a gas that causes a large and rapid increase in
volume. In most commercial systems, adequate provi-
sions are designed into the system for pressure relief in
case the fuid pressures exceed the design limits. This
may not be the case for experimental systems, so the
experimenter must ensure that adequate precautions
are taken so the pressures do not exceed the design lim-
its. This may include the addition of a properly designed
relief system and some type of warning system to signal
the operators before potentially excessive pressures are
Figure 2.20
Typical ear plugs. (From Baukal, C.E., Industrial Combustion Pollution
and Control, New York: Marcel Dekker, 2004.)
Figure 2.21
Typical ear muffs. (From Baukal, C.E., Industrial Combustion Pollution
and Control, New York: Marcel Dekker, 2004.)
Figure 2.22
Ear muffs designed to be used with hard hats. (From Baukal, C.E.,
Industrial Combustion Pollution and Control, New York: Marcel Dekker,
2004.)
© 2011 by Taylor and Francis Group, LLC
52 Industrial Combustion Testing
reached. Adequate sensors should be provided so the
operators can monitor the pressures so corrective action
can be taken if necessary. One type of hazardous over-
pressure condition is a rapid defagration, which is con-
sidered separately next.
2.3.5 explosion
Danger of explosion may come from many sources,
but explosions most often occur when the equipment
involved is in a state of change such as startup, shut-
down, or maintenance. Since furnaces are made pri-
marily of metals, maintenance often involves welding.
The welding process provides an ignition source for
any combustible gases or liquids that might remain in
the work area. Typically, tests for the presence of fam-
mable gases or liquids are conducted before any main-
tenance is allowed to commence. However, thorough
testing to assure the absence of fammable materials can
be complex and diffcult.
Two principles should be followed when testing for
combustibles in a planned, enclosed maintenance area.
First, testing should be done not only in the immedi-
ate work area but also in any connecting plumbing or
equipment. Second, testing should be done immediately
before the work begins and periodically during the time
the ignition source is in use.
Stacks are designed to vent exhaust gases from
industrial processes. If the gases are combustible, fare
stacks may be used to burn the exhaust gases before
they are vented into the atmosphere. In the case of com-
bustible gases, operational procedures should be used
that prevent the infltration of oxygen into the stack
where it could mix with the exhaust gases and cause an
explosion. Stacks should be built with welded seams
rather than bolted joints. Bolted joints, especially joints
having surfaces that are not machined, can leak air into
the stack in areas of negative pressure. Operational
procedures should be used that assure a continuous
fow of gas is moving up the stack. A continuous fow
of gas will sweep away small leaks of air and prevent
air from moving down into the stack from the top.
The most common source of furnace explosions is the
use of an improper lighting procedure. Even procedures
that have been used for years without incident can cause
explosions if conditions change. For example, if the fuel
source is not completely isolated from the furnace before
the ignition source is ignited, a leaky valve can allow
enough gas into the furnace to cause an explosion. A safe
furnace lighting procedure for a furnace containing piloted
burners should include the following generalized steps.
1. Confrm that the furnace fuel lines are com-
pletely isolated from the furnace (either by dis-
connection, blind fange, or a double block and
bleed valve assembly).
2. Confrm that all auxiliary furnace equipment is
functioning properly, including all instrumen-
tation and measurement devices; open both the
burner air inlet register/damper and the fur-
nace stack damper to the fully open position.
3. Purge the furnace of any combustible or fam-
mable substances; follow the NFPA recommen-
dation of purging the furnace with four furnace
volumes of fresh air or inert gas (N
2
or CO
2
) [22].

4. Test the atmosphere inside the furnace, just
before ignition will be attempted, to assure that
there are no combustibles present.
5. Connect the pilot fuel line to the burner; activate
the permanent pilot igniter or insert a portable
pilot igniter or premixed ignition torch.
Figure 2.23
Pressurized gas cylinders located outside a building where gases are used in a lab inside the building.
© 2011 by Taylor and Francis Group, LLC
Testing Safety 53
6. Slowly open the pilot fuel control valve to the
manufacturer-specifed pilot fuel pressure;
visually confrm stable ignition of the pilot
fame.
7. Reestablish the main burner fuel supply (either
by connecting the main fuel line, removing a
blind fange, or reversing the double block and
bleed valve assembly) to provide a fuel source
to the furnace.
8. Slowly open the main burner fuel control valve
to supply the burner with the manufacturer-
specifed ignition fuel pressure; visually con-
frm stable ignition of the main burner fame.
If the burner ignition attempt is unsuccessful after a rel-
atively short time, or if the procedure is aborted prior to
successful burner ignition, the burner fuel supply should
be immediately disconnected. Subsequent attempts to
successfully light the burner must begin at step 1.
Another source of explosions in furnaces is improper
air management. If the furnace is starved for air, a pul-
sating huffng sound may result. The fame will be
unsteady, changing from long to short or wide to nar-
row. The variations are the result of the available air
being completely consumed without burning all of
the fuel. The air-starved fame will then be reduced in
volume until more air is available. It will then increase
in size until the available air is again consumed. This
alternating cycle causes the huffng sound. The correct
action, when a furnace is huffng, is to reduce the fow
of fuel until there is enough air for full combustion. If
the operator incorrectly increases the air without frst
reducing the fuel fow, the increased air supply may mix
with the large volume of unburned fuel already in the
furnace causing an explosion.
Obviously, the best plan is to prevent an explosion
from occurring, but appropriate plans should be made
in the event one should occur. For example, it is recom-
mended that all personnel be located away from the
furnace during light-off in case there is an incident. It
may also be advisable to have some type of relief system
built into the furnace, for example pressure relief doors
(sometimes referred to as explosion doors) that can open
in the case of overpressurization.
2.3.6 Flame instability
There are multiple potential problems that could cause
fame instability. The fre triangle (see Figure 1.5) shows
that fuel, oxygen, and an ignition source are required
for a fre. However, these are not enough to guarantee a
stable fame. Combustion chemistry is very complicated.
The combustion of methane with air includes over 350
chemical reactions and dozens of species, most of which
only survive for fractions of a second in the fame before
they recombine into more stable species such as CO
2
and
H
2
O. This chemistry must be allowed to proceed to com-
pletion because any interruptions could extinguish the
fame or cause the fame to become unstable. For exam-
ple, halon fre extinguishers disturb the chemistry in a
fame and cause it to go out. The velocity of the oxidizer
and fuel are critical to fame stability. If these velocities
are too slow or too fast, the fame can become unstable.
If the fuel and oxidizer mixing is inadequate, this can
produce an unstable fame. The more common causes of
fame instability are briefy discussed here.
If the exit velocity from a premixed burner is below
the fame speed for the oxidizer-fuel mixture, fashback
(sometimes referred to as blowback) can occur. This is
where the fame travels back inside the burner. The fame
can only travel as far as the oxidizer-fuel mixture is fam-
mable. The burner design also dictates how far the fame
can travel as shapes inside the burner can act as fame
holders or fame arrestors. Figure 2.24 shows a radiant
wall burner that has been damaged from a fashback.
Some design features that prevent fashback include:
making the outlet velocity profle as uniform as •
possible
maximizing the oxidizer-fuel exit velocity from •
the burner
making the oxidizer-fuel mixture fuel lean, •
which reduces the fame speed
making the oxidizer-fuel mixture infammable •
(either above the higher fammability limit or
below the lower fammability limit)
promoting laminar fow out of the burner •
because the laminar fame speed is much slower
than the turbulent fame speed
Figure 2.24
Burner damaged from a fashback.
© 2011 by Taylor and Francis Group, LLC
54 Industrial Combustion Testing
Flashback is normally easy to visually and aurally detect.
The general response to fashback is to either shut off
the fuel or increase the fring rate so the exit velocity is
above the mixture fame speed. Experimenters should
pay particularly close attention to any conditions that
might result in fashback. If fashback conditions are
likely to occur, appropriate precautions should be taken.
These might include using stronger and higher temper-
ature materials in the burner and shielding the equip-
ment in case the burner fails.
If the oxidizer-fuel mixture exiting the burner is much
higher than the fame speed, liftoff (sometimes referred
to as blowoff) may occur. Figure 2.25 shows an example
of a premix radiant wall burner lifting off. In this type of
burner design, the fame exits the burner at a 90° angle
and impinges on the refractory. When it is operating
correctly, the refractory is uniformly heated and radi-
ates to the process tubes in the furnace. In Figure 2.25,
the burner is being overfred and the fame is partially
lifting off, causing the fower-shaped heating pattern.
Liftoff typically occurs when the oxidizer-fuel mixture
velocity exiting the burner is much higher than the fame
speed and there is inadequate fame stabilization in the
burner to anchor the fame. There are some fames that
are designed to be lifted, but most industrial fames are
not lifted because of the possibility of the fame blowing
out and reigniting causing an explosion. One indicator
of this type of instability is a bouncing or pulsing fame.
This phenomenon is commonly referred to as “huffng.”
This type of low frequency instability can cause dam-
age to the furnace, supporting structure, and associated
equipment due to vibration. The normal procedure to
handle huffng is to slowly reduce the fuel fow until the
huffng stops and then to make whatever adjustments
(e.g., repairing or cleaning the burner) are necessary to
prevent this from occurring again. If the cause is due to
overfring the burner, then the burner design may need
to be modifed to reduce the outlet velocities and pre-
vent liftoff. If the fame were to completely lift off and go
out, the fuel should immediately be shut off to prevent
an explosion or limit the damage that could occur from
an explosion. This is typically done with some type of
optical fame detector (see Figure 2.26), such as an ultra-
violet fame scanner. Some burner design features that
minimize the chance for liftoff include:
reducing the oxidizer-fuel outlet velocities (e.g., •
reducing the fring rate)
designing fameholders into the burner to anchor •
the fame and reduce the outlet velocities
improving the oxidizer-fuel mixing (e.g., chang- •
ing the fuel injection angles)
increasing the oxidizer-fuel turbulence that •
increases the fame speed
2.3.7 environmental
There are many potential environmental problems from
combustion processes that can be safety issues [27].
Incomplete combustion of carbon-containing fuels can
produce carbon monoxide (CO) and smoke. Carbon mon-
oxide is usually a regulated pollutant that can asphyxi-
ate people in high enough concentrations. Smoke is also
a commonly regulated pollutant that can damage the
lungs and block or impair vision. These pollutants are
typically more prevalent with liquid and solid fuels.
Both can generally be easily controlled with suffcient
oxygen to complete combustion, proper mixing of the
oxygen with the fuel, and adequate temperatures for the
reactions to go to completion. Incomplete combustion of
hydrocarbon fuels can produce a wide range of prod-
ucts of incomplete combustion (PICs), including in the
extreme allowing some fuel to go through the combustor
Figure 2.25
Premix burner lifting off.
Figure 2.26
Coen iScan
®
fame detector.
© 2011 by Taylor and Francis Group, LLC
Testing Safety 55
unchanged. These PICs are also generally easily con-
trolled with suffcient oxygen, temperature, and mixing.
Fuels that contain potentially harmful chemicals may
produce effuents that contain hazardous materials.
For example, fuels containing benzene could produce
benzene in the exhaust under certain conditions. Since
benzene is a hydrocarbon, it is generally easy to com-
pletely destroy if there is adequate oxygen, temperature,
and mixing. Fuels containing chlorine or fuorine can
produce dioxins and furans that are toxic. These are
not as easy to eliminate from the exhaust, so some type
of posttreatment system is often required to minimize
these pollutants to acceptable levels.
Appropriate analyzers should be used to detect the
emissions from combustion processes. There are two
general types of analysis systems (see Chapter 7): extrac-
tive (see Figure 2.27) and in situ (see Figure 2.28). The
extractive system pulls a small sample from the exhaust
and sends it frst to a conditioning system that cools,
cleans, and dries the sample, before sending the con-
ditioned sample to the analyzers. These systems have
some lag time because they are remotely located from
the stack, usually in some type of conditioned build-
ing such as the control room. In situ analyzers measure
the gas composition at the sample location, where the
sample is often hot and contains water. These systems
provide very rapid response as there is very little delay
in getting the measurement.
2.4 Accident Prevention
The best safety strategy is to prevent accidents from
occurring in the frst place, thus preventing equipment
damage and personnel injury. Quintana, Camet, and
Deliwala (2001) describe the application of a predictive
risk analysis model to a large-scale combustion test facil-
ity [34]. The model uses continuous data sampling and
analysis of an experiment to predict potential hazards
that could lead to an incident, a system malfunction or
unacceptable risk conditions. Operators are warned of a
potential problem so they can take appropriate corrective
action. If the proper corrective action is not implemented
quickly enough, then the system can be designed to take
appropriate action. This type of automated safety analy-
sis system is particularly benefcial in combustion testing
because of the fast and dynamic environment that can
be present. Some of the more common strategies for pre-
venting accidents are briefy discussed in this section.
2.4.1 ignition Control
Ignition is the process through which combustion is
initiated, and occurs when a fammable mixture of fuel
and oxidizer comes in contact with a suitable ignition
source. Direct contact with a spark or fame is a very
common energy source often used for the intentional
ignition of industrial combustion equipment. For many
types of burners, the ignition source may be in the form
of a small premixed pilot burner, a portable electrostatic
ignitor, or a portable premixed gas torch.
Static electricity is a common potential ignition source
in chemical processing plants. An electrostatic charge is
Figure 2.27
Typical continuous emission measurement system. (From Baukal,
C.E., Industrial Combustion Pollution and Control, New York: Marcel
Dekker, 2004.)
Figure 2.28
Typical in situ analyzer.
© 2011 by Taylor and Francis Group, LLC
56 Industrial Combustion Testing
formed whenever two dissimilar surfaces move relative
to each other. A relevant example is liquid fowing
through a pipeline, moving past the walls of the pipe. In
this example, one charge is formed on the pipe surface,
while another equal but opposite charge is formed on the
surface of the moving liquid. When the voltage becomes
strong enough, the static electricity will discharge in
the form of an electrical spark. The spark can ignite
any combustible and fammable materials present. The
NFPA (1993) [35] presents detailed explanations of design
fundamentals for the prevention of fres and explosions
due to electrostatic discharge. Some methods used to
prevent unintended ignition include proper grounding
of electrical equipment, using nonsparking tools, pre-
venting smoking, and keeping electrical devices that are
not intrinsically safe (e.g., cell phones) out of areas in the
plant where fammables could be present.
2.4.2 general
There are many general precautions that can be taken
to prevent an accident during combustion testing. A
sample of such precautions is given here. Others may be
implemented as appropriate.
Some test rigs are temporary for one-time experi-
ments, which may require specifc precautions to prevent
equipment damage and personnel injury. Figure 2.29
shows a photo of a rubber mat over hoses being used for
a temporary combustion test. For test rigs that will be
permanent, cables and hoses should be located so they
will not be tripping hazards. For temporary rigs that
are commonly used in testing, appropriate precautions
should be taken to prevent injuries that could be caused
by these types of setups. Figure 2.30 shows safety tape
around a portion of a combustion test setup to prevent
Figure 2.30
Safety tape around test apparatus.
Figure 2.29
Rubber mat over tripping hazard.
© 2011 by Taylor and Francis Group, LLC
Testing Safety 57
unauthorized personnel from accidentally entering an
area where they could possibly be injured.
Monitoring and control equipment can be used to
rapidly stop a test if a problem is detected. The con-
ventional method used in commercial combustion sys-
tems is some type of burner management system that
includes a fame detector (e.g., Figure 2.26). A furnace
camera (Figure 2.31; see Chapter 16) may be useful for
remote monitoring of the burner, fame, and furnace.
These cameras are specially designed for use in furnaces
and may include air cooling, water cooling, and air
purging of the lens. The camera monitor can be located
in the control room so the testing can be rapidly and
remotely stopped. Figure 2.32 shows some emergency
stop push buttons in a control room that can be used to
rapidly stop a test if necessary. It may also be advisable
to have emergency stop push buttons at or near the test
rig. Figure 2.33 shows one of these push buttons located
next to a sight port on a test furnace so someone watch-
ing the testing through the sight port can rapidly stop
the test if a problem is detected.
Another example of a precaution that should be taken
is to prevent liquid fuel in storage tanks from leaking
into the ground and possibly contacting an ignition
source that could lead to a fre or explosion. Figure 2.34
shows an example of a metal containment system at
the base of elevated diesel storage tanks. The contain-
ment needs to be capable of holding the entire volume
of the tanks plus include some additional capacity for
rain and any other materials that could possibly enter
the containment system. Some type of monitoring sys-
tem to detect leaks may be advisable depending on the
circumstances, particularly in enclosed locations. As an
example, Figure 2.35 shows gas analyzers mounted in a
combustion laboratory wall inside a building.
Figure 2.31
Furnace camera.
Figure 2.32
Emergency stop push buttons in a control room.
Figure 2.33
Emergency stop push button next to a sight port on a test furnace.
Figure 2.34
Liquid fuel containment for diesel storage tanks.
© 2011 by Taylor and Francis Group, LLC
58 Industrial Combustion Testing
Personnel must be properly trained to be able to iden-
tify potential hazards, to operate the equipment, and to
respond in the event of an accident. Appropriate safety
signs (see Figure 2.36) should be used to alert personnel
of potential hazards in a particular area.
2.5 Accident Mitigation
While every effort should be made to prevent an acci-
dent, appropriate plans should also be in place in case
there is an accident to minimize the consequences if one
should occur. If a fre should occur, a fre alarm system
(see Figure 2.37) should be in place to alert test personnel
and to notify the local fre department. If the fre is not
too large and personnel have been trained to extinguish
fres, appropriate fre extinguishers (see Figure 2.38)
should be located in the vicinity of combustion testing.
Figure 2.37
Fire alarm system in a combustion test facility.
Figure 2.38
Large portable fre extinguisher.
Figure 2.35
NO, CO, O
2
, and combustibles analyzers in a combustion laboratory.
Figure 2.36
Example of safety signs on the outside of a building.
© 2011 by Taylor and Francis Group, LLC
Testing Safety 59
There may be multiple levels of shutdown depending
on the severity of an incident. In a multiburner test, the
lowest level of shutdown might be to shut down a single
burner, for example, in the event that burner fames out.
Another level might be to shut off all the burners to a
single test heater. Depending on how the test facility is
designed, it might be desirable to shut off an entire sec-
tion of the facility. For example, if a building in the facil-
ity contains multiple test furnaces, it might be desirable
to rapidly shut off all the furnaces in the building in the
event of a fuel leak in the building. The highest level of
shutdown might be to shut down the entire test facility
in the event of a very serious incident. Figure 2.39 shows
an emergency pull ring designed to shut down an entire
combustion test facility in case of a serious incident.
These shutdown systems are designed to minimize the
impact of an incident that could become much worse if
it adversely affects nearby equipment and personnel.
Figure 2.40 shows some examples of safety and medi-
cal kits to treat personnel who might be injured in an
incident. A well-equipped medical kit should be readily
available and personnel properly trained and certifed
in frst aid to treat minor injuries and provide temporary
treatment until emergency medical personnel arrive in
the event of a serious injury. It may also be advisable to
have an emergency shower (see Figure 2.41) available in
case personnel are exposed to chemicals or hot fuids.
Appropriate PPE should be used during industrial com-
bustion testing. This may include steel-toed shoes, safety
Figure 2.39
Emergency pull ring to shut down the entire facility.
Figure 2.40
Examples of safety and medical kits.
Figure 2.41
Emergency shower.
© 2011 by Taylor and Francis Group, LLC
60 Industrial Combustion Testing
glasses, a hard hat, gloves, and fame-retardant clothes. It
may also include a face shield, special glasses for view-
ing bright objects such as fames in furnaces, and possi-
bly even a special heat resistant suit (see Figure 2.8). The
PPE can prevent or minimize personnel injury that could
result during testing.
Figure 2.42 shows a photo of a wind sock located near
fuel gas piping. This sock can be used to determine what
direction leaking chemicals might be taken by the wind
in the event of a leak. This is particularly important for
toxics that could harm personnel and fammable mate-
rials that could start or feed a fre. In general, personnel
would normally want to go upwind of leaking equip-
ment unless that path would be more hazardous (e.g.,
they would have to go through a fre to get upwind).
2.6 Recommendations
Industrial combustion processes have the potential to
be very dangerous because of the large quantities of
fuels being combusted. Industrial combustion testing
can be even more dangerous due to the potential of
using unproven equipment designs and confgurations.
Rigorous safety reviews are strongly recommended to
identify potential hazards so appropriate precautions
can be taken to prevent equipment damage and
personnel injury. There are many potential hazards that
could be present during industrial combustion testing,
some of which have been discussed in this chapter.
Appropriate measures should be taken to prevent
accidents, but plans should also be made to minimize
the consequences of an accident if one should occur.
Safety should be the primary concern for any industrial
combustion experiments.
References
1. Faisal. I., and Abbasi, S. A. “Major accidents in process
industries and an analysis of causes and consequences.”
Journal of Loss Prevention in the Process Industries 12, no. 5
(1999): 361–78.
2. Center for Chemical Process Safety. Explosions in the
Process Industries. New York: American Institute of
Chemical Engineers, 1994.
3. Sanders, R. E. Chemical Process Safety: Learning from Case
Histories. Boston: Butterworth-Heinemann, 1999.
4. Kletz, T. Learning from Accidents. 2nd ed. Oxford:
Butterworth-Heinemann, 1994.
5. Kletz, T. An Engineer’s View of Human Error. 2nd ed.
Rugby, UK: Institution of Chemical Engineers, 1991.
6. Center for Chemical Process Safety. Guidelines for
Preventing Human Error in Process Safety. New York:
American Institute of Chemical Engineers, 1994.
7. Eckhoff, R. K. Explosion Hazards in the Process Industries.
Houston, TX: Gulf Publishing, 2005.
8. Crocker, W. P., and Napier, D. H. “Thermal Radiation
Hazards of Liquid Pool Fires and Tank Fires,” IChemE
Symp. Series No. 97.159–84. Oxford, UK: Institute of
Chemical Engineers, Pergamon Press, 1986.
9. Center for Chemical Process Safety. Thermal Radiation 1:
Sources and Transmission. New York: American Institute
of Chemical Engineers, 1989.
10. Center for Chemical Process Safety. Thermal Radiation 2:
The Physiological and Pathological Effects. New York:
American Institute of Chemical Engineers, 1996.
11. Kletz, T. What Went Wrong: Case Histories of Process Plant
Disasters. 4th ed. Houston, TX: Gulf Publishing Company,
1998.
12. Lewis, B., and Von Elbe, G. Combustion, Flames and Explosions
of Gases. 3rd ed. New York: Academic Press, 1987.
13. Cox, G. Fire Safety Science. UK: Routledge, 1990.
14. Jones, J. C., and Williams, A. Topics in Environmental and
Safety Aspects of Combustion Technology. Scotland: Whittles
Publishing, 1997.
Figure 2.42
Wind sock.
© 2011 by Taylor and Francis Group, LLC
Testing Safety 61
15. Marshall, V. C., and Ruhemann, S. Fundamentals of Process
Safety. UK: Institute of Chemical Engineers, 2000.
16. Korver, W.O.E. Electrical Safety in Flammable Gas/Vapor
Laden Atmospheres. Norwich, CT: William Andrew
Publishing, 2001.
17. Thomson, N. G. Fire Hazards in Industry. Oxford, UK:
Butterworth-Heinemann, 2002.
18. BP Safety Group. Safe Furnace and Boiler Firing. 4th illus-
trated ed. Rugby, UK: Institute of Chemical Engineers,
2005.
19. NFPA 30: Flammable and Combustible Liquids Code.
Quincy, MA: National Fire Protection Association,
2008.
20. NFPA 54: National Fuel Gas Code. Quincy, MA: National
Fire Protection Association, 2009.
21. NFPA 85: Boiler and Combustion Systems Hazard Code.
Quincy, MA: National Fire Protection Association, 2007.
22. NFPA 86: Standard for Ovens and Furnaces. Quincy, MA:
National Fire Protection Association, 2007.
23. Baukal, C. E. Industrial Combustion Heat Transfer. Boca
Raton, FL: CRC Press, 2000.
24. Slavejkov, A. G., Baukal, C. E., Joshi, M. L., and
Nabors, J. K. “Advanced Oxygen/Natural Gas Burner
for Glass Melting.” Proceedings of 1992 International Gas
Research Conference, Orlando, FL, November 16–19, 1992.
Rockville, MD: Government Institutes, Inc., 1993.
25. Owens, P. D. “Health Hazards Associated with Pollution
Control and Waste Minimization.” In Process Engineering
for Pollution Control and Waste Minimization, edited by
D. L. Wise and D. J. Trantolo. New York: Marcel Dekker,
1994.
26. Sams, G., and Jordan, J., “How to design low-noise burn-
ers.” Hydrocarbon Processing 75, no.12 (1996): 101–108.
27. Baukal, C. E. Industrial Combustion Pollution and Control.
New York: Marcel Dekker, 2004.
28. Bussman, W., and Jaykaran, J. D. “Noise.” In The John
Zink Combustion Handbook, edited by C. E. Baukal. Boca
Raton, FL: CRC Press, 2001.
29. Liu, D. H. F. “6.7 Noise Control in the Transmission
Path.” In Environmental Engineers’ Handbook, edited by
D. H. F. Liu and B. G. Lipták, 2nd ed. Boca Raton, FL:
Lewis Publishers, 1997.
30. Moulder, R. “Sound-Absorptive Materials.” In Handbook
of Acoustical Measurements and Noise Control, edited by
C. M. Harris, 3rd ed. Woodbury, NY: Acoustical Society
of America, 1998.
31. Reed, R. D. Furnace Operations. 3rd ed. Houston, TX: Gulf
Publishing, 1981.
32. Nixon, C. W., and Berger, E. H. “Hearing Protection
Devices.” In Handbook of Acoustical Measurements and
Noise Control, edited by C. M. Harris, 3rd ed. Woodbury,
NY: Acoustical Society of America, 1998.
33. Royster, L. H., and Royster, J. D. “Hearing Conservation
Programs.” In Handbook of Acoustical Measurements and
Noise Control, edited by C. M. Harris, 3rd ed. Woodbury,
NY: Acoustical Society of America, 1998.
34. Quintana, R., Camet, M., and Deliwala, B. “Application
of a Predictive Safety Model in a Combustion Testing
Environment.” Safety Science 38 (2001): 183–209.
35. National Fire Protection Association. NFPA 77:
Recommended Practice on Static Electricity. Quincy, MA:
National Fire Protection Association, 1993.
© 2011 by Taylor and Francis Group, LLC
63
The goal of experimental design is effciency: collecting
the maximum amount of information with the mini-
mum of experimental cost. We do this by making use
of symmetry in the experimental design. For example,
consider Figure 3.1. The design has three major symme-
try planes, denoted by the dotted lines—namely, hori-
zontal (x
1
), vertical (x
2
), and diagonal (x
1
x
2
)—and admits
the following equation:
y = a
0
 + a
1
x
1
 + a
2
x
2
 + a
12
x
1
x
2
 + ε, (3.1)
where
y is some response of interest
a
0
through a
12
are constants to be determined
x
1
and x
2
are factors of interest
ε is the experimental error.
3.1 Factor Coding
To simplify the solutions of Equation 3.1, we shall code
the factor extrema to ±1. This is easily done by means of
the following linear transforms:

x
k
k
k k
k k
=

+ 






ξ
ξ ξ
ξ ξ
,max ,min
,max ,min
2
2
(3.2)
where
x
k
is the kth coded factor,
ξ
k
is the factor in its natural dimension,
ξ
k,max
is the maximum value of the factor, and
ξ
k,min
is the minimum value of the factor.
The quantity in parenthesis will be recognized as the
average factor value while the denominator represents
the half-range. We illustrate the use of the transform
with an example. Suppose ξ
1
is the oxygen fraction (from
1 to 3%) and ξ
2
is the furnace temperature (from 1450 to
1600°F). Then the transforms become x
1
 = (ξ
1
  – 2%)/1%,
x
2
 = (ξ
1
 – 1525°F)/75°F, and the reader may easily verify
that the transform results in factors of having a dimen-
sionless range of ±1. With such coding, we shall show
that the solutions for the coeffcients in Equation 3.1 are
as follows:
a
y y y y y
a
y y y y
0
00 11 12 21 22
1
12 22 11 2
=
+ + + +
=
+ ( ) − +
5
,
11
2
21 22 11 12
12
11 22
4
4
( )
=
+ ( ) − + ( )
=
+ ( )
,
, a
y y y y
a
y y −− + ( ) y y
12 21
4
(3.3)
where y
00
through y
22
are defned per Figure 3.1. That
is, a
0
is the mean value of all observations and the
3
Experimental Design
Joseph Colannino
CONTENTS
3.1 Factor Coding ................................................................................................................................................................ 63
3.2 Using Matrix Equations ............................................................................................................................................... 64
3.3 Factorial Designs ........................................................................................................................................................... 65
3.4 About Experimental Error ........................................................................................................................................... 66
3.5 Sums of Squares ............................................................................................................................................................ 67
3.6 Degrees of Freedom ...................................................................................................................................................... 67
3.7 The ANOVA ................................................................................................................................................................... 67
3.8 Error ................................................................................................................................................................................ 68
3.9 Using Center Points and Replicates ........................................................................................................................... 69
3.10 Fractional Factorials ..................................................................................................................................................... 71
3.11 Generating the Fractional Design ............................................................................................................................... 73
3.12 Calculating the Confounding Pattern ....................................................................................................................... 73
3.13 Central Composite Designs ......................................................................................................................................... 73
3.14 Blocking .......................................................................................................................................................................... 75
© 2011 by Taylor and Francis Group, LLC
64 Industrial Combustion Testing
remaining coeffcients are contrasts along various axes
of symmetry. For example, if

y
y
y
y
y
00
11
12
21
22
57
40
50
60
80
















=




























= then
a
a
a
a
0
1
2
12
577 5
7 5
12 5
2 5
.
.
.
.












In fact, these are the least squares solutions as we shall
demonstrate.
3.2 Using Matrix Equations
We may recast Equation 3.1 as follows:
y
y
y
y
y
00
11
12
21
22
57 4 7 5 0 















=
+ ( ) . . ++ ( ) + ( )( )
+ − ( ) + − ( ) +
12 5 0 2 5 0 0
57 4 7 5 1 12 5 1 2
. .
. . . .55 1 1
57 4 7 5 1 12 5 1 2 5 1 1
− ( ) − ( )
+ − ( ) + + ( ) + − ( ) + ( ) . . . .
557 4 7 5 1 12 5 1 2 5 1 1
57 4 7 5
. . . .
. .
+ + ( ) + − ( ) + + ( ) − ( )
+ +11 12 5 1 2 5 1 1 ( ) + + ( ) + + ( ) + ( )
















+

. .
00 4
0 1
0 1
0 1
0 1
57
40
50
60
80
.
.
.
.
.
















=
















(3.4)
Or, making use of matrix algebra, we may write Equation
3.4 in the following way:

y
y
y
y
y
00
11
12
21
22
1 0 0 0
1 1 1 1
















=
− − −
11 1 1 1
1 1 1 1
1 1 1 1
57 4
7 5
12
− −
− −
















.
.
.55
2 5
0 4
0 1
0 1
0 1
0 1
.
.
.
.
.
.












+
− 















=
















57
40
50
60
80

(3.5)
Noting that in matrix multiplication, columns are mul-
tiplied by rows, the reader should satisfy himself that
Equations 3.5 and 3.4 are equivalent. This is standard
linear algebra.* Equation 3.5 may be further condensed
by making use of matrix notation.
y = Xa + ε (3.6)
In this chapter, we use the convention that bold lower
case letters represent column vectors (i.e., n × 1 matri-
ces), while bold upper case letters represent matrices
(i.e., general m × n matrices). For row vectors (i.e., 1 × n
matrices) we use the transpose operator, which switches
rows and columns for any matrix or vector (e.g., x
T
).
Equation 3.7 demonstrates the principle:

X =
x x x
x x x
x x x
m
m
n n m
0 1 1 1 1
0 2 1 2 2
0 1
, , ,
, , ,
, ,


   

,,
, ,
n n
x
x
x
x












=












=
x
x
1
2
1

T
xx x
n 2
 ( )
(3.7)
where x
j,k
represents the matrix element in the jth
column and the kth row. For purposes of convenience,
we index the columns beginning with 0 and the rows
beginning with 1.
Returning now to Equation 3.6, we premultiply by the
transpose of X. This assures that the matrix to be solved
will be square (equal number of rows and columns) and
therefore a candidate

for a unique solution.
X
T
y = X
T
X a + X
T
ε (3.8)
As a principle of the least squares process, a is cho-
sen such that X
T
ε  is driven to 0. This leaves X
T
y = X
T
X
a, which is known as the normal equation. To fnd a with
*

For a review of matrix algebraic concepts, please see J. Colannino,
Modeling of Combustion Systems, A Practical Approach, Chapter 1,
Taylor & Francis/CRC Press, Boca Raton, FL, 2006, or any standard
text on linear algebra.

All matrices having general solutions are square; however, not all
square matrices have solutions. Notwithstanding, the experimental
designs we present will always be soluble in the way shown.
Vertical symmetry plane
y
12
y
11
y
22
y
0,0 x
1
direction
y
21
x
2
direction
Diagonal
symmetry
plane (1 of 2
equivalent planes)
Horizontal
symmetry
plane
Figure 3.1
An experimental design. The design comprises fve points and three
main symmetry planes: horizontal, vertical, and diagonal.
© 2011 by Taylor and Francis Group, LLC
Experimental Design 65
this property, we multiply by the inverse matrix (X
T
X)
−1
.
This gives (X
T
X)
−1
X
T
y = (X
T
X)
−1
(X
T
X)a. Since by defni-
tion (X
T
X)
−1
(X
T
X) = I (the identity matrix) and Ia = a, we
have our fnal result:
a = (X
T
X)
−1
X
T
y (3.9)
This is a general result for any experimental design
whatsoever. However, owing to symmetry in statisti-
cally cognizant designs, all nondiagonal elements in
X
T
X and (X
T
X)
−1
are zero. This results in diagonal matri-
ces that are not only easy to solve (e.g., Equation 3.3),
but also results in coeffcient solutions that are indepen-
dent and maximally effcient. To see this, we begin by
expanding X
T
X:

X X
T
=
x x x
x x x
x x
n
n
m m
0 1 0 2 0
1 1 1 2 1
1 2
, , ,
, , ,
, ,


   
 xx
x x x
x x x
m n
m
m
,
, , ,
, , ,












0 1 1 1 1
0 2 1 2 2


   
 x x x
x x
n n m n
k
k
n
0 1
0
2
1
0
, , ,
,












=
=

,, , , ,
, , ,
k k
k
n
k m k
k
n
k k
k
n
k
x x x
x x x
1
1
0
1
0 1
1
1
= =
=
∑ ∑


22
1
1
2
0
1
1
k
n
k m k
k
n
k m k
k
n
k
x x
x x x x
= =
=
∑ ∑


   
, ,
, , , mm k
k
n
m k
k
n
x
, ,
= =
∑ ∑























2
2
1


(3.10)
where
k indexes the rows,
m is the total number of columns (factors), and
n is the total number of rows (observations).
For the designs we consider, ∑
= k
n
k
x
1 0
2
,
 = n, the num-
ber of observations (always), because x
k,1
 = 1 for every k
(see Equation 3.5). In all cases, X
T
X is symmetrical as
refected about the principal diagonal; that is, X
T
X is a
symmetrical matrix. Now for symmetrical experimen-
tal designs centered at zero (also known as balanced
designs), all nondiagonal elements are zero (the reader
should verify this using the matrix of Equation 3.5, for
example). This results in the following matrix:
X X
T
=

















=
=


n
x
x
k
k
n
m k
k
n
1
2
1
2
1
,
,




(3.11)
Solving for X
T
y, we obtain
X y
T
=
x x x
x x x
x x
n
n
m m
0 1 0 2 0
1 1 1 2 1
1 2
, , ,
, , ,
, ,


   
 xx
y
y
y
x y
m n n
k k
,
,
























=
1
2
0

kk
n
k k
k
n
m k k
k
n
x y
x y
=
=
=

















1
1
1
1
,
,











(3.12)
Since x
0,k
 = 1 for all values of k, then ∑ = ∑
= = k
n
k k k
n
k
x y y
1 0 1 ,
.
Substituting Equations 3.11 and 3.12 into Equation 3.8
and dropping the subscripts, we obtain:

y
x y
x y
n
x
x
m
m





















=





1
1
2
2























a
a
a
m
0
1

. (3.13)
Since the matrix is diagonal, the solutions are inde-
pendent and easy to solve: a
0
 = ∑y/n and a
j
 = ∑x
j
y/∑x
2
j
.
Indeed, this is the basis for the solutions in Equation
3.3, which the reader should now be able to verify.
Spreadsheet software is generally able to perform
matrix operations including the inverse, multiplication,
and transpose operations.
3.3 Factorial Designs
The type of design shown in Figure 3.1 is known as a
full factorial two-level experimental design with a center
point. Not including center points, full factorial two-
level designs require 2
f
points, directed in f dimensions.
In two dimensions, the design points occupy the verti-
ces of a square, in three dimensions—a cube, in four or
more dimensions—a hypercube. Although diffcult to
visualize, higher dimensionality presents no mathemat-
ical challenge beyond what has already been presented.
These designs can ft a 2
f
-coeffcient model of the fol-
lowing type:
y a a x a x x a x x
j j
j
f
jk j k
k
f
j k
f
hjk h j
= + + +
= = <

∑ ∑ ∑
0
1 1
1
xx
k
k
f
j k
f
h j
f
= <

<

∑ ∑ ∑
+
1
1 2
.
(3.14)
© 2011 by Taylor and Francis Group, LLC
66 Industrial Combustion Testing
No model can ft more coeffcients than observations,
and often, the model is truncated far short of 2
f
coef-
fcients as it is rarely necessary to ft anything greater
than a second-order model. Therefore, full factorial
designs are typically unnecessary for f > 3 or so. To
accommodate fewer experimental points, fractional fac-
torial designs have been developed. We shall discuss
each in turn.
A two-level full factorial is an experimental design
that investigates all possible factor combinations at
two levels: high and low. These levels are coded to ±1
using the transforms presented in Equation 3.2. Since
the designs use ±1, generally, only the sign of the fac-
tor is reported. To construct a full factorial in f factors,
prepare f columns of design points, with minus and
plus signs alternating in blocks of 2
f−k
where k is the fac-
tor subscript. For example, for f = 3 there will be 2
f
 = 8
design points. Then x
1
alternates in blocks of four (2
3−1
),
x
2
alternates in blocks of two (2
3−2
), and x
3
alternates in
blocks of one (2
3−3
). Table 3.1 gives the design.
Once the design is selected, the experiments are
run, the responses are measured, and the coeffcients
calculated. As an example, the reader should dupli-
cate the least squares solutions calculated earlier.
However, something very important is missing from
the analysis—an analysis of experimental error.
3.4 About Experimental Error
Cognizance of experimental error is critical for under-
standing the meaning of the coeffcients and conduct-
ing the experiments themselves. As a frst principle,
factorial experimental designs should be run in fully
randomized order. This is necessary in order to break
any serial correlations that may exist. For example, sup-
pose the design of Table 3.1 were run in the order pre-
sented. Further suppose that ambient temperature is an
important but unrealized factor affecting the responses.
If the frst four experimental points were run before the
lunch break where cooler temperatures prevailed and
the second four experimental points were run in the
afternoon where warmer temperatures prevailed, then
x
1
would be confounded with ambient temperature. If a
1

were judged to be signifcant, it would be impossible to
know whether the effect was due to variations in x
1
or
variations in the temperature. Even worse, since we are
unaware that ambient temperature affects the results,
we would ascribe the results to a
1
even though there
may be no such association! Therefore, randomizing the
run order is an important prophylactic against serial
correlation.
The presumption in Equation 3.1 is that there is only
one error structure to worry about. That is, it presumes
that all experiments are run in such a way that they dif-
fer only with respect to changes associated with various
levels of the factors. This would be likely with ran-
domized runs occurring over a short span of time but
unlikely if, for example, half the experiments were run
in January and the other half in October. Unknown fac-
tor variations in feedstock, ambient conditions, differ-
ences in equipment used, and so on could all affect the
results. If, however, we can presume that Equation 3.1 is
an accurate model, then we can estimate the statistical
signifcance of each coeffcient. To do this, let us con-
sider the example given in Table 3.2.
We postulate the following model
y a a x a x x
k k
k
jk j k
k j k
= + + +
= = <
∑ ∑ ∑
0
1
3
1
3 2
ε (3.15)
We have truncated the design at second-order. This
is equivalent to presuming that the x
1
x
2
x
3
interaction is
insignifcant. Presuming higher order interactions are
negligible is usually a good assumption. However, if
we would like an independent error assessment, we can
duplicate some points. We discuss this later. For now, we
shall continue on the basis of our assumption. Equation
3.16 shows the normal equation and the solution to the
Table 3.1
A Full Factorial Two-Level Design in Three
Factors
Point x
1
x
2
x
3
1 − − −
2 − − +
3 − + −
4 − + +
5 + − −
6 + − +
7 + + −
8 + + +
Table 3.2
2
3
Factorial Design Using Hypothetical Data
Point
Oxygen
x
1
Temperature
x
2
H
2
in Fuel
x
3
NOx, ppm
y
1 − − − 26.4
2 − − + 30.9
3 − + − 38.1
4 − + + 45.3
5 + − − 29.9
6 + − + 35.3
7 + + − 56.3
8 + + + 57.7
© 2011 by Taylor and Francis Group, LLC
Experimental Design 67
coeffcients. Table 3.2 results in the following normal
equation and coeffcient estimates.

318.5
120.9
82.6
57.0
38.9
3.1
0.9























=






















8
8
8
8
8
8
8
0
1
a
a
aa
a
a
a
a
a
a
a
2
3
12
13
23
0
1
2






















; aa
a
a
a
3
12
13
23
39 8
15 1






















=
.
.
110 3
7 1
4 9
0 4
0 1
.
.
.
.
.
.























(3.16)
The problem now is to decide if the model as a whole
is signifcant, and if so, which coeffcients are signif-
cant. This is most easily done in an analysis of variance
table (ANOVA).
3.5 Sums of Squares
ANOVA is constructed using the following equation:

ˆ ˆ
y y y y y y
k
k
k
k
k
k
− ( ) + − ( ) = − ( )
∑ ∑ ∑
2 2 2
, (3.17)
where
k indexes the observations,
y
k
is the kth observation,
y is the average of all observations, ∑
k
y
k
/n, and,
y
ˆ
k
is the kth predicted observation (least squares
solution).
The frst term is known as the sum of squares, model
(SSM). The second term is known as the sum of squares,
residual (SSR), and the fnal term is known as the sum
of squares, total (SST). Equation 3.16 is true for any
least squares solution whatsoever. However, SSM can
be apportioned by component (i.e., a
0
, a
1
, a
2
, …) only
for so-called orthogonal models or data sets—that is,
those that generate diagonal X
T
X matrices (as is the case
for factorial designs). Equation 3.16 also has a matrix
formulation:

ˆ ˆ ˆ ˆ
y y y y y y y y y y y y
T T T T T T
− ( ) + − ( ) = − ( ) . (3.18)
Also, since y = Xa, by the rules of matrix algebra,
Equation 3.19 also holds

ˆ ˆ ˆ
y y a X X a a X y a X y
T T T T T T T
= ( ) = ( ) = ( ) (3.19)
Therefore, any of the quantities in Equation 3.19 may
be used interchangeably for y
ˆ
T
y
ˆ
in Equation 3.18. Note
also the following equalities that come in handy when
constructing the ANOVA.

SSM
SSR
T T T T
T T
= − ( ) = ( )
= − ( )
ˆ ˆ
,
ˆ ˆ
y y y y a X y
y y y y
1
1


== − ( )
= − ( ) = − ( )
y y a X y
y y y y y y a X y
T T T
T T T T T
SST
,
0
0
(3.20)
where
a
1
T
is the coeffcient row vector, starting from 1
(the second element) rather than 0 (the frst
element),
X y
T
( )
1
is the X
T
y column vector starting from 1 (the
second element) rather 0 (the frst element),
a
0
T
is a
0
, the frst element of the coeffcient vector,
X y
T
( )
0
is the frst element of the X
T
y column vector
(the one associated with a
0
).
3.6 Degrees of Freedom
The sums of squares are an important component of
the ANOVA table that we shall construct. Another
important component is known as the degrees of free-
dom (DF). The total degrees of freedom (DFT) is given
by n – 1, because if we know y and n – 1 other of the
y
k
values, we can calculate the remaining y
k
value.
Likewise, there are m – 1 degrees of freedom for the
model (DFM) because y
ˆ
k
is determined from m param-
eters (i.e., the model coeffcients) and we must subtract
1 DF for y. Finally, there are n – m degrees of freedom
in the residual (DFR) because the last term comprise n
values of y
k
and m values to determine y
ˆ
k
. We are now
ready to build our ANOVA (Table 3.3).
3.7 The ANOVA
Making use of these terms, the general ANOVA table is
organized as shown in Table 3.3. Applying this to the
data of Table 3.2 and making use of Equation 3.15 gives
the ANOVA shown in Table 3.4.
We shall take this opportunity to point some
salient features of the ANOVA table. The frst three
columns of cell entries are defned by appending the
left-hand column entry (M, R, or T) to the top row
© 2011 by Taylor and Francis Group, LLC
68 Industrial Combustion Testing
entry (SS, DF, or MS). For example, SSR = 5.7, DFM = 6,
and MSM = 546.0. The mean square is MS and F is
Snedecor’s ratio. We shall have more to say about both
of these presently. But frst, we digress to a discussion
of error.
3.8 Error
If the model accounts for all of the major infuences in
the response then the residual cumulates all the minor
errors. The central limit theorem states that these errors
will be distributed as a normal distribution, character-
ized by a mean (µ) and a standard error (σ) according to
Equation 3.21.

P y e P z e
N
y
N
z
( ) , ( ) ∼ ∼ =

− 




 − 1
2
1
2
1
2
1
2
2
2
2
πσ π
µ
σ
22
2
π
µ
σ
e
z
y
z ( )
=

,
and

(3.21)
where P
N
(y) is the probability that the normal distribu-
tion will have the value at y. In Equation 3.21, the param-
eter µ indicates where the distribution will be centered
(central tendency) and the parameter σ indicates the
dispersion (how narrow or broadly the distribution
will spread with y). This is the familiar bell-shaped curve
shown in so many statistics books. Generally the curve
is normalized in terms of standard error units (σ) and
centered at zero by letting z = y – µ/σ. Unfortunately,
we typically do not know µ and σ precisely. However,
if our sample is suffciently large, we can estimate them
with good confdence. In that case, we use y to estimate
the mean and s to estimate the standard error. These are
given by the following formulas:
y
y
N
k
k
=

, (3.22)
s
y y
n
k
k
=
− ( )


2
1
. (3.23)
Note that in Equation 3.23, the denominator is reduced
by 1 in order to subtract the degree of freedom used to
estimate the mean, as discussed previously.
Equation 3.22 presumes that each observation differs
only by random error, ε. Since the long run average of
ε is 0, then averaging with a suffciently large sample
will give a good estimate for µ by way of y. However,
an arithmetic average may not be used to estimate σ
because the result will always tend to zero. Rather,
we shall use a root mean square (RMS) procedure to
derive the dispersion parameter. If the various y
k
differ
only by experimental error, then y y
k
− ( )
2
is a measure
of σ
2
, known as the variance. The long run average of
these measures will tend toward σ
2
. However, a short
run average will be biased if we divide by n. This
is because we have used the data themselves to esti-
mate y. Therefore, we must subtract one degree of free-
dom in order not to double count one data point and
unfairly shrink our estimate of σ
2
. The true variance is
estimated by s
2
:
s
y y
n
k
k
2
2
1
=
− ( )


. (3.24)
Returning to the ANOVA, let us consider what
would happen if the model were not statistically sig-
nifcant. In that case there would be no difference
between y
ˆ
and y, except for experimental error; that
is, y y − =
ˆ
ε. In that case, both MSM and MSR would
estimate σ
2
. In such a case, the ratio MSM/MSR ~ 1.
Indeed, there would be no reason to use two separate
estimates of the same quantity and we would simply
pool both estimates. By inspection, the reader may see
that adding SSM + SSR and dividing by DFM + DFR,
we arrive at Equation 3.24. If, on the other hand,
the model were statistically signifcant, it would be
improper to pool the estimates and Equation 3.24
Table 3.4
ANOVA for Table 3.2 and Equation 3.15
Parameter SS DF MS F
M 3275.7 6 546.0 103.4
R 5.3 1 5.3
T 3281.0 7
Table 3.3
The General ANOVA for Model and Residual Terms
Parameter SS DF MS F
M
∑ − ( )
k
k
y y
ˆ
2
m − 1
∑ − ( )

k
k
y y
m
ˆ
2
1
∑ − ( )
∑ − ( )








k
k
k
k
y y
y y
m
N m
ˆ
ˆ
2
2
1
R
∑ − ( )
k
k
y y
ˆ
2
n − m
∑ − ( )

k
k
y y
N m
ˆ
2
T
∑ − ( )
k
k
y y
2
n − 1
© 2011 by Taylor and Francis Group, LLC
Experimental Design 69
would not be appropriate for estimating σ
2
. In such
a case, MSR would be the appropriate estimate for σ
2

because it alone averages the random error. Moreover,
MSM/MSR >> 1 in such a case. The question now
arises. How much greater than 1 should MSM/MSR
be in order to decide that the model is signifcant?
To decide, we must know the probability distri-
bution for MSM/MSR. One thing is certain, it will
not be distributed as P
N
because the range of P
N
is
–∞ < P
N
 < ∞, while σ
2
can never take on negative val-
ues. In fact, the ratio of two variances is distributed as
Snedecor’s F distribution. If both belong to the same
population then F ~ 1. If MSM and MSR belong to dif-
ferent populations (indicating that MSM and MSR
are not derived from the same distribution of σ
2
) then
F >> 1. Tabulations exist for critical values (F
crit
) of the
F distributions. These are shown in Appendix A. To
determine F
crit
from the table, we need to know three
things. First, how many degrees of freedom are in the
model (DFM)? Second, how many degrees of freedom
are in the residual (DFM)? Third, since we can never
be perfectly confdent that the values do not differ by
some chance event, what probability of a false posi-
tive (α) are we willing to tolerate? Generally α is set
to 0.05 (sometimes noted as p = 0.05; i.e., 95% conf-
dence that if F exceeds F
crit
, the model accounts for
signifcantly more variance than could occur merely
by chance). From Appendix A we see that F
crit
(6, 1,
0.05) = 234.0; however, F = 103.4. Since F fails to equal
or exceed F
crit
, the model is not statistically signif-
cant as a whole. However, perhaps some parameters
are statistically signifcant. An advantage of facto-
rial designs is that they are orthogonal and therefore
permit separate assessment of each model parameter.
We can reconstruct the ANOVA table (see Table 3.5)
using the k
th
sum of squares as a
k
(X
T
y)
k
and noting that
SSM
T
= ∑ ( )
=1 k
m
k
k
a X y ; that is, excluding a
0
(X
T
y)
0
. Note we
are asking the following statistical question. Do any
parameters besides a
0
add some signifcant advantage
to the model? Even if there are good theoretic reasons
to exclude a
0
in our model, we shall always compare
the fnal model with a model comprising a single
parameter—the mean—which for factorial designs is
a
0
. Therefore, a
0
is excluded from the analysis. Table 3.6
shows the values.
From Table 3.6 we see that a
1
through a
12
are statis-
tically signifcant as indicated by an asterisk in the
F column; a
0
, in the absence of other parameters, is
equivalent to y the hypothesis against which we are
testing all other effects. However, a
13
and a
23
are not
statistically signifcant. As such, they can be pooled
into the residual. Since a
1
through a
12
are signifcant,
they could be pooled into the model. If this were done,
the reduced model would now show itself to be sig-
nifcant (see Table 3.7).
3.9 Using Center Points and Replicates
The residual error may be divided into two com-
ponents—random error (also called pure error, the
Table 3.5
ANOVA with an Accounting of Individual Parameters
Parameter SS DF MS F
a
1
a
1
(X
T
y)
1
1 a
1
(X
T
y)
1
/1
a y y
k
k 1
2
1 ( )
T
1
X y /
ˆ
/ ∑ − ( )
a
2
a
2
(X
T
y)
2
1 a
2
(X
T
y)
2
/1
a y y
k
k 2
2
1 ( )
T
2
X y /
ˆ
/ ∑ − ( )
a
3
a
3
(X
T
y)
3
1 a
3
(X
T
y)
3
/1
a y y
k
k 3
T
3
( ) X y /
ˆ
/ ∑ − ( )
2
1
a
12
a
12
(X
T
y)
12
1 a
12
(X
T
y)
12
/1
a y y
k
k 12
T
12
( ) X y /
ˆ
/ ∑ − ( )
2
1
a
13
a
13
(X
T
y)
13
1 a
13
(X
T
y)
13
/1
a y y
k
k 13
T
13
( ) X y /
ˆ
/ ∑ − ( )
2
1
a
23
a
23
(X
T
y)
23
1 a
23
(X
T
y)
23
/1
a y y
k
k 23
T
23
( ) X y /
ˆ
/ ∑ − ( )
2
1
R
∑ − ( )
k
k
y y
ˆ
2
1
∑ − ( )
k
k
y y
ˆ
/
2
1
T 3281.0 7
Table 3.6
ANOVA for Table 3.2 and Equation 3.15 with
an Accounting of Individual Parameters
Parameter SS DF MS F
a
0 
= 39.8 (presumed signifcant) *
a
1 
= 15.1 1827.2 1 1827.2 345.9*
a
2
 = 10.3 851.9 1 851.9 161.3*
a
3
 = 7.1 406.3 1 406.3 76.9*
a
12
 = 4.9 189.0 1 189.0 35.8*
a
13
 = −0.4 1.2 1 1.2 0.2
a
23 
= 0.1 0.1 1 0.1 0.0
R 5.3 1 5.3
T 3281.0 7
* Signifcant effects.
Table 3.7
ANOVA for Table 3.2 and Equation 3.15
with a Reduced Model:
y = a
0
+ a
1
x
1
+ a
2
x
2
+ a
3
x
3
+ a
12
x
1
x
2
+ ε
Parameter SS DF MS F
M 3274.5 3 1091.5 498.5*
R 6.6 3 2.2
T 3281.0 6
* Signifcant effects.
© 2011 by Taylor and Francis Group, LLC
70 Industrial Combustion Testing
accumulation of manifold minor sources that act in
no organized way) and bias error (a systematic error
that is not random). Bias error is also called lack of ft.
Now if we replicate a point exactly and randomize the
run order, any difference in responses to the replicate
points should be due to random error alone. If we sub-
tract this from the residual, then what remains is an
estimate for bias or lack of ft. We can test the bias com-
ponent in the same way as any other and see whether
it is indeed signifcant. If it is signifcant, we can aug-
ment the model with additional points to eliminate
any bias caused by an incorrect model specifcation. If
the bias component is not signifcant then we will have
additional confdence that our model is adequate and
its terms are signifcant.
We want to add replicate points in a way that will
not disturb the orthogonal properties of the X
T
X
matrix. One way to do this is to replicate all points in
a design. Another way is to add several center points.
Replicating center points helps distinguish bias error
in two ways—replicating them allows for an inde-
pendent assessment of pure error and their location
allows us to detect any curvature or systematic bias in
our model. Additionally, replicating center points is
typically less expensive experimentally as we do not
need to double or triple the existing number of runs.
Table 3.8 shows the design augmented with three
additional points.
We must now add two additional rows (four addi-
tional entries) to our ANOVA. We need a row for bias
comprising the appropriate entries for sums of squares,
bias (SSB) and the associated degrees of freedom
(DFB). Additionally, we need the pure error terms (SSE
and DFE). Regarding the bias, we have the following
formulations:
SSB DFB
T T
= − ( ) = − = −
=

y y u m
k k
k
n
k k
ˆ ˆ ˆ
,
2
1
y y y y (3.25)
where k indexes each response, y
k
is the mean of all
the runs performed at the design coordinate associ-
ated with y
k
. In the case of nonreplicated points, y
k
 = y
k

because a single value is its own mean. The number of
unique points is u (i.e., number of coordinates or unique
positions in factor space), and m is the number of model
parameters (includes a
0
).
Regarding the pure error, we may account for it
thus:
SSE DFB
T T
= − ( ) = − = −
=

y y n u
k k
k
n
k k
2
1
y y y y , (3.26)
The total number of points is n, and u is the number
of unique points (i.e., number of coordinates or unique
positions in factor space).
Constructing the ANOVA from Table 3.8 and
Equation 3.15 we note that n = 11; that is, there are a
total of 11 runs, u = 9; that is, there are nine unique
coordinate points (8 factorial coordinates and 1 center
point coordinate); therefore, u = 9, and m = 7; that is,
there are seven model parameters—a
0
, a
1
, a
2
, a
3
, a
12
, a
13
,
a
23
. If we chose to ft the reduced model—a
0
, a
1
, a
2
, a
3
,
a
12
—then m = 4.
Table 3.9 shows the generic ANOVA table for individ-
ual model parameters and with residual split into bias
and pure error components. Table 3.10 gives the actual
numbers.
For convenience, we have appended the values of F
to the ANOVA table and added an asterisk for statisti-
cally signifcant effects. From the table in Appendix A,
we have F
crit
(0.05, 1, 1) = 18.5 and F
crit
(0.05, 2, 2) = 19.0. As
before, a
0
through a
12
are signifcant; but notably, the
bias is not signifcant since 2.7 < 19.0. In that case, our
original decision to use the x
123
effect as an estimate of
error seems justifed. In such a case, we can pool the bias
with the pure error and x
13
and x
23
effects, thus obtain-
ing Table 3.11.
Note that the critical threshold has fallen from 18.5
to 6.0 owing to the additional degrees of freedom
F
crit
(0.05, 1, 6). Thus, we can assess statistical signifcance
with better precision. Finally, since MSR is our best esti-
mate of σ
2
, we estimate σ ~ √2.7 ≈ 1.6 ppm (having the
same units as the response).
Table 3.8
2
3
Factorial Design Plus 3 Center Points
(Hypothetical Data)
Point
Oxygen
x
1
Temperature
x
2
H
2
in Fuel
x
3
NOx, ppm
y
1 − − − 12.7
2 − − + 25.8
3 − + − 21.7
4 − + + 38.6
5 + − − 32.3
6 + − + 47.2
7 + + − 64.1
8 + + + 76.1
9 0 0 0 39.5
10 0 0 0 41.4
11 0 0 0 39.3
© 2011 by Taylor and Francis Group, LLC
Experimental Design 71
3.10 Fractional Factorials
Full factorials are a good way to ft models such as
Equation 3.15. However, they quickly become unwieldy
for f > 5. Moreover, in principle, one needs only as
many unique data points as there are coeffcients to
ft, plus additional replicates as necessary. In such
cases, full factorials may have many more runs than
are necessary. For example, consider that we wish to
investigate performance in response to the following
factors:
ξ •
1
: oxygen concentration in the furnace, from
1 to 3%
ξ •
2
: hydrogen concentration in the fuel, from 0 to
25%
ξ •
3
: propane concentration in the fuel, from 0 to
25%
ξ •
4
: furnace temperature, from 1400 to 1800°F
ξ •
5
: fuel injection angle, from 5 to 15°
Suppose further that we are only interested in the
frst-order terms per Equation 3.27:

ˆ
y a a x
k k
k
= +
=

0
1
5
. (3.27)
Table 3.9
ANOVA for Factorial with Replicate Points
Parameter SS DF MS F
Model, M
SSM = a
1…
(X
T
y)
1…
 = 
∑ − ( )
k
k
y y
ˆ
2
a
1
a
1
(X
T
y)
1
1
a
1
1
1
X y
T
( )
a
y y
n u
k
n
k k
1
1
1
2 1
X y
T
( )
∑ − ( )
− 





=
a
2
a
2
(X
T
y)
2
1
a
2
2
1
X y
T
( )
a
y y
n u
k
n
k k
2
2
1
2 1
X y
T
( )
∑ − ( )
− 





=
    
DFM = m – 1 a
m–1
a
m–1
(X
T
y)
m–1
1
a
m
m


( )
1
1
1
X y
T
a
y y
n u m
m
k
n
k k


=
( )
∑ − ( )
− 





1
1
1
2 1
X y
T
Residual, R
SSR = ∑ − ( )
k
k
y y
ˆ
2
Bias, B
∑ − ( )
= k
n
k
y y
1
2
ˆ
u – m
∑ − ( )
∑ − ( )








=
=
k
n
k k
k
n
k k
y y
y y
n u
u m
1
2
1
2
ˆ
DFR = n – m Pure
error, E
∑ − ( )
= k
n
k k
y y
1
2
n – u
∑ − ( )

= k
n
k k
y y
n u
1
2
Total, T
∑ − ( )
k
k
y y
2
n – 1
Table 3.10
ANOVA for Factorial with Replicate Points with Table 3.8
and Equation 3.15 as the Basis
Parameter SS DF MS F F
crit
a
0
 = 39.9* (presumed signifcant) *
a
1
 = 15.1* 1827.2 1 1827.2 1827.2* 18.5
a
2
 = 10.3* 851.9 1 851.9 851.9* 18.5
a
3
 = 7.1* 406.3 1 406.3 406.3* 18.5
a
12
 = 4.9* 189.0 1 189.0 189.0* 18.5
a
13
 = –0.4 1.2 1 1.2 1.2 18.5
a
23
 = 0.1 0.1 1 0.1 0.1 18.5
B 5.4 2 2.7 1.4 19.0
E 2.6 2 1.3
T 3283.8 10
* Statistically signifcant, α = 0.05.
Table 3.11
ANOVA for Factorial with Replicate Points with Table 3.8
and Equation 3.15 as the Basis
Parameter SS DF MS F F
crit
a
0
 = 39.9* (presumed signifcant) *
a
1
 = 15.1* 1827.2 1 1827.2 1827.2* 6.0
a
2
 = 10.3* 851.9 1 851.9 851.9* 6.0
a
3
 = 7.1* 406.3 1 406.3 406.3* 6.0
a
12
 = 4.9* 189.0 1 189.0 189.0* 6.0
R 9.3 6 2.7
T 3283.8 10
* Statistically signifcant, α = 0.05.
© 2011 by Taylor and Francis Group, LLC
72 Industrial Combustion Testing
The formula has 1 zero-order coeffcient (a
0
), 5 frst- order
coeffcients (a
1
→ a
5
), and needs at least one degree of
freedom to estimate of error. Then the minimum number
of runs required is 7. However, a full factorial would
have 2
5
 = 32 experimental points and be able to ft the
following 32 coeffcients associated with the respective
factors and interactions.
a
0
a
1
a
2
a
3
a
4
a
5
a
12
a
13
a
14
a
15
a
23
a
24
a
25
a
34
a
35
a
45
a
123
a
124
a
125
a
134
a
135
a
145
a
234
a
235
a
245
a
345
a
1234
a
1235
a
1245
a
1345
a
2345
a
12345
This seems excessive. Can we reduce the number of
required experiments? The answer is “Yes”, if we are
willing to give up on higher-order effects. Fractional
factorial coeffcients offer one method for doing this
and require 1/2
k
·

2
f
or 2
f−k
runs where 1/2
k
represents
the fraction. For example, if we run the half fraction
(k = 1), this will require only 16 runs. We do this by
frst preparing a 2
4
factor design and then we add a
ffth column. We shall describe presently how the
sign order is determined in this last column. For now,
please examine Table 3.12.
This design has all the desirable orthogonal proper-
ties of the full factorial. However, we have not gotten
something for nothing. Although we will be able to
determine up to 16 coeffcients, they will all be con-
founded with others. We show the confounding pattern
presently. Rather than list the coeffcients, we merely list
the factor numbers. For example, the x
1
x
2
x
5
interaction
requiring determination of a
125
coeffcient will be repre-
sented merely by the factor list 125.
0 = 12345 1 = 2345 2 = 1345 3 = 1245 4 = 1235
5 = 1234 12 = 345 13 = 245 14 = 235 15 = 234
23 = 145 24 = 135 25 = 134 34 = 125 35 = 124
45 = 123
From this list, we see that every coeffcient is con-
founded with another. For example, a
34
is confounded
with a
125
. This means though it will be possible to
determine the a
34
coeffcient (or the a
125
coeffcient—
both are identical), it will not be possible to separately
know the infuence of x
34
from x
125
or the individually
associated coeffcients. Both effects will be merged.
Notwithstanding, third- and higher-order effects
are rarely important. Therefore, this is not a signif-
cant loss, and this is what makes fractional factorials
so powerful. In this design, we see that no two-factor
interaction is confounded with another two-factor
interaction. Why not fractionate further? Let us exam-
ine the confounding pattern for the ¼ 2
5
factorial (see
Table 3.13).
Confounding pattern for Table 3.13:
0 = 145 = 235 = 1234 1 = 45 =  234 = 1235 2 = 35 = 134 = 1245
3 = 25 = 124 = 1345 4 =  15 =  123 = 2345 5 = 14 = 23 = 12345
12 = 34 = 135 = 245 13 = 24 = 125 = 345
We see from the confounding pattern that main and
two-factor effects are confounded with other two-factor
effects. Since two-factor interactions are often signif-
cant, highly fractionated designs like this are typically
used for screening. In other words, we believe that frst-
order effects will likely show up if signifcant; however,
their values are subject to error. The reason for using
highly fractionated designs is to weed out less impor-
tant factors and reduce the design to fewer factors that
can be investigated in greater detail, perhaps with a full
factorial.
Table 3.12
A ½ 2
5
Fractional Factorial
Point x
1
x
2
x
3
x
4
x
5
1 − − − − +
2 − − − + −
3 − − + − −
4 − − + + +
5 − + − − −
6 − + − + +
7 − + + − +
8 − + + + −
9 + − − − −
10 + − − + +
11 + − + − +
12 + − + + −
13 + + − − +
14 + + − + −
15 + + + − −
16 + + + + +
Table 3.13
A ¼ 2
5
Fractional Factorial
Point x
1
x
2
x
3
x
4
x
5
1 − − − − +
2 − − + + −
3 − + − + −
4 − + + − +
5 + − − + +
6 + − + − −
7 + + − − −
8 + + + + +
© 2011 by Taylor and Francis Group, LLC
Experimental Design 73
3.11 Generating the Fractional Design
So how do we generate the fractional factorial? As
stated before, we frst prepare a 2
f−k
and determine the
k remaining columns by selecting k defning contrasts. A
defning contrast is any factor or factor interaction that
occurs in the full factorial (e.g., 123, 1234, 245, etc.). We
choose the defning contrasts to be as large as possible
without signifcant overlap. For example, the ½ 2
5
facto-
rial has k = 1. So we need to select 1 defning contrast;
we do not need to worry about a second defning con-
trast or its overlap with the frst. In such a case, we shall
use 12345, the largest possible contrast in the 2
5
facto-
rial. That is how column 5 in Table 3.13 was derived: by
multiplying 1234 (that is x
5
 = x
1
x
2
x
3
x
4
). For example, the
value for x
5
in row 12 is found by x
5
 = (+)(−)(+)(+) = (−).
This is equivalent to setting 12345 = + 1 in all rows. (We
could have just as easily set 12345 to −1 as well. In that
case, all the signs in column x
5
would be reversed.)
3.12 Calculating the Confounding Pattern
The method for calculating the confounding pattern
also involves the defning contrast. The rule is that we
shall multiply the contrast with the effect of interest
and eliminate the squared components. For example
(125)(12345) = 1
2
2
2
345
2
 = 34. Therefore 125 = 34, and the
reader may verify that multiplying either the x
1
x
2
x
5
or
x
3
x
4
columns together gives an identical factor pattern.
We repeat this for every possible factor combination so
the entire confounding pattern may be found.
In order to fnd the ¼ fraction, k = 2, and f – k = 5 − 2 = 3,
we start with a 2
3
full factorial and we choose two defn-
ing contrasts. We attempt to choose them with as little
overlap as possible because with multiple contrasts we
need to worry not only about the defning contrasts but
the implicit contrasts created by their interaction. For
example, if we choose 1234 and 235 we also have, tacitly,
12
2
3
2
45 = 145, and this is how Table 3.12 was generated.
Other choices fare no better (124 and 235 give 1345) and
some fare much worse (e.g., 124 and 125 give 5, mean-
ing that we will be confounding a
5
with a
0
– a very
bad idea).
3.13 Central Composite Designs
Generally, for large experimental design problems it is
useful to tackle the job sequentially. For example, for 11
factors, one uses a highly fractionated factorial. Suppose
such a screening design identifes two factors as impor-
tant. Then we investigate them with a full factorial hav-
ing a few center points. But when we analyze the data
via ANOVA, we fnd signifcant lack of ft (bias). In such
a case we may wish to augment the design with addi-
tional points that will allow us to ft a full second- order
design to include pure quadratics. A very good design
for doing this is known as a central composite. It com-
prises a full or fractional factorial augmented with axial
points. Figure 3.2 shows a central composite in two fac-
tors. The design comprises four factorial points (con-
nected in the fgure for illustration by a dotted square),
four axial points, and as many center point replicates as
necessary (typically three to six). The central composite
design allows one to ft a full second-order model of the
following kind:
y a a x a x x a x
k k
k
f
jk j k
k
f
j k
f
kk k
k
= + + +
= = <

=
∑ ∑ ∑
0
1 1
1
2
11
f

+ ε. (3.28)
Note the last summand indexes pure quadratic terms;
that is, a x a x
11 1
2
22 2
2
+ +. Central composites interrogate
the factor space at fve levels along each axis. The ques-
tion becomes what axial levels should we use? We have
two main choices.
1. Make the design strictly rotatable (e.g., circular,
spherical, or hyperspherical) by using α = f ,
y
10
y
12
y
02
y
22
y
0,0
y
11
y
01
y
21
y
20
Figure 3.2
A central composite design. In two dimensions, the design comprises
eight points and as many centerpoint replicates as required.
© 2011 by Taylor and Francis Group, LLC
74 Industrial Combustion Testing
where all axial points lie along an axis ±α from
the center.
2. Make the design strictly orthogonal by choosing
α = − nn n
f f
/ , 2 where n is the total number
of points in the design, and n
f
is the number of
factorial design points.
Typically the frst option is chosen and the quadratics
are assessed as a single orthogonal entry by ftting the
model:

ˆ
y a a x a x x a x
k k
k
f
jk j k
k
f
j k
f
kk k
= + + + −
= = <

∑ ∑ ∑
0
1 1
1
2
qq
k
k
f
( )
=

1
(3.29)
where q x n
k k
n
k
= ∑
=1
2
/ . For example, consider the central
composite design of Table 3.14.
For this design, we shall ft the model of Equation 3.30:

ˆ
y a a x a x a x x a x q a x = + + + + − ( ) + −
0 1 1 2 2 12 1 2 11 1
2
1 22 2
2
qq
2
( ) [ ]
(3.30)
where q
1
 = q
2
 = 2/3. Subtracting the mean square from
x
1
and x
2
keeps them orthogonal to a
0
and the rest of
the model terms. Although the bracketed terms are
orthogonal to all other model terms, a
11
is not orthog-
onal with respect to a
22
. Therefore, we shall assess the
entire bracketed quantity separately in the ANOVA. If it
is signifcant we shall not know if one or both quadratic
terms are signifcant. There is also a small chance that
both terms could be signifcant but of exactly opposite
effect so that they do not appear signifcant when pooled
in the ANOVA. This is not very likely but it is possible.
Table 3.14 generates the coeffcients and ANOVA given
in Table 3.15.
From the ANOVA, all the parameters are signifcant
including the entry for the quadratics. However, the
quadratic coeffcients are widely divergent with a
11
 = 0.2
and a
22
 = 3.9. It appears that a
22
could be the only reason
that the quadratic entry is signifcant. It is also appar-
ent that there is no signifcant bias, meaning the model
fts the data adequately. Therefore, we can pool the bias
and pure error into a single residual, MSR. This gives
Table 3.16.
In this case MSR = 0.08 and s = √0.08 = 0.29. To assess
the quadratic effects separately, we can compare each
parameter with its standard error. The standard error is
defned by Equation 3.31:
s s
k
k k
= ( )

2
1
X X
T
,
(3.31)
where s
k
is the standard error for the k
th
effect, s is the
standard error given by the best estimate of error (√MSR
if the bias and error can be pooled or √MSE if it cannot),
and X X
T
( )

k k ,
1
is the k
th
diagonal element of the inverse
matrix.
Table 3.14
A Central Composite Design in Two Factors
Point x
1
x
2
Y
1 − − 11.6
2 − + 21.5
3 + − 31.4
4 + + 61.6
5 0 −√2 21.0
6 0 √2 49.2
7 −√2 0 6.4
8 √2 0 48.9
9 0 0 27.0
10 0 0 27.8
11 0 0 27.2
12 0 0 27.0
Table 3.15
ANOVA with Coeffcients for Table 3.14 Data
Parameter SS DF MS F F
crit
a
1
 = 30.1* 1800.12 1 1800.12 12,559* 10.13
a
2
 = 15.0* 799.62 1 799.62 5579* 10.13
a
12
 = 5.1* 103.02 1 103.02 719* 10.13
a
11
 = 0.2* + a
22
 = 3.9 100.85 2 50.43 352* 9.55
B 0.06 3 0.02 0.1 9.28
E 0.43 3 0.14
T 2805.79 11
* Signifcant at p = 0.05.
Table 3.16
ANOVA with Coeffcients for Table 3.14 Data Having a
Pooled Residual
Parameter SS DF MS F F
crit
a
1
 = 30.1* 1800.12 1 1800.12 12,559* 5.99
a
2
 = 15.0* 799.62 1 799.62 5579* 5.99
a
12
 = 5.1* 103.02 1 103.02 719* 5.99
a
11
 = 0.2* + a
22
 = 3.9 100.85 2 50.43 352* 5.14
R 0.49 6 0.08
T 2805.79 11
* Signifcant at p = 0.05.
© 2011 by Taylor and Francis Group, LLC
Experimental Design 75
Equation 3.32 shows the inverse matrix, (X
T
X)
−1
.
Note the matrix is not quite diagonal with the quadratic
entries mutually biasing one another:
X X
T
( ) =













 −1
1 12
1 8
1 8
1 4
5 32 1 32
1 32 5 32




.
(3.32)
Table 3.17 presents the results from an analysis of the
standard errors of each effect. Dividing the coeffcient
value by the standard error for each effect gives the t
ratio. The critical t ratio is given by t
crit
 = √F
crit
(p, 1, DFR)
ratio. Here we see that a
22
exceeds t
crit
while a
11
does not.
This method is not foolproof since a
11
and a
22
are mutu-
ally biased. However, for central composite designs the
degree of bias is often slight.
3.14 Blocking
Designs having large numbers of experimental runs
or experiments run sequentially often cannot be run
under homogeneous conditions. This is because dur-
ing intermittent periods, important factor changes
may occur. These include ambient factors, time-based
factors such as settling or aging of raw materials, and
batch effects. If we presume that such effects will alter
the mean response but not dramatically affect other
coeffcient values, we can eliminate bias in the analy-
sis by a technique known as blocking. In general one
should “block whenever possible and randomize the
remainder.”
For example, suppose we wish to test a burner using
a full factorial design in four factors. However, we only
have a limited time to run the tests. Perhaps the burner
is only available for two days separated by some inter-
vening gap of days or weeks. During the intervening
time many things could change. Perhaps the oil fuel
quality changes over time, or unbeknown to us, some
ambient conditions are important and these condi-
tions have changed. Anything that is not specifcally
accounted for in the declared model will accumulate in
the error term, infating it. That is, in a fully random-
ized experiment, the error term will contain not only
random error but also systematic bias of which we are
unaware. This kind of bias cannot be measured because
we do not even know that it exists. Notwithstanding,
our ignorance will not prevent this infuential step
change from affecting our results. Table 3.18 shows a
suggested way of blocking the design. In the frst set of
tests we run points 2, 3, 5, 8, 9, 12, 14, 15 in randomized
order. In the second set of tests we run the remainder,
also in a randomized order.
Before describing how we arrived at such a block-
ing arrangement, let us examine a very poor blocking
arrangement—namely, Block I = Points 1–7, Block II  = 
Points 8–16. If we use this blocking arrangement, and
we fail to randomize runs, we will have committed two
sins and completely confounded any block effect with
the factor effect, x
1
. The block effect will be ascribed
falsely to x
1
because x
1
is run at its low level in tests 1–7
and at its high level in runs 8–16. In short, x
1
changes
when the block changes and the effects are hopelessly
entangled. What we desire is a block effect that will be
orthogonal to the effects of interest.
However, we tacitly learned how to block orthogo-
nally when we learned how to fractionate a design. We
Table 3.17
Analysis of Standard Error of Effects
Coeffcient s
k
t Ratio t
crit
a
0
 = 30.1* 0.08 365.68* 2.45
a
1
 = 15.0* 0.10 149.04* 2.45
a
2
 = 10.0* 0.10 99.34* 2.45
a
12
 = 5.1* 0.14 35.66* 2.45
a
11
 = 0.2 0.11 2.11 2.45
a
22
 = 4.0* 0.11 35.21* 2.45
* Signifcant at p = 0.05.
Table 3.18
A 2
4
Factorial Design in Two Blocks
Point x
1
x
2
x
3
x
4
Block
1 − − − − II
2 − − − + I
3 − − + − I
4 − − + + II
5 − + − − I
6 − + − + II
7 − + + − II
8 − + + + I
9 + − − − I
10 + − − + II
11 + − + − II
12 + − + + I
13 + + − − II
14 + + − + I
15 + + + − I
16 + + + + II
© 2011 by Taylor and Francis Group, LLC
76 Industrial Combustion Testing
should assign our blocks in the same way: use a defn-
ing contrast. Since there is only one block, we use the
largest possible contrast—1234B where B stands for the
block effect. For this reason, the blocking pattern takes
on a similar pattern to x
5
in the fractional factorial of
Table 3.12. We can also add some center points to each
block and randomize them along with the rest of the
runs in order to assess pure error within each block.
The axial and factorial points of a central composite
design are also orthogonal to one another, so one can
run those designs sequentially and subtract the block
effects from the error, thus making the F tests more
sensitive.
What if we need four blocks instead of two? Then,
we would add two blocking factors (b
1
, b
2
) in the same
way that we would add two additional factors in a frac-
tional factorial. We would then assign the four com-
binations to a block. For example, (− −) = I, (− +) = II,
+ −) = III, and (+ +) = IV. Table 3.19 gives an example
with defning contrasts being 124b
1
, 34b
2
, and 123b
1
b
2
.
Therefore, 124, 34, and 123 are confounded with block
effects. Let us presume that we can tolerate no more
than six runs at any time and that there is a possibility
that signifcant changes that affect our response may
occur between blocks. Table 3.20 gives the factor effects.
Tables 3.20 and 3.21 show the associated t and F tests.
For these data, lack of ft turns out not to be signifcant.
Therefore, we may use the pooled residual to perform
our tests.
Table 3.19
A 2
4
Full Factorial Design in Four Blocks with Eight
Center Points
Point x
1
x
2
x
3
x
4
b
1
b
2
Block y
1 + + + − − −
I
19.5
2 − − + − − − 16.7
3 + − − + − − 26.4
4 − + − + − − 12.1
5 0 0 0 0 0 0 17.5
6 0 0 0 0 0 0 18.1
7 − + + + − +
II
27.2
8 + − + + − + 41.0
9 + + − − − + 7.0
10 − − − − − + 4.5
11 0 0 0 0 0 0 20.7
12 0 0 0 0 0 0 19.4
13 − − − + + −
III
16.5
14 + − + − + − 31.9
15 − + + − + − 15.6
16 + + − + + − 21.1
17 0 0 0 0 0 0 20.5
18 0 0 0 0 0 0 20.2
19 + + + + + +
IV
35.3
20 + − − − + + 17.5
21 − + − − + + 1.5
22 − − + + + + 30.3
23 0 0 0 0 0 0 20.2
24 0 0 0 0 0 0 21.0
Table 3.20
t Tests for Factor Effects for Table 19 Data
Coeffcient Value s
k
t
a
0
* 20.07 0.16 126.1*
a
1
* 4.71 0.19 24.1*
a
2
* −2.84 0.19 −14.6*
a
3
* 6.93 0.19 35.6*
a
4
* 5.98 0.19 30.7*
a
12
* −1.39 0.19 −7.1*
a
13
0.03 0.19 0.2
a
14
0.01 0.19 0.0
a
23
0.06 0.19 0.3
a
24
* 0.53 0.19 2.7
a
34
−0.34 0.34 −1.0
a
123
0.16 0.34 0.5
a
124
0.18 0.34 0.5
a
134
−0.04 0.19 −0.2
a
234
0.06 0.19 0.3
a
1234
0.13 0.19 0.7
Block I −1.90 0.48 −4.0
Block II 0.35 0.48 0.7
Block III 0.65 0.48 1.4
Block IV 0.90 0.48 1.9
* Signifcant at p = 0.05, t
crit
 = 2.57.
Table 3.21
ANOVA for Table 3.19 Data
Coeffcient DF SS MS F F
crit
a
1
1 354.38 354.38 582.82 6.61
a
2
1 129.39 129.39 212.80 6.61
a
3
1 768.68 768.68 1264.18 6.61
a
4
1 572.41 572.41 941.39 6.61
a
12
1 31.08 31.08 51.12 6.61
a
13
1 0.02 0.02 0.03 6.61
a
14
1 0.00 0.00 0.00 6.61
a
23
1 0.05 0.05 0.08 6.61
a
24
1 4.52 4.52 7.43 6.61
a
34
1 0.63 0.63 1.04 6.61
a
123
1 0.13 0.13 0.21 6.61
a
124
1 0.18 0.18 0.29 6.61
a
134
1 0.03 0.03 0.05 6.61
a
234
1 0.05 0.05 0.08 6.61
a
1234
1 0.28 0.28 0.45 6.61
Blocks 3 9.93 3.31 5.44 5.41
R 5 3.04 0.61 0.09
T 23 1874.78
© 2011 by Taylor and Francis Group, LLC
77
4.1 Introduction
The hydrocarbon and petrochemical industries employ
the fow of many fuids. For example, combustion equip-
ment requires the fow of air and fuel. Then there are the
process streams themselves comprising hydrocarbon
liquids and gases fowing under pressure, through
pipes and fttings, to their delivery point. Calculation
of fuid fow, therefore, is of fundamental importance.
The purpose of this chapter is to give the reader a fun-
damental understanding of some of the fuid dynamic
concepts and equipment that are important in testing
combustion systems.
4
Fluid Flow
Wes Bussman and Joseph Colannino
CONTENTS
4.1 Introduction ..................................................................................................................................................................... 77
4.2 Gas Properties ................................................................................................................................................................. 78
4.2.1 Density ................................................................................................................................................................. 78
4.2.2 Ratio of Specifc Heats and Related Concepts ................................................................................................ 78
4.2.3 Viscosity ............................................................................................................................................................... 78
4.2.4 Fuel Heating Value ............................................................................................................................................. 78
4.3 Basic Fluid Dynamic Concepts ..................................................................................................................................... 79
4.3.1 Defnition and Units of Pressure ...................................................................................................................... 79
4.3.2 Gauge and Absolute Pressure ........................................................................................................................... 79
4.3.3 Draft ...................................................................................................................................................................... 79
4.3.4 Compressible Flow through an Orifce ........................................................................................................... 80
4.3.5 Pressure Drop in Pipes and Fittings ................................................................................................................ 81
4.4 Pressure Measurement Techniques ............................................................................................................................. 82
4.4.1 Manometer ........................................................................................................................................................... 82
4.4.1.1 U-Tube Manometer .............................................................................................................................. 82
4.4.1.2 Inclined Manometer ............................................................................................................................ 82
4.4.2 Bourdon Tube Gauge .......................................................................................................................................... 83
4.4.2.1 Design of the Bourdon Tube Gauge .................................................................................................. 83
4.4.2.2 Common Failure Mechanisms ........................................................................................................... 84
4.4.2.3 Calibration of Pressure Gauges .......................................................................................................... 85
4.4.2.4 Selection ................................................................................................................................................. 85
4.4.2.5 Installation ............................................................................................................................................ 85
4.5 Differential Producing Flow Meters ............................................................................................................................ 86
4.5.1 Orifce Meter ........................................................................................................................................................ 86
4.5.2 Venturi Meter ...................................................................................................................................................... 89
4.5.3 Turbine Flow Meter ............................................................................................................................................ 89
4.5.4 Vortex Flow Meter .............................................................................................................................................. 89
4.5.5 Magnetic Flow Meter ......................................................................................................................................... 90
4.5.6 Ultrasonic Flow Meter ........................................................................................................................................ 90
4.5.7 Thermal Mass Meter .......................................................................................................................................... 91
4.5.8 Positive Displacement Meter ............................................................................................................................. 91
4.5.9 Pitot Tube ............................................................................................................................................................. 92
4.5.10 Annubar ............................................................................................................................................................... 94
References .................................................................................................................................................................................. 95
© 2011 by Taylor and Francis Group, LLC
78 Industrial Combustion Testing
4.2 Gas Properties
4.2.1 Density
Density (ρ) is defned as mass (m) divided by volume
(V). Accordingly, units of density are g/cm
3
, kg/m
3
,
lbm/ft
3
, and so forth. When the volume is macroscopic
then, technically, the density is an average density. The
general way to defne the density at a point is to use the
limit at infnitesimal volume.
ρ
dm
dV
; (4.1)
however, this defnition breaks down as we near the
molecular level. Therefore, we shall presume that in all
cases that matter this may be considered a continuum.
4.2.2 ratio of Specific Heats and related Concepts
The specifc heat represents the amount of energy
required to raise a unit mass one unit in temperature.
For gases, the specifc heat differs depending on whether
the gas is allowed to do work by expanding against an
atmosphere (constant pressure defnition, c
p
) or is con-
strained within a volume (constant volume defnition, c
v
).
Examples of units of c
p
and c
v
are J/g K or Btu/lbm °F. If
we wish to express this on a molar basis we shall use the
upper case: that is, C
p
or C
v
having units of J/mol K or
Btu/lbmol °F.
The universal gas constant (R) is defned for an ideal
gas as follows:
R = PV/nT, (4.2)
where
P is the pressure,
V is the volume,
n is the number of moles, and
T is the absolute temperature.
By ideal gas we mean a gas whose atomic and molecu-
lar volumes and attraction entities exert no signifcant
infuence upon the bulk volume or pressure. All real
gases deviate from ideality at very low temperatures
and very high pressures. In many cases (and especially
so for combustion) the ideal gas law is suffciently accu-
rate for engineering purposes. Remarkably, the univer-
sal gas constant and heat capacities are related:
C
p
− C
v
 = R. (4.3)
It also turns out that the dimensionless ratio of heat
capacities (k) is an important thermodynamic parameter:
k
c
c
C
C
p
v
p
v
. (4.4)
4.2.3 Viscosity
Shear stress (τ) is the force per unit area exerted due to
friction of a fowing fuid. If a stationary wall is exposed
to a fowing fuid, the velocity at the wall will be zero
and the velocity far from the wall will be the free stream
velocity. Between these two points a velocity gradient
exists. The viscosity (µ) is the constant of proportional-
ity between the velocity gradient (dv/dy) and the shear
stress. Figure 4.1 shows the coordinate directions and
the variation in velocity from zero at the plate to the free
stream velocity, v
s
. The velocity is fowing in the x direc-
tion and the velocity gradient varies in the y direction.
The shear stress for ideal fowing gases is given by the
following equation and gauges the resistance to estab-
lishing a rate of strain (dislocation of the fuid)
τ µ
dy
dx
. (4.5)
4.2.4 Fuel Heating Value
The fuel heating value is the total amount of energy
that can be liberated by combustion of a fuel. The heat-
ing value may be normalized by mass, moles, or vol-
ume; accordingly, it will have units of kJ/kg, J/mol, or
J/Nm
3
. Customary units include Btu/lbm, Btu/lbmol,
and Btu/scf. In the case of volume, some standard
temperature and pressure must be chosen, such as 25°C
and 1 atm. Two examples of unit volumes referenced to
such conditions are the normal meter cubed (Nm
3
) and
the standard cubic foot (scf).
For hydrocarbons, combustion always produces water
and carbon dioxide according to the following equation:
C
x
H
y
 + (x + y/4)O
2
 = xCO
2
 + y/2 H
2
O. (4.6)
For combustion reactions producing water, we may
consider two different standard states for calculating
the heating value: the standard state may be taken to
be either a liquid or vapor at the standard temperature
and pressure. If the standard state considers water as
a vapor, the heating value so determined is called the
y
x
v
s
Figure 4.1
Fluid fow shear stress.
© 2011 by Taylor and Francis Group, LLC
Fluid Flow 79
lower heating value (LHV) or net heating value. If the
standard state considers water as a liquid, then the heat-
ing value so determined is called the higher heating
value (HHV) or gross heating value. It is important to
understand that this determination is set by convention
and has nothing to do with the actual state of the water.
The difference between the higher and LHVs is the
standard enthalpy of vaporization for water. Some com-
bustion produces no water (e.g., combustion of CO) and
therefore such higher and LHVs are identical.
4.3 Basic Fluid Dynamic Concepts
4.3.1 Definition and units of Pressure
Pressure (P) is defned as a normal force per unit area.
For example, the pressure of a standard atmosphere
at sea level is ~14.7 lbf/in
2
(psi). Other pressure units
include kilopascals (kPa), bar (101.3 kPa), and atmo-
spheres (atm).
4.3.2 gauge and absolute Pressure
Although the pressure at sea level is 14.7 psi, pressure
gauges are calibrated to read 0 at 1 atmospheric pressure.
Therefore, when one measures pressure in an automo-
bile tire, say 32 psig, one is measuring the pressure above
atmospheric. In this example, the total pressure (also
called the absolute pressure) in the tire is actually 46.7 psia
because the pressure at sea level is 14.7 psia. In order to
distinguish between gauge (g) and absolute (a) pressure,
the respective letter is appended to the psi designation.
4.3.3 Draft
Draft is a differential pressure between the outside atmo-
sphere and the internal pressure in a furnace. Differential
pressure such as units of draft are sometimes referenced
as psid. However, draft pressures are ordinarily so low
that smaller units are needed. Two common units are mm
w.c. or in w.c. (also w.c. or i.w.c.); that is, millimeters, water
column or inches, water column, respectively. For example,
in order to displace a fuid from a glass to one’s mouth
three inches above the liquid surface via a straw, one must
cause a differential pressure of 3 in. w.c. Draft pressures
are often smaller. For example, a typical natural draft
refnery furnace operates anywhere from 0.25 in. w.c. to
0.50 in. w.c. This seemingly small pressure difference can
cause great volumes of air fow. Gas pressure and velocity
are related by Bernoulli’s equation:

∆P v
ρ

2
2
. (4.7)
Bernoulli’s equation, as written above, considers no
losses due to friction or microscopic recirculation.
In order to calculate the total draft for a furnace of
a given height (h) and temperature, one may equate
the potential energy of the gas owing to its mass in a
gravitational feld with that of the ambient fuid outside
the furnace; that is, air in the usual case. This gives
ΔP/Δρ = gh or ΔP = Δρgh where Δρ is the difference in
densities between the two fuids. Noting that Δρ = PM/
RΔT where Δρ is the density, P is the ambient pressure,
M is the molecular weight, R is the universal gas con-
stant, and ∆T is the absolute temperature difference, we
may recast the equation as

∆P
PMgh
R T T
f

|
.

`
}


1 1
, (4.8)
where T

is the absolute ambient temperature and T
f
is
the absolute furnace temperature.
example 4.1: Calculation of Draft Pressure
Problem statement: What is the maximum draft pressure at
the foor of a 10-m furnace operating at 800°C? Presume
that the control system adjusts the pressure at the top of
the furnace to a draft of 3 mm w.c. via a stack damper.
Solution: From Equation 2.110, at 1 atm of pressure (101 kPa)
we have

]
]
]

]
]
P
g
101 29 [ ] kPa
mol
8.314
m Pa
mol K
3
]]
|
.

`
}

+

+
|
.

`
}
1
800 273
1
25 273 ( )[ ] ( )[ ] K K

]
]
]

9 8 10
54 6
. [ ]
. [ ][ ]
m
s
m
Pa mmw.c.
2
The negative sign reminds us that the pressure is below
atmospheric. Adding the −3 mm w.c. pressure (draft) we
have at the bridgewall gives a total of −8.6 mm w.c. pressure
at the foor (or 8.6 mm w.c. draft).
We can also write an equation for the mass fow (m˙ ) as
a function of velocity and density:


m Av ρ , (4.9)
where A is the fow area. Coupling these two equa-
tions one arrives at mass fow as a function of draft pres-
sure through an area:

m A P 2ρ ∆ . For purposes of the
calculation, the absolute value of the pressure is used
with the understanding that the fow direction is always
defned as fow from the higher to the lower pressure.
However, as previously noted, Bernoulli’s equation does
not consider friction loss. We can account for friction by
premultiplying by a loss coeffcient, C
o
. In that case, the
© 2011 by Taylor and Francis Group, LLC
80 Industrial Combustion Testing
mass fow rate as a function of pressure and density
becomes


m C A P
o
2ρ ∆ . (4.10)
For British units, one should note that 32.174 lbm.ft/lbf−s
2

= 1 = g
c
, where lbm is the pound mass unit and lbf is
the pound force unit. Force and mass are related by
Newton’s second law,
F = mg, (4.11)
where
F is the force,
m is the mass, and
g is the gravitational acceleration (9.8 m/s
2
or 32.174
ft/s
2
).
By defnition, 1 lbm at 1 g of gravity exerts 1 lbf.
Therefore, 1 lbf = 32.174 lbm ft/s
2
. Since g
c
 = 1 by defni-
tion, the incompressible fow equation is often written as


m C A g P
o c
2 ρ ∆ , (4.12)
when using British units. This would not be necessary
except that engineers before Newton did not recognize
the difference between mass and force and developed
an inconsistent set of units. SI units, or any other consis-
tent set of units, do not require the use of g
c
.
example 4.2: Calculation of Mass
Flow Through Hole in a Furnace
Problem Statement: A furnace has a leak equating to a total
fow area of 4 in
2
at an elevation in the heater corresponding
to −0.25 in. w.c. of draft. Presuming an orifce coeffcient
of 0.80, what is the mass fow of air that this represents?
Presume that the ambient air density is 0.075 lbm/ft
3
.
Solution: We use the equation

m C A g P
o c
2 ρ| | ∆ with
the following information:
C
o
 = 0.80
A = 4 in
2
ρ = 0.075 lbm/ft
3
∆P = −0.25″ w.c.
Therefore, we have
We also note that there are 144 in
2
per 1 ft
2
and 27.69 in
w.c. per 1 lbf/in
2
. With these conversions we have
The incompressible fow equation used here is accurate
for natural draft devices where the resulting air velocity is
much lower than the speed of sound. However, it is not
a good assumption for fuel fow from a fuel orifce as fuel
pressures are normally 25 psig or so and the exit velocity
is sonic. For these cases, we must make use of a more
inclusive form of the Bernoulli equation.
4.3.4 Compressible Flow through an Orifice
We shall fnd it convenient to defne a pressure ratio (r
o
)
that is a function of the absolute pressure upstream of
the orifce (P
o
) and the absolute pressure surrounding the
orifce into which the fuid is exhausting (P

):
r
P
P
o
o


. (4.13)
As the pressure increases, the velocity through the
orifce also increases until it reaches a natural limit
known as the speed of sound. This limit occurs at a
critical pressure ratio (r
c
). This critical ratio is given by
the following equation:
r
k
c
k
k

+
( )
− 1
2
1
. (4.14)
Considering compressible fow through an orifce the
following equation gives the mass fow. An exact deriva-
tion is available at http://www.combustion-modeling.com


m
C A P
M
RT
k
k
r
r
o o o
o
k
k
o

( )

( )

]
]
]


β
β
γ
2
1
2
2
1
1
44
2
1
4
2
1
1
2
r r
C A P
M
RT
k
k
k
r r
o c
o o o
k
k
o c
<
( )
+
+
( )



β
β
¹¹
´
¹
¹
¹
¹
¹
¹
¹
¹
¹
¹
¹
,
(4.15)


m ( )( )[ ] ( )

]
]
0 80 4 2 32 174 . . in
lbm ft
lbf s
2
2 ]]

]
]
]
[ ] 0 075 0 25 . . .
lbm
ft
in w.c.
3

m ( )( )[ ]
[ ]
[ ]
( ) 0 80 4
1
144
2 32 174 . . in
ft
in
lbm
2
2
2
ft
lbf s
lbm
ft
in w.c
2 3

]
]
]

]
]
]
0 075 0 25 . . ..
1
lbf
in
in w.c.
in
ft
2
2
2
[ ]

]
]
]
[ ]

27 69
144
.
]]
]
]
= 0.056 lbm/s (or 200.4 lbm/h).
© 2011 by Taylor and Francis Group, LLC
Fluid Flow 81
where β is the ratio of the orifce diameter to the upstream
pipe or duct diameter, A
o
is the area of the orifce, and
T is the upstream temperature. For monoatomic gases
such as helium or argon, k ≈ 5/3. For diatomic gases
such as air, oxygen, or nitrogen, k ≈ 7/5. For polyatomic
gases k approaches 1 as the number of atoms increases to
infnity; however, for common polyatomic gases having
3 to 5 atoms, 9/7 < k < 8/6 and an average value of 1.3 is
commonly used. Although k is not a strong function of
temperature (much weaker than either C
p
or C
v
alone)
at higher temperatures more energy states become
available and this also tends to drive k lower. However,
for any particular gas or temperature, exact values of k
are available.
4.3.5 Pressure Drop in Pipes and Fittings
Although velocity through the fnal gas orifce is often
sonic, for industrial combustion the usual practice is to
make the upstream diameter of the duct or pipe many
times larger than the orifce and to keep the fow much
below sonic. Also, liquids, even under high pressure, may
be generally regarded as incompressible. Therefore, we
may treat pressure drop in pipes and fttings as incom-
pressible fow. A friction factor ( f ) is used to correct for
friction. The general relation is
∆P f
L
D
K
v
g
c
+
|
.

`
}


ρ
2
2
, (4.16)
where L is the length of the duct, D is its diameter, and
∑K is the sum of correction factors for bends and fttings.
example 4.3: Calculation of Pressure
Drop Through a Pipe
Problem Statement: Consider a liquid hydrocarbon mixture
fowing at 50 ft/s and having the following properties.
ρ = 50 lbm/ft3
f = 0.023
L = 100 ft
D = 0.5 ft
ΣK = 4.5
What is the pressure drop in psi?
Solution: Applying the pressure drop equation gives the
following:
∆P
[ ]
[ ]
+
|
.

`
}

0 023
100
0 5
4 5
50
.
.
.
ft
ft
lbm
ft
3
]]
]
]

]
]
]
( )

]
]
50
2 32 174
2
2
ft
s
lbm ft
lbf s
2
.
]]
( )

]
]
]

144
122 8
in
ft
lbf in
2
2
2
. / .
Figure 4.2 gives the Moody Friction Factor Chart. This
chart allows f to be read as a function of pipe roughness, ε,
divided by pipe diameter (ε/D, the so called relative rough-
ness) and the Reynolds number (Re = Dvρ/µ), where µ is
the viscosity of the fuid. One can also solve the Colebrook
equation iteratively to fnd f:

1
2
3 7
2 51
10
f
D
f
− +
( )
log
.
.
Re
.
ε
(4.17)
0.08
0.1
0.07
0.06
0.05
0.04
0.03
0.02
0.01
0.009
0.09
1000
Reynolds number
10
6
10
4
10
5
10
7
10
8
F
r
i
c
t
i
o
n

f
a
c
t
o
r

f
0.05
0.04
0.03
0.02
0.01
0.006
0.000 01
z/D
0.000 05
0.000 1
0.000 2
0.000 4
0.000 6
0.001
0.002
0.004
0.015
0.000 005
0 (smooth)
f = 64/Re
2 3 4 6 8 2 3 4 6 8 2 3 4 6 8 2 3 4 6 8 2 3 4 6 8
Figure 4.2
U-tube manometer.
© 2011 by Taylor and Francis Group, LLC
82 Industrial Combustion Testing
The Haaland equation provides a useful approximation
that does not require iteration:

1
1 8
3 7
6 9
10
1 11
f
D

( )
+

]
]
]
. log
.
.
Re
.
.
ε
(4.18)
example 4.4: Calculation of Friction Factor
Problem Statement: Use the Colebrook and Haaland
equations to calculate the friction factor with ε/D = 0.008
and Re = 8 × 10
5
.
Solution: We begin with the Haaland equation, as it does
not require any iteration. This gives

1
1 8
0 0008
3 7
6 9
2 10
10
1 11
6
f

( )
+

]
]
]
. log
.
.
.
.
×
77 18
0 0194
. ,
. .
and
f
The Colebrook equation yields

1
2
3 7
2 51
0 0188
10
f
D
f
− +
( )
log
.
.
Re
. .
ε
Thus, the Haaland equation is a bit more than 3% too high.
This is well within the accuracy of the calculation; indeed,
the relative roughness is never known with certainty and can
change with use.
4.4 Pressure Measurement Techniques
4.4.1 Manometer
One of the simplest and most useful pressure measuring
devices is the manometer. There are several variations
of the manometer. In this section we will discuss two
types: U-tube and inclined manometer.
4.4.1.1 U-Tube Manometer
In its most basic form, the manometer takes the form of
a U-shaped tube that is partly flled with some liquid,
such as water, oil, or mercury, as shown in Figure 4.3.
The differential pressure in a U-tube manometer can be
expressed as
P h
d
γ , (4.19)
where
P
d
 ≡ differential pressure (lb
f
/ft
2
),
γ ≡ specifc weight of the fuid (lb
f
/ft
3
),
h ≡ height of the liquid (ft).
The specifc weight of water and mercury is 62.4 lb
f
/ft
3

and 847lb
f
/ft
3
, respectively.
Example: A U-tube manometer, flled with mercury, is
connected to a pressurized vessel. The difference in the
height of mercury reads 60-inch. (a) What is the pres-
sure in units of pounds per square inch? (b) What would
the difference in height be if the manometer was flled
with water?
(a)
lb
ft
60 in
1ft
12in
1ft
144in
f
3
2
2
P h
d
× × γ 847 229 4 . psi,

(b)
mercury
water
mercury
water
water
h
h
P
P
h
d
d

γ
γ
hh
mercury
mercury
water
in
γ
γ
× 60
847
64 2
814 4
.
. in.
This example illustrates that mercury-flled, U-tube
manometers are reasonable to use in pressure ranges
as high as 30 psig; at this pressure, a 5-foot (60 inches)
manometer is required. A water-flled manometer,
however, is not reasonable at this pressure because it
would require a U-tube manometer over 68-feet (815
inches) long.
4.4.1.2 Inclined Manometer
In combustion applications it is often important to be
able to measure very small pressure levels accurately.
Clear tube Oil-filled Water-filled
Pressure
differential
Liquid-filled
Scale
(in. WC)
4
3
2
1
0
1
2
3
4
Figure 4.3
U-tube manometer.
© 2011 by Taylor and Francis Group, LLC
Fluid Flow 83
For example, it is common to be able to measure the draft
inside of a heater with an accuracy of 1/100th of an inch
of water column (0.00036 psi). With U-tube monometers,
it is not possible to read pressures this accurate; instead,
inclined manometers are commonly used.
Inclined manometers essentially expand the scale of
a U-tube manometer by orienting it at an angle, relative
to the horizontal, as illustrated in Figure 4.4. Although
the actual vertical rise of the liquid in the U-tube and
inclined manometer are identical, the scale on the
inclined manometer is spread out over a much larger
horizontal distance; this expanded scale increases the
instruments accuracy.
Inclined manometers consist of a clear tube positioned
at a slight angle relative to the horizontal as shown in
Figure 4.5. The tube is typically flled with red oil with a
specifc gravity of approximately 0.8. Located at the top
of the manometer are two pressure taps: a high pres-
sure side and low pressure side. The low pressure side
tap contains the oil reservoir. For accurate measurement
it is important that the manometer is level with the
horizontal. Typically inclined manometers are designed
with a bubble level to assist in accurate leveling.
The units typically used in the combustion industry
to measure furnace draft are expressed in inches or
millimeters of water column (in WC or mm WC). A col-
umn of water 27.68 inches high will create a hydrostatic
pressure of 1 pound per square inch.
4.4.2 bourdon Tube gauge
In 1849 Eugene Bourdon, a French watchmaker and
engineer, patented an instrument used to measure
pressure. This instrument is called the Bourdon tube
gauge and is one of the most common instruments
used in industrial plants. The purpose of this section
is to discuss the design, failure, and calibration of
these gauges.
4.4.2.1 Design of the Bourdon Tube Gauge
Bourdon tube gauges are available in a variety of sizes
(see Figure 4.6). The diameter of the dial faces can vary
from less than 1 inch (2.5 centimeters) to as large as 3 feet
(0.9 meter), with the largest typically used for extreme
accuracy [1]. Bourdon tube gauges are also available in
a variety of pressure ranges: from 5 psig full-scale to as
much as 10,000 psig full-scale [2].
The Bourdon gauge consists of a fattened, thin-walled
metal tube bent into the form of an arc or C shape as
illustrated in Figures 4.7 and 4.8 [3]. One end of the tube
is fxed and securely sealed and bonded to a threaded
connection called a socket. The other end is tightly
sealed and free to move. At atmospheric pressure (zero
gauge pressure) the tube is undefected and for this
condition the gauge pointer is calibrated to read zero
pressure [4]. However, when pressure is applied into
the fxed end of the tube, the effect of the forces tends
Figure 4.6
Basic components of a Bourdon pressure gauge.
Expanded scale
Inclined manometer
U-tube manometer
1 cm 1 cm
0
0
.2
0
.4
0
.6
0
.8
1
.0
Figure 4.4
Showing how an inclined manometer expands the measurements
scale.
Clear tube
High pressure-side tap
Oil reservoir
Bubble level
Low pressure-side tap
Figure 4.5
Basic components of a Bourdon pressure gauge.
© 2011 by Taylor and Francis Group, LLC
84 Industrial Combustion Testing
to uncoil or straighten the tube causing the end to move
in a slightly curved path. The end of the Bourdon tube
does not move a great distance within its pressure range:
typically 0.125 to 0.25 inch (0.31 to 0.63 centimeter) [5].
Using a gear and linkage system (called the movement)
the motion of the tube is magnifed and converted into
rotary movement of the indicating needle. The indicat-
ing needle rotates around a circular dial calibrated for
pressure [6].
The indicating needle is commonly fxed to a spiral
spring that tightens up as more pressure is applied. The
purpose of the spiral spring is to assist the tube and
pointer to quickly respond to a drop in pressure after
reading some higher value.
The most important part of the gauge is the Bourdon
tube. Bourdon tubes are made of many materials: beryl-
lium copper, phosphor bronze, and various alloys of steel
and stainless steel [7]. Beryllium copper is typically used
for high pressure applications. Most gauges in air, light
oil, or water applications use phosphor bronze. Stainless
steel alloys usually add cost to the gauge if specifc cor-
rosion resistance is not required.
The socket is usually made of brass, steel, or stainless
steel. The most common socket size is usually ¼-inch
and ½-inch National Pipe Thread (NPT) tapered pipe
threads. For dial sizes smaller than 4.5 inches, a ¼-inch
NPT is common; however, in some cases a 1/8-inch NPT
will be used. For pressure-gauge dial sizes smaller than
2 inches, a 1/8 NPT is common.
The movement mechanism can be made from a variety
of material such as nickel–silver, brass, or stainless steel;
brass is a very common material. It is important that the
material used allows for a friction-free assembly. The
movement mechanism is protected within an enclosure
commonly made of brass, aluminum, steel, or plastic.
Most lenses, protecting the indicating needle and dial
are made of plastic instead of glass for safety reasons
due to breakage.
4.4.2.2 Common Failure Mechanisms
Since a typical plant will use many pressure gauges, they
do not always receive the proper maintenance causing
them to be unreliable. Three common reasons for gauge
failure are harsh environments, vibrations, and water
condensation.
Harsh environments such as corrosive and dirty fuids
can cause a pressure gauge to fail. Bourdon tubes can
be made from a variety of materials and are most often
manufactured from 316 stainless steel, phosphor bronze,
Monel, and Inconel [8]. The selection of material depends
on the fuid for the application. If the application consists
of a highly corrosive fuid then a diaphragm seal can be
used with the gauge; the seal prevents the fuid from
coming in contact with the Bourdon tube. Diaphragm
seals not only protect a pressure gauge from corrosion
attack, it can also prevent dirt and scale from clogging
the tube. The diaphragm seal will typically reduce the
accuracy of the gauge.
If a gauge is used in a vibrating environment, it can
cause excessive motion of the movement mechanism
causing it to wear and fail prematurely. Vibrating envi-
ronments also makes it diffcult to read the pressure
gauge accurately because the indicating needle oscil-
lates. To help prevent this problem, pressure gauges are
often flled with a dampening fuid such as glycerin or a
silicon fuid. The fuid helps cushion the tube and move-
ment mechanism against damage from sudden impact
and vibration. Many industrial plants have adopted liq-
uid-flled gauges as a standard.
Excessive motion of the movement mechanism can
also occur if rapid pressure changes occur within a pro-
cess. Pressure snubbers are commonly used to protect
pressure gauges from the system pressure shock and
surges and can prolong the life of a gauge. The pressure
snubber is placed inline between the gauge and the pro-
cess as shown in Figure 4.9 [9] and are typically designed
with an orifce or a porous metal disk to help dampen
pressure fuctuations.
Tube movement
Bourdon
tube
Spiral
spring
Indicating
needle
Socket
Fluid in
Gear and
linkage
system
(movement)
Figure 4.8
Basic components of a Bourdon tube pressure gauge. (Adapted from
Hydraulic Energy Transmission and Control, Hill Learning Systems,
2001–2007, http://www.hetacfuidpower.com/screen_shot_gallery.
htm)
Figure 4.7
Internal view of the Bourdon pressure gauge.
© 2011 by Taylor and Francis Group, LLC
Fluid Flow 85
4.4.2.3 Calibration of Pressure Gauges
To insure accuracy, it is customary to calibrate pres-
sure gauges periodically. Depending on their use and
importance to a given process, gauges might be cali-
brated monthly, quarterly, yearly, or every several years.
Deadweight testers are the primary standard for accu-
rate pressure calibration. A typical oil-type deadweight
tester is shown in Figure 4.10.
The basic components of an oil-type deadweight
tester are shown in Figure 4.11. These testers consist
of an accurately honed piston of a known cross-sec-
tional area inserted into a cylinder. Known weights
(standard masses) are placed on a freely rotating plat-
form attached to the top of the piston; the platform is
designed to freely rotate to reduce friction between
the piston and cylinder. When the oil pump supplies
enough pressure to raise the weights, the force exerted
by the oil pressure over the piston area is balanced by
the force of the weights. The pressure is calculated as
follows:
P
gauge
 = F/A
piston
. (4.20)
When using a deadweight tester, it is important to
apply corrections for local gravity, elevation, buoyancy,
or thermal expansion of the piston, since these errors
may be signifcant [10]. Other sources of error include
weights that are dirty, corroded, or chipped or using a
tester that is not properly leveled [11].
4.4.2.4 Selection
When selecting a pressure gauge, it is recommended
to use one that has a range of twice the normal work-
ing pressure. The maximum operating pressure, in all
cases, should not exceed 75% of the maximum range of
the gauge. If gauges are used in applications where pul-
sations are encountered, the pressure should be limited
to 67% of the maximum range of the gauge.
The ambient temperature range for most standard dry
or silicone flled gauges is about −40°F to + 140°F (−40°C
to 60°C). The ambient temperature range for glycerine
flled gauges is −4°F to + 140°F (−20°C to 60°C). The error
caused by temperature change is + /−0.3% per 18°F rise
or fall, respectively, based on the temperature of the
gauge.
4.4.2.5 Installation
Bourdon gauges are commonly used to measure the static
pressure of a fuid fowing in a pipe. To reduce the poten-
tial for error, it is important that static pressure taps are
Figure 4.10
An oil-type deadweight tester.
Test gauge
Standard masses
Platform
Piston and
cylinder
Oil reservoir
Displacer pump
Check
valve
Oil pressure
Figure 4.11
Basic components of a deadweight tester. (Adapted from Miller, R. W.,
Flow Measurement Engineering Handbook, New York: McGraw-Hill,
Inc., 1989.)
Female pipe thread
Orifice or porous
metal disk
Male pipe thread
Pressure snubber
Process
Figure 4.9
Pressure snubber. (Adapted from O’Keefe Controls Co., Trumbull,
CT, 2003, http://www.okcc.com/PDF/Pressure%20Snubber%20pg.42.
pdf)
© 2011 by Taylor and Francis Group, LLC
86 Industrial Combustion Testing
designed correctly. Below are several important design
considerations for static pressure taps:
The hole should be drilled perpendicular to •
the pipe.
The hole should be free of burrs. •
The edge of the hole should be square with the •
inner pipe surface with no rounded corners or no
concave or convex surface surrounding the hole.
The drilling of the hole should be done after all •
fttings for attaching the pressure gauge have
been welded to the pipe.
Ferron [12] states that countless static pressure mea-
surements are in error because of poor pressure tap
construction. Brunkalla [13] suggests a ¼ inch (6.4 mm)
diameter hole for pipe sizes greater than 2 inches (50 mm).
Departure from the recommended hole size, inclination,
or edge condition results in errors of −0.5 to + 1.1 percent
of the velocity or dynamic pressure. [10]
4.5 Differential Producing Flow Meters
The differential producing fow meter is a device used to
create varying static pressure, within a fow stream that
can be used to determine the fow rate of the fuid. These
devises have been used for over 1000 years [10]. Today,
differential fow meters are the most common and reli-
able fow meters used in industry. This section will dis-
cuss three common types of differential producing fow
meters: Orifce meter, Venturi meter, and Nozzle meter.
4.5.1 Orifice Meter
Orifce meters have been in commercial use since the
early 1900s and are the most common differential pro-
ducing fow meter used today [14,15]. The orifce meter
consists of a fat, thin plate with a circular hole machined
into the center of it as shown in Figure 4.12 and is held in
place with a holding device such as a pipe fange.
When a fuid fows through an orifce plate, the
static pressure within the pipe varies as illustrated in
Figure 4.13. Notice that just upstream of the orifce plate
the static pressure reaches a maximum value. When the
fuid passes through the bore of the plate it accelerates.
As the fuid jet exits the other side of the plate it continues
to accelerate and decrease in cross-sectional area; hence,
the minimum fow area is actually smaller than the
area of the orifce. The location where the jet reaches its
minimum cross-sectional area is referred to as the vena
contracta and is where the velocity is the highest and the
static pressure the lowest. As the fuid continues to fow
downstream, it de-accelerates, where some, but not all of
the static pressure is recovered.
Pressure taps are required on each side of the orifce
plate to allow the measurement of the pressure drop
across the plate when the fuid is fowing. Knowing the
differential pressure, the ratio of the orifce plate bore to
pipe inside diameter (beta ratio) and fuid density, the
fow rate can be calculated.
The differential pressure depends on the location of
the pressure taps. Pressure taps are located in four gen-
eral arrangements as illustrated in Figure 4.14; fange
taps, D and D/2 taps, corner taps, and pipe taps. Flange
taps are typically placed on the pipe fange and are
located 1 inch upstream and 1 inch downstream from
the face of the plate to centerline of the tap. The D and
D/2 taps are located one pipe inner diameter upstream
and one-half downstream from the face to the fange.
Corner taps are located at the face of the orifce plate.
Pipe taps are located 2.5 pipe inner diameters upstream
Figure 4.12
Orifce plate.
P2 P1
Static pressure profile
Orifice plate
Flow
Differential pressure (1c p)
P1–P2
Figure 4.13
Static pressure trend across an orifce plate. (Adapted from Smith
Metering, Inc., “Fundamentals of Orifce Metering,” http://www.
afms.org/Docs/gas/Fundamenatls_of_Orifce.pdf)
© 2011 by Taylor and Francis Group, LLC
Fluid Flow 87
and 8 diameters downstream from the face of the orifce
plate; these arrangements measure the pressure drop at
the point of full pressure recovery.
When fuid fows through elbows, tees, and valves a
swirling fow with an uneven velocity profle develops.
For accurate fow measurement, the fow upstream of
a differential pressure meter should be free of these
disturbances. In order to eliminate these disturbances,
a long length of straight pipe is required. For example,
swirl fow inside a pipe can require 50 to 100 diameters
of straight pipe to eliminate the spin [16].
Swirl-free conditions are defned as where the angle
of the swirl is less than 2° [16]. An acceptable velocity
profle condition exits when the velocity at each point
within the pipe cross-section agrees to within ± 5% of
the velocity of a very long, straight length (over 100
diameters) of similar pipe.
Flow conditioners are commonly placed upstream of
the orifce metering device to help straighten the fow
profle and reduce swirl, thus allowing shorter lengths
of upstream pipe. In general, there are three standard
types of fow conditioners: tube bundle conditioner,
Zanker straightener, and Sprenkle straightener [16]. The
tube bundle conditioner consists of a number of parallel,
thin-walled tubes connected together. The tube bun-
dles do a good job of eliminating swirl, but do a poor
job of straightening the velocity profle. The Zanker
straightener, consists of a plate with holes of a certain
size located just upstream of a number of parallel, thin-
walled square channels connected together; one channel
for each hole. The Sprenkle straightener consists of the
three perforated plates, in series, spaced one pipe diam-
eter apart. The plates are held together by studs or bolts
around the perimeter of the plates. Research and devel-
opment efforts in this area are ongoing in order to gain
better insight into the best design and location of fow
conditioners upstream of a differential metering device.
The equation for the mass fow rate through an orifce
plate starts with the Bernoulli equation:

P
g
V
g
z
P
g
V
g
z
o o
o
ρ ρ
+ + + +
2
1 1
2
1
2 2
, (4.21)
where the subscripts “o” represent the location at the
plane of the orifce bore and “1” represents the location
upstream of the orifce plate. Setting z
o
 = z
1
, ΔP = P
1
− P
o
,
and solving Equation 4.21 for V
o
, gives
V V
P
o
+
× |
.

`
}
1
2
1
2
2 ∆
ρ
. (4.22a)
Flow direction
Flow direction
D
Corner taps Pipe taps
Tap D
Flow direction
2.5D 8D
Tap
D and D/2 taps
±0.02D for d/D <= 0.6
±0.01D for d/D > 0.6
D±0.1D
±0.04 in for d/D <= 0.6
±0.02 in for d/D <= 0.6
0.5D
Taps
d D
Flow direciton
1 in 1 in
Tap Tap
d D
Flange taps
Taps
Figure 4.14
Common pressure-tap arrangements.
© 2011 by Taylor and Francis Group, LLC
88 Industrial Combustion Testing
Now, V
1
is eliminated by means of the continuity equa-
tion V
1
A
1
 = V
o
A
o
; and solving for V
o
gives
V
P
A
A
o
o

×

|
.

`
}

]
]
]
]
2
1
1
2

ρ
. (4.22b)
The mass fow rate can be written as


m V A
o o
ρ . (4.23a)
Substituting Equation 4.22b into Equation 4.23a, letting
β = d
o
/d
1
, and rearranging gives


m A
P
o

× ×

2
1
4
∆ ρ
β
. (4.23b)
Equation 4.23b represents the theoretical or ideal mass
fow rate through the orifce; that is, an incompressible,
inviscid fow. To determine the actual mass fow rate,
this equation needs to be multiplied by two correction
terms: gas expansion factor (Y
1
), and discharge coeff-
cient (C).
When a gas fows through an orifce plate it takes a
pressure drop causing the gas to expand and reduce the
density. To correct for the density change the equations
are multiplied by a gas expansion factor (Y
1
). The gas
expansion factor is based on laboratory tests and is
defned as follows:
Y
P
k P
1
4
1
1 0 41 0 35 − + ( ) . . , β

(4.24)
where k is the isentropic coeffcient.
The discharge coeffcient corrects the theoretical
equation for the effects of the approaching velocity
profle, the assumption of no energy loss between the
pressure taps and pressure tap location and is defned
as
C
true flo rate
theoretical flow rate
w
. (4.25)
The discharge coeffcient is determined from labora-
tory experiments and varies with tap arrangement and
diameter of pipe and orifce plate. For example, the
discharge coeffcient, C, for an orifce plate with corner
taps is
C C
b
R
D
n
+

(4.26)
where
C

 = 0.5959 + 0.0312β
2.1
−0.184β
8
b = 91.71β
2.5
n = 0.75
R
D
 ≡ Reynolds number based on the pipe inner diam-
eter and upstream velocity

R
V d
D

1
υ
.
For a complete listing of discharge coeffcient equations
see Miller [10]. The fnal form of the actual mass fow rate
through an orifce is the theoretical mass fow rate mul-
tiplied by the gas expansion factor and orifce discharge
coeffcient and is written as


m A Y C
P
o
× × ×
× ×
− ( )
1
1
4
2
1
∆ ρ
β
. (4.27)
Notice that the discharge coeffcient, C, is a function of
the upstream velocity in the pipe (V
1
). Since the upstream
velocity is not known, the solution to this equation
requires an iterative procedure.
For applications where the fuid temperature is very
high or low, it is important to include another correction
factor. Pipe and orifce plate bore diameters are typically
provided by the manufacturer and are based on room
temperature. As the temperature of the orifce plate and
pipe vary from room temperature, they expand and
contract. To correct for the expansion and contraction
of the material, a thermal-expansion factor is applied to
the dimensions; these new dimensions are then used in
the calculation. The thermal-expansion correction factor
for a temperature relative to 68°F is
F T
a PE P F
+

− ( ) − ( ) 1
2
1
68
4
4
β
α β α , (4.28)
where α
PE
and α
F
are the thermal-expansion factor of
the orifce plate and pipe, respectively, T
f
is the mate-
rial temperature in °F, and β is the diameter ratio of the
orifce bore to the inner diameter of the pipe. The ther-
mal expansion factor for several materials is given in
Table 4.1 [17].
Table 4.1
Thermal-Expansion Factors
Material (10
−6
in/in-°F)
Austenitic stainless steel 10.2 to 9
Martensitic stainless steel 6.5 to 5.5
Aluminum 16.3 to 14.4
Copper 9.8 to 7.7
© 2011 by Taylor and Francis Group, LLC
Fluid Flow 89
4.5.2 Venturi Meter
The orifce meter is a simple and accurate device for
measuring fow rates; however, the pressure drop for an
orifce meter can be quite large [4]. A meter that operates
on the same principle as the orifce meter, but with a
much smaller pressure drop, is the Venturi meter.
The Venturi meter is designed with gradual upstream
converging cone with an angle of 15° to 20° as illustrated
in Figure 4.15. Downstream of the converging cone the
fuid enters into a throat section. The fow rate is deter-
mined based on the pressure difference between the
upstream side of the converging cone and the throat sec-
tion. After leaving the throat, the fuid enters into a diverg-
ing section designed with an angle of 5° to 7°. Due to the
gradual upstream contraction there is no vena contracta
(like for the fow through an orifce plate) resulting in a
low pressure drop; again, this makes the Venturi meter
suitable where only small pressure heads are available.
The equation for the mass fow rate through a Venturi
meter is the same as for the orifce meter, Equation 4.27.
The discharge coeffcient for a Venturi meter is larger
and typically varies from 0.95 to 0.975; for the orifce
meter the discharge coeffcient typically varies from 0.6
to 0.7. For more information on the design and discharge
coeffcients of a Venturi meter see ASME MFC-3M-1989
[16].
4.5.3 Turbine Flow Meter
Turbine fow meters, when installed and calibrated
properly, are capable of providing the highest accura-
cies attainable by any currently available fow sensor for
both liquid and gas volumetric fow measurement [18].
The turbine meter consists of a cylindrical housing con-
taining a balanced, free-spinning rotor designed with
a series of curved blades that mounts on the centerline
of the pipe. The number of blades typically varies from
about 6 to 20 or more depending on the design with a
pitch angle of 30° to 45° [18]. As fuid passes over the
blades, it forces the rotor to spin.
The rotational speed of the rotor, or turbine, increases
linearly with fow velocity: within ± 0.5% over a wide fow
range from 10:1 to 20:1 [10]. The turbine speed is typically
measured by detecting the pulse of each blade electrically,
mechanically, or optically; each pulse represents a certain
volume of fuid. The fow rate of the fuid is determined
by integrating the total number of pulses over a period of
time. The accuracy of these meters is better than 1% over
a wide range of fow rates [4]. Accuracy is compromised,
however, with blade wear, bearing friction due to wear,
and when a vapor enters the line for liquid fows [10].
4.5.4 Vortex Flow Meter
When a fuid fows past a bluff body, the wake down-
stream will form rows of vortices that shed continuously
from each side of the body as illustated in Figure 4.16.
These repeating patterns of swirling vorticies are
referred to as Karman vortex streets: named after the
fuid dynamicist Theodore von Karman. Vortex shed-
ding is a common fow phenomenon that causes car
antennas to vibrate at certain wind speeds and also lead
to the collapse of the famous Tacoma Narrows Bridge
in 1940. Each time a vortex is shed from the bluff body
it creates a sideways force causing the body to vibrate.
The frequency of vibration is linearly proportional to
the velocity of the approching fuid stream and is inde-
pendent of the fuid density.
The frequency of vortex shedding or vibration of the
bluff body is determined as follows:
f
St V
d

×
average
body
, (4.29)
where St is the Strouhal number (dimensionless), V
average

is the average upstream velocity, and d
body
is diameter
of the bluff body. For fow over a cylindrical body, the
Strouhal number is constant for Reynolds numbers from
10,000 to 1,000,000: based on the diameter of the bluff
body. As the Reynolds number increases, the wake
becomes more complex and turbulent, but the alternate
Pressure sensing ports
Throat section
∆P
Figure 4.15
Typical Venturi meter. (Adapted from Roberson, J. A., and Crowe, T.,
Engineering Fluid Mechanics, p. 45, Dallas, TX: Houghton Miffin Co.,
1980.)
Karman vortex streets
Bluff body
Cycle
Figure 4.16
Pair of vortices shed downstream of a bluff body.
© 2011 by Taylor and Francis Group, LLC
90 Industrial Combustion Testing
shedding can still be detected at a Reynolds number
of 10,000,000 [19]. Knowing the Strouhal number, fre-
quency of vibration, and the diameter of the bluff body,
the average velocity can be determined.
Vortex fow meters are designed to partially obstruct
the fuid stream using a bluff body in order to create vor-
tex shedding. The frequency of the vortex shedding is
measured by various methods and converted to an aver-
age velocity. Today, most Vortex meters operate accu-
rately over a wide range of fows and are available for
line sizes from ½ inch to 16 inches. The main advantage
of this meter is that it has no moving parts, but it does cre-
ate pressure drop comparable to other obstruction-type
meters [10]: turbine meter, orifce plate, and so forth.
4.5.5 Magnetic Flow Meter
All the fow meters mentioned so far require that some
sort of obstruction, such as a turbine or orifce, be placed
within the fow stream. Magnetic fow meters, however,
do not obstruct the fow of the fuid. The magnetic fow
meter operates on the basic principle of magnetic induc-
tion; that is, when a conducting material moves through
a magnetic feld, it produces an electromagnetic force as
illustrated in Figure 4.17. It is interesting to note that the
basic principle of the magnetic fow meter was investi-
gated by Faraday in 1832; however, it was not practically
used until a century later [10].
When a current is applied to the coil, a magnetic feld
is produced that is at right angles to the pipe. As the con-
ductive liquid fows through the pipe, an electrical volt-
age is induced and measured by two electrodes mounted
fush to the pipe wall. This induced voltage is propor-
tional to the average velocity of fow and therefore to the
volume fow rate. In order for a magnetic fow meter to
work, the fuid must be conductive; therefore, they are
not suitable for gas fow [20].
It is important that the strength of the magnetic feld
remain constant; this is accomplished with a micropro-
cessor built into the transmitter controls. The micro-
processor also detects the voltage of the electrodes
and converters the signal to a fuid fow rate. The main
advantages of the magnetic fow meter are that the out-
put signal varies linearly with the fow rate and that the
meter does not obstruct the fow. The major disadvan-
tages are its relative high cost and that it is not suited for
gas fow [4].
4.5.6 ultrasonic Flow Meter
There are two types of ultrasonic fow meters: time-of-
fight and Doppler. The time-of-fight meter sends pulses
of high frequency sound waves diagonally across the
pipe as illustrated in Figure 4.18 [21]. The time required
for the sound wave to reach the opposite wall depends
on whether it is moving with the fow or against the
fow; if the wave is moving with the fow it will travel
faster than the wave moving against the fow. The time
difference is a measure of the fuid fow rate. The ultra
sonic fow meter illustrated in Figure 4.18 is referred
to as a single-path meter because only one beam path
crosses the pipe. Some ultrasonic fow meters send
sound waves along several paths to improve accuracy
as illustrated in Figure 4.19.
The Doppler fow meter sends out sound waves
similar to the time-of-fight meter, but the waves are
refected back to a detector by small particles or bubbles
moving with the fuid. The sound wave refected back
from the fuid will have a slightly lower frequency than
the transmitted sound wave due to the Doppler effect.
The difference between transmitted frequency and the
refected frequency is used to determine the fow rate.
Flow
Pulses of high
frequency sound
Figure 4.18
Time-of-fight ultrasonic fow meter: single-path type.
Magnetic field
Electrode
plates
Flow of
conductive fluid
Electrode voltage
Coil
V
V
Voltage applied
to produce
magnetic field
Figure 4.17
Principle of magnetic induction. (Adapted from “Precision Measuring
PD 340 Flow Transmitter,” http://www.instruments-gauges.co.uk/
pd340.htm)
© 2011 by Taylor and Francis Group, LLC
Fluid Flow 91
Commercially available Doppler fow meters requires
that the liquid contain at least 100 parts per million of 100
micron or larger suspended particles or bubbles [22].
4.5.7 Thermal Mass Meter
A typical thermal mass meter is shown in Figure 4.20.
The operation of these meters is based on the principle
that the rate of heat absorbed by a fowing fuid is directly
proportional to the mass fow rate [23]. In general there
are three types of thermal mass meters: constant tem-
perature, constant power, and energy balance.
Energy balance thermal mass meters require one heat-
ing element located between two temperature sensors as
illustrated in Figure 4.21 [24]. Although several design
variations exist, their operation is basically similar. As
the fuid fows past the heating element, it absorbs heat.
This heat is carried downstream where it is transferred
to the downstream temperature sensor. The tempera-
ture difference between the upstream and downstream
sensor is detected. This output signal is then converted
into a mass fow rate. These meters typically have a turn
down ratio of 10:1 while the constant temperature and
constant power meters have a turndown ratio of 1000:1
and 100:1, respectively.
4.5.8 Positive Displacement Meter
Positive displacement fow meters, or PD meters, are used
more than all other fow measurement devices [10]; a typ-
ical PD fow meter is shown in Figure 4.22. Millions of PD
Figure 4.20
Thermal mass meter.
Temperature sensors
Heating
element
Flow
Figure 4.21
Energy balance type thermal mass meter. (Adapted from “Flow
Measurement,” http://en.wikipedia.org/wiki/Flow_Measurement.)
Flow Flow
Flow
Figure 4.19
Time-of-fight ultrasonic fow meter: multipath type. (Adapted from Miller, R. W., Flow Measurement Engineering Handbook, New York: McGraw-
Hill, Inc., 1989.)
© 2011 by Taylor and Francis Group, LLC
92 Industrial Combustion Testing
meters are used daily to meter gasoline, natural gas, and
water into our homes. Although there are many different
PD meter designs commercially available, they all share
the same principle of operation: they measure the volume
of fuid by separating the fow into known volumes and
then counting, repeatedly, the flling and discharging of
volumes passing through the meter over time.
The pumping of individual parcels of fuid is accom-
plished using rotating parts, pistons, or diaphragms
that form moving seals between each other and/or the
fow meter body [25]. A few of these mechanical units
designed with rotating parts is illustrated in Figure 4.23.
As the fow rate of the fuid increases, these parts turn
or reciprocate proportionally faster and are sensed by
mechanical registers or electronic transmitters. The rate
of revolution or reciprocation determines the volume
of fuid passing through the meter. It is important that
these mechanical units have tight tolerances to prevent
fuid from passing through the fow meter without being
measured; this is referred to as slippage. Due to these
tight tolerances, flters are typically installed upstream
to prevent dirt from entering the meter; this helps reduce
mechanical wear and prevent plugging. Typically, flters
are used to remove particles larger then 100 µm as well
as gas bubbles from the liquid fow [26]. Mechanical wear
can result in substantial fow error; therefore, PD meters
should be periodically calibrated.
Since PD meters only measure the volume of fuid that
passes through, they are rarely used as fow rate meters;
however the average fow rate can be determined by
measuring the volume of fuid that passes through in
a given amount of time. It should also be mentioned
that PD meters create a considerable pressure drop that
should be considered for any application.
4.5.9 Pitot Tube
A Pitot tube is an instrument used to measure the
velocity of a fowing fuid. The Pitot tube was invented
by French engineer Henri Pitot in the early 1700s.
Today, it is widely used in industrial applications
and is manufactured in a variety of sizes as shown in
Figure 4.24.
Figure 4.22
Positive displacement meter.
Flow
Oval
gear
Oval gear
Flow
Lobe
Rotating lobe
Retractable
rotor blade
Flow
Rotating vane Rotating impeller
Impeller
Flow
Figure 4.23
Types of positive displacement meters. (Adapted from Universal Flow
Monitors (UFM), “Positive Displacement Flowmeters,” Hazel Park,
MI, http://www.fowmeters.com/ufm/index.cfm?task =  positive_
displacement)
12-inch ruler
Figure 4.24
Various sizes of Pitot-static tubes.
© 2011 by Taylor and Francis Group, LLC
Fluid Flow 93
The simple Pitot tube consists of a tube bent at a right
angle as illustrated in Figure 4.25. One end of the tube
is placed directly into a fowing fuid stream, parallel to
the fow direction or streamlines. As the fuid impacts
the open end of the tube it is brought to rest, with zero
velocity, directly in front of the opening of the Pitot
tube. The pressure at this point is the total or stagnation
pressure. By attaching a pressure sensing device on the
opposite end of the Pitot tube one can measure the total
pressure in the fowing stream directly. The velocity
pressure cannot be determined unless the static pressure
is known within the fow stream. Recall that the total
pressure is written as
Total Pressure = Static Pressure + Velocity Pressure.
The static pressure is generally measured using the
Pitot-static tube design as shown in Figures 4.26 and
4.27. Pitot-static designs consist of a coaxial tube placed
around the outside of the simple Pitot tube. Several
pressure ports located around the perimeter of the out-
side coaxial tube are used to measure the static pressure
within the fow stream; these holes are called the static
taps. By attaching a pressure sensing device and measur-
ing the differential pressure between the total and static
pressure allows one to determine the velocity pressure.
Knowing the velocity pressure, the velocity of the fuid
at the point of measurement can be determined. This is
illustrated in the following example.
Example: Pitot-static tube is placed within duct fow-
ing air with a temperature of 200°F. The static pressure
within the duct is 1 psig and the atmospheric pressure is
14.3 psia. A U-tube manometer measures a differential
pressure of 4 inches of WC between the total pressure
tap and static tap on the Pitot tube. Determine the veloc-
ity of the air at the measurement point.
First, determine the density of the air fowing within
the duct.
ρ
air,duct
3
density of air at sta
lb
ft
0 0765 .
m
nndard
temperature and pressure
(59 F and 14. ° 6696 psia)
psia psig
14.69
    
×
+ 14 3 1 .
66 psia
correcting for the actual air
pressure e inside the duct
F
    

×
+ 59 4460
200 460

F
correcting for the actual air
t
+
eemperature inside the duct
    
×
1 sslug
32.2 lb
converting pounds
mass to slugs
m
    
0 001944 . .
slugs
ft
3
Next, convert the differential pressure, dP, to units of
lb
f
/ft
2
.
dP
f
×
×
4
144
inches of WC
lb
inch
inch
ft
27.6
2
2
2
88 inches of WC
lb
ft
2
20 81 . .
f
Stagnation point
Streamlines
Velocity
Simple pitot tube
Total pressure
U-tube manometer
Figure 4.25
The simple Pitot tube.
Streamlines
Stagnation point
Velocity
Pitot-static tube
Velocity pressure
U-tube manometer
Static taps around
circumference
Figure 4.26
Illustration of the Pitot-static tube.
Total pressure tap
or impact hole Fittings for used mounting
Static pressure side
Total pressure side
Static tap
Figure 4.27
Photograph of the Pitot-static tube.
© 2011 by Taylor and Francis Group, LLC
94 Industrial Combustion Testing
Finally, the velocity can be calculated as follows:
V
dP
f

× |
.

`
}

× ×
2
2 20 81
1
2
ρ
.
lb
ft
1
slug-ft
s
1
2
2
llb
slugs
ft
f
2
0 001944
146
1
2
.
.
|
.

`
}

33
ft
s
.
The static reading on a Pitot tube is accurate to 0.5%
for Mach numbers up to 0.5. For Mach numbers between
the values of 0.5 to 0.7 the accuracy increases to 1.5%.
Above a Mach number of 0.7 the error can increase as
much as 10% due to the formation of shock waves on and
around the tip of the probe. Above a Mach number of 1.0
both the total and static readings vary signifcantly for
the actual values [27]. The Mach number in the above
example is calculated as follows:

Ma
V
c
c k R T

≡ × × ( ) speed of sound
1
2
(4.30)
where

k
R

ratio of specific heat 1.4 for air
univveral gas constant
mole weight
ft-lb

1545 32 .
ff
m
m
f
m
T
lb -mol-R
lb
lb mol
ft lb
lb R
m
29
53 3
20


.
00
1 4 53 3 660

F 460 660 R
ft lb
lb R
R
+
×


×
|
.
c
f
m
. .

× ×
`
}

32 2
1
1
2
. lb
1 slug
1
slug-ft
s
1 lb
2
m
f
2259 3
146 3
1259 3
0 116
.
.
.
. .
ft
s
ft
s
ft
s
Ma
Since the Mach number in this example is less than 0.7
we would not expect any error from formation of shock
waves around the probe.
Although Pitot-static tubes are not extremely sensitive
to their angle with respect to the fow streamline (angle
of attack), it is good practice to position them as parallel
as possible to the fow stream. Errors as high as 2% can
result if the angle of attack varies from 5° to 30°.
If the probe is positioned too close to a wall, the fuid
can accelerate between the probe and wall decreasing
the static pressure on one side. It is recommended that
the probe be positioned at least 10 probe diameters away
from the wall.
If the volume or mass fow rates are required in a
round or rectangular ducts then one must take mea-
surements at various locations across the duct. It is
important that the Pitot traverse be conducted at least
8 duct diameters downstream or 2 diameters upstream
of an obstruction such as fans, elbows, dampers, expan-
sions, and so on. Figure 4.28 shows the locations for a
Pitot tube traverse in a round or rectangular duct, based
on centroids of equal area, for determining the volume
or mass fow rate [28].
After measuring the velocity at each point in the duct
using the locations provided in Figure 4.28, the total vol-
ume and mass fow rate of the fuid can be determined
as follows:

Q A
n
V
m Q
i
i
n

×


duct
1
1

ρ
(4.31)
where
Q ≡ volume fow rate in the duct
A
duct
 ≡ duct cross-sectional area
n ≡ total number of points traversed
V
i
 ≡ velocity at each measurement point
m˙  ≡ mass fow rate in the duct.
4.5.10 annubar
The annubar is widely used as a fow measurement
device in industry. In general, an annubar consists of
a hollow tube or strut, with various cross-sectional
designs, which spans across a fow stream as illustrated
in Figure 4.29. The strut is designed with several holes
strategically placed along its length so that the average
of the total and static pressure is measured. The differ-
ence in the average, total, and static pressure provides
the mean velocity across the pipe; the volume fow rate
of the fuid is determined by multiplying the mean
velocity by the cross-sectional area of the pipe. A single
Pitot tube measurement will give a similar reading if it
is located at a point in the pipe cross section where the
fowing velocity is near the average velocity [29].
The strut can consist of various designs such as a
cylinder, diamond, or triangular cross section. The
© 2011 by Taylor and Francis Group, LLC
Fluid Flow 95
position of the pressure ports is critical to the accu-
racy of the device. If the ports are not located in the
proper location then the average velocity profle will
be in error. It is also important that the velocity profle
is symmetric; if skewed to one side signifcant errors
can result. Typically, as the pipe or duct size increases,
the number of total pressure ports and static ports
increases.
Annubars are commonly used to measure the fow
rate in ducts and large pipes. The advantages of the
annubar is that they are easy to install, relatively inex-
pensive, and do not take a large pressure drop [10].
The disadvantage is their poor performance with dirty
fuids.
References
1. Betts, D. “Pressure Gauge. How Products Are Made.”
Gale Research Inc. Encyclopedia.com November 20, 2009,
http://www.encyclopedia.com/doc/1G2-2896500079.
html
2. Zigrang, D. J. “Elements of Engineering Measurements.”
Department of Mechanical Engineering, University of
Tulsa, Oklahoma, 1985.
3. Hydraulic Energy Transmission and Control. Hill
Learning Systems. 2001–2007. http://www.hetacfuid-
power.com/screen_shot_gallery.htm
4. Roberson, J. A., and Crowe, T. Engineering Fluid Mechanics,
p. 45. Dallas, TX: Houghton Miffin Co., 1980.
5. Betts, D. “Pressure Gauge. How Products Are Made.”
Gale Research Inc. Encyclopedia.com November 20, 2009,
http://www.encyclopedia.com/doc/1G2-2896500079.
html
6. Cornforth, J. R. Combustion Engineering and Gas Utilization.
London: British Gas plc, 1992.
7. Betts, D. “Pressure Gauge. How Products Are Made.”
Gale Research Inc. Encyclopedia.com November 20, 2009,
http://www.encyclopedia.com/doc/1G2-2896500079.
html
8. Jankura, R. Seven Steps to Pressure Gage Selection (Tech
Check). Dresser Instrument: Encyclopedia.com July 1, 2002,
http://www.encyclopedia.com/doc/1G1-90331703.html
9. O’Keefe Controls Co., Trumbull, CT, 2003, http://www.
okcc.com/PDF/Pressure%20Snubber%20pg.42.pdf
Distance from centerline (CL), x/W or y/H # Point/row
Round ducts
3
4
5
6
4
5
6
7
±0.25
±0.083
±0.375
±0.20
±0.25
±0.143 ±0.286
±0.417
±0.429
Distance from CL, x/W or y/H
±0.40 0
0
0.204
0.177
0.158
0.144
0.353
0.306
0.274
0.25
0.457
0.395
0.354
0.323
0.468
0.418
0.382
0.474
0.433 0.479
CL
x
y
H
W
CL
Rectangular ducts
# Points/row
CL
y
CL
x
Example of 6 points/row
Example of 4 points/row
Figure 4.28
Pitot tube traverse locations for round and rectangular ducts based on centroids of equal areas. (Adapted from Flow Kinetics LLC, “Using a Pitot
Static Tube for Velocity and Flow Rate Measurement.,” http://www.fowmeterdirectory.com/fowmeter_artc/fowmeter_artc_02111201.html)
Static
pressure
Total
pressure
Flow
Flow
Various strut designs
Figure 4.29
Annubar.
© 2011 by Taylor and Francis Group, LLC
96 Industrial Combustion Testing
10. Miller, R. W. Flow Measurement Engineering Handbook.
New York: McGraw-Hill, Inc., 1989.
11. Deadweight Testers. “Improving Deadweight Tester
Accuracy.” www.transcat.com 2007, http://www.tran-
scat.com/PDF/DeadweightTesters.pdf
12. Ferron, A. G. “Construction of Static Pressure Taps.”
Technical Memo. Alden, MA: Alden Research Lab,
September 1986.
13. Brunkalla, R. L. “Effects of Fabrication Technique on the
Discharge Coeffcient for Throat Tap Nozzles,” ASME
Paper 84-JPGC-PTC-10, October 1985.
14. Buckley, B. “Fundamentals of Orifce Metering.”
Houston, TX: Daniel Measurement and Control. http://
help.intellisitesuite.com/ASGMT%20White%20Papers/
papers/002.pdf
15. Smith Metering, Inc., “Fundamentals of Orifce Metering,”
http://www.afms.org/Docs/gas/Fundamenatls_of_
Orifce.pdf
16. “Measurement of Fluid Flow in Pipes Using Orifce,
Nozzle, and Venturi,” ASME MFC-3M-1989, The
American Society of Mechanical Engineers, 345 East 47th
Street, New York.
17. “Tool Design,” http://www.wisetool.com/designation/
te.htm
18. Wadlow, D. “Turbine and Vane Flowmeters.” in The
Measurement, Instrumentation and Sensors Handbook,
edited by J. G. Webster. Boca Raton, FL: CRC Press, 1998.
http://www.sensors-research.com/articles/turbines.
htm
19. White, F. M. Viscous Fluid Flow, p. 10. St. Louis, MO:
McGraw-Hill, 1991.
20. “Precision Measuring PD 340 Flow Transmitter,” http://
www.instruments-gauges.co.uk/pd340.htm
21. “Why and How to Measure Flare Gas, Roxar,” http://
www.fowmeterdirectory.com/fowmeter_artc/fowme-
ter_artc_02021401.html
22. “Introduction to Ultrasonic Doppler Flow Meters,”
http://www.omega.com/prodinfo/ultrasonicfowmeters.
html
23. “A Flow Measurement Primer: Thermal Mass Flow Meters,
Cross Instrumentation,” http://www.crossinstrumenta-
tion.com/ga/mfg/Flow/tmf_meters.htm
24. “Flow Measurement,” http://en.wikipedia.org/wiki/
Flow_measurement
25. Universal Flow Monitors (UFM). “Positive Displacement
Flowmeters.” Hazel Park, MI. http://www.fowmeters.
com/ufm/index.cfm?task = positive_displacement
26. eFunda. “Positive Displacement Flowmeters.” 2009.
http://www.efunda.com/designstandards/sensors/
fowmeters/fowmeter_pd.cfm
27. United Sensor. “Pitot-Static Pressure Probes: For Measuring
Total and Static Pressures of a Moving Fluid,” Bulletin 1,
6–83 Amherst, NH. www.unitedsensorcorp.com
28. Flow Kinetics LLC. “Using a Pitot Static Tube for Velocity
and Flow Rate Measurement.” http://www.fowmeterdi-
rectory.com/fowmeter_artc/fowmeter_artc_02111201.
html
29. Annubar. 2009. http://en.wikipedia.org/wiki/Annubar
© 2011 by Taylor and Francis Group, LLC
97
5.1 Introduction
Temperature is a particularly important variable in
industrial combustion applications because it directly or
indirectly affects a number of other important variables.
The product temperature is often a critical parameter
in most processes. While there is usually a minimum
temperature that must be reached for adequate pro-
cessing, there may also be a maximum temperature
above which product quality is reduced. Higher than
necessary product temperatures not only increase fuel
costs, but they may also increase cooling costs after
the product exits the combustion process. Temperature
affects the heat transfer in a furnace [1]. Thermal NOx
emissions are exponentially dependent on fame tem-
peratures [2]. Combustion chemistry is very compli-
cated and dependent on temperature. High exhaust
gas temperatures mean reduced thermal effciency [3].
Low combustion chamber temperatures may be safety
issues for some technologies like fameless combustion
(see Chapter 23). Low combustion chamber tempera-
tures during startup can signifcantly impact pollutant
emissions [4]. Excessively high combustion chamber
temperatures could cause damage to the refractory lin-
ing and to equipment inside the chamber such as process
tubes. Figure 5.1 shows impinging fames on the convec-
tion tubes in a process heater. Prolonged impingement
overheats the tubes and may eventually damage these
tubes, possibly leading to a rupture. Figure 5.2 shows a
composite image consisting of multiple infrared scans
of the process tubes in a heater. Many plants with pro-
cess heaters take regular scans to ensure there are no
hot spots that could lead to damage and premature shut-
down. Bussman and Baukal have shown that even the
ambient air temperature can have a signifcant impact
on both thermal effciency [5] and pollution emissions
in industrial combustion processes [6].
5
Temperature
Charles E. Baukal, Jr.
CONTENTS
5.1 Introduction ..................................................................................................................................................................... 97
5.2 Temperature Measurement Examples ......................................................................................................................... 98
5.3 Thermocouples ................................................................................................................................................................ 98
5.4 Gas Temperature ........................................................................................................................................................... 101
5.4.1 Suction Pyrometer ............................................................................................................................................ 101
5.4.2 Optical Techniques ........................................................................................................................................... 102
5.4.2.1 Light Scattering .................................................................................................................................. 102
5.4.2.2 Emission .............................................................................................................................................. 102
5.4.3 Fine Wire Thermocouples ................................................................................................................................ 103
5.4.3.1 Coatings ............................................................................................................................................... 103
5.4.3.2 Corrections .......................................................................................................................................... 104
5.4.3.3 Example Uses ...................................................................................................................................... 107
5.4.4 Line Reversal ..................................................................................................................................................... 107
5.4.5 Acoustic Pyrometry .......................................................................................................................................... 108
5.5 Surface Temperature .................................................................................................................................................... 108
5.5.1 Embedded Thermocouple ............................................................................................................................... 108
5.5.2 Infrared Detectors ............................................................................................................................................. 109
5.5.3 Other Techniques ...............................................................................................................................................111
5.6 Future ..............................................................................................................................................................................111
References ................................................................................................................................................................................ 112
© 2011 by Taylor and Francis Group, LLC
98 Industrial Combustion Testing
5.2 Temperature Measurement Examples
Some examples of temperature measurements for indus-
trial combustion processes will serve to illustrate the
various methods that have been used that are described
in some detail in this chapter. Hayes et al. [7] describe
glass surface temperatures made in a production regen-
erative fat glass furnace (see Chapter 32). The optical
measurements were made with a water-cooled, two-
color pyrometer inserted through holes in the crown
(roof) at six locations along the length of the furnace.
Both average and time-resolved glass surface tempera-
tures were measured during the transient reversal cycles
in the furnace. Average temperatures ranged from about
1700 K to a peak of about 1900 K depending on the time
in the cycle and the location in the furnace. These mea-
surements are discussed in more detail in Chapter 32 of
this book.
Honner, Vesely, and Svantner [8] made tempera-
ture measurements in a continuous pusher-type metal
reheating furnace using thermocouples imbedded in a
billet used specifcally as a “sensor”. These were par-
ticularly challenging measurements, not only because
of the harsh high temperature environment inside a
reheat furnace, but also because the sensor billet was
moving, which means the electrical wiring and battery-
powered data acquisition system also moved through
the furnace.
Rafdi and Blasiak [9] measured local temperature in a
test furnace using four high temperature air combustion
(HiTAC) regenerative burners (see Chapter 21). A suc-
tion pyrometer was used to measure gas temperature
and a thin-wire thermocouple (80 µm diameter) with a
short response time (time constant = 25 ms) was used
to measure temperature fuctuations. Fast response was
particularly important in this application because of
the dynamic nature of regenerative burners that cycle
between fring and exhausting. An important objective
of this study was to determine spatial temperature uni-
formity and comparing it to conventional nonregenera-
tive burners.
5.3 Thermocouples
Many different types of devices are used to measure
temperature, depending on the temperature range and
specifc application. The most common device is the
thermocouple. There are many references on using ther-
mocouples in combustion applications. Some are briefy
discussed here. Eckert and Goldstein [10], Goldstein
and Chiang [11], Holman [12], and Deshmukh [13] are
Flames into
roof tubes
Figure 5.1
(See color insert following page 424.) Flame impingement on the
process tubes in the convection section of a process heater.
Outlets
ermal
probe
Tubeskin Tubeskin
View
port
Figure 5.2
(See color insert following page 424.) Example of an infrared scan of
process tubes in a heater.
© 2011 by Taylor and Francis Group, LLC
Temperature 99
some examples of references that discuss general tem-
perature measurement techniques. Bowman [14] dis-
cusses the use of various types of temperature probes
in fames. Cornforth [15] devotes a chapter to instru-
mentation and measurement, including temperature,
in combustion applications. The report by Okoh and
Brown [16] reviews and compares many techniques for
the measurement of temperature in combustion sys-
tems. The report also includes equipment specifcations
and suppliers.
Thermocouples are the primary device used to make
temperature measurements in industrial combustion
applications. A variety of other techniques have been
used to measure both gas and surface temperatures
that are briefy discussed later in this chapter. Due to
the importance of thermocouples, they are considered
in some detail in this section.
Temperature measurement using thermocouples is
based on the thermoelectric effect. Two dissimilar metals
are joined together at a junction where an electromotive
force (emf) is generated according to the Seebeck effect.
The emf level depends on the junction temperature. The
Peltier effect causes an emf to be generated when the
dissimilar metals are connected to an electrical circuit.
A third emf is produced if there is a temperature gradi-
ent along either of the materials. The emf generated by
the Peltier effect at the junction can be used to measure
temperature, knowing the thermoelectric properties of
the metals.
There are many different types of thermocouples,
depending on the temperature range needed. Table 5.1
shows the common thermocouple types and the alloys
used in each. Each combination generates a specifc
millivolt output as a function of temperature. These
millivolt reference tables are based on some reference
junction temperature which is usually 0°C. Most elec-
tronic data acquisition systems have these tables built
into the software, usually through a curve ft of the
data, so the millivolt readings are automatically con-
verted to temperature. For best accuracy, a thermocou-
ple with the appropriate range should be used for the
expected temperatures. For example, type T thermo-
couples are typically used for lower temperature mea-
surements such as the water temperature in a cooling
system. In industrial combustion, Type J and K ther-
mocouples are commonly used to measure interme-
diate temperatures, such as the temperature inside a
refractory wall or temperatures inside a burner. Type
Table 5.1
Application Characteristics of Some Common Thermocouple Alloys
Max T
°F
Max T
°C
Allowable
Atmos. (Hot) Material Names
ANSI
Type
a
Color
Code
Output
mV/100°F
Accuracy, %
Standard
a
Precision
a
5072 2800 Inert, H
2
,
vacuum
Tungsten/tungsten 26%
rhenium
— — 0.86 — —
5000 2760 Inert, H
2
,
vacuum
Tungsten 5% rhenium/tungsten
26% rhenium
— — 0.76 — —
4000 2210 Inert, H
2
Tungsten 3% rhenium/tungsten
35% rhenium
— — 0.74 — —
3720 1800 Oxidizing
b
Platinum 30% rhodium/
platinum 6% rhodium
B — 0.43 1/2 1/4
2900 1600 Oxidizing
b
Platinum 13% rhodium/
platinum
R — 0.64 1/4 1/4
2800 1540 Oxidizing
b
Platinum 10% rhodium/
platinum
S — 0.57 1/4 1/4
2372 1300 Oxidizing
b,c
Platinel II (5355)/Platinel II
(7674)
— — 2.20 5/8 —
2300 1260 Oxidizing Chromel/Alumel,
d
Tophel/
Nial,
e
Advance T1/T2,
f
K Yellow red 2.20 4°F, or 3/4% 2°F, or 3/8%
Thermo-Kanathal P/N
g
1800 980 Reducing
a
Chromel/constantan E Purple red 4.20 1/2 3/8
1600 875 Reducing Iron/constantan J White red 3.00 4°F, or 3/4% 2°F, or 3/8%
750 400 Reducing Copper/constantan T Blue red 2.50 3/4 3/8
a
Per ANSI C96.1 Standard.
b
Avoid contact with carbon, hydrogen, metallic vapors, silica, reducing atmosphere.
c
@ Engelhard Corp.
d
@ Hoskins Mfg. Co.
e
Wilber B. Driver Co.
f
Driver-Harris Co.
g
The Kanthal Corp.
© 2011 by Taylor and Francis Group, LLC
100 Industrial Combustion Testing
R, S, and B (platinum–rhodium) thermocouples are
commonly used to measure higher temperatures, such
as hot face temperatures and gas temperatures inside
a furnace. Type L thermocouples can be used for even
higher temperatures, although they are not as readily
available as the platinum– rhodium thermocouples.
Another important factor to consider when selecting a
thermocouple is the environment where it will be used.
Some thermocouples can handle oxidizing and reducing
conditions, while others cannot. For example, Type R, S,
and B thermocouples can handle the high temperature
oxidizing conditions that are often present in industrial
combustion processes. While a Type C thermocouple
(tungsten–rhenium) can measure higher temperatures
than platinum–rhodium thermocouples, Type C ther-
mocouples cannot be used in oxidizing environments—
which severely limits their use in industrial combustion
applications.
Thermocouple installation is important to ensure
the proper temperature will be measured. Figure 5.3
shows a thermocouple located at the top of the radiant
section in a process heater used to measure the tem-
perature of the heater. As will be discussed later, this
type of thermocouple measurement needs to be cor-
rected to get the actual temperature. The location of
this thermocouple is important because if it is located
too close to the wall, then the temperature will be lower
due to the lower temperature tubes that are cooled
by process fuid. That lower temperature would not
be representative of the average heater temperature.
Figure 5.4 shows a photo of the thermocouples used to
measure the water outlet temperatures from calorim-
eters in a fame impingement heating study [17]. The
thermocouples were positioned so that water would
have to fow over the junctions, regardless of the fow
rate. If the thermocouples were positioned, for exam-
ple, perpendicular to a vertically downward fow of
water, there is a good chance the junction would not
always contact the water fow, particularly at low fow
rates. This would lead to measurement error in deter-
mining the actual water temperature.
As with any instrument, calibration is important for
thermocouples. Although manufacturers supply accu-
racy data for typical thermocouples, calibration can be
used to check that these data apply to the specifc instru-
ment being used in an experiment. The accuracy can
be improved by calibrating the specifc thermocouple
being used. More importantly, a thermocouple calibra-
tor (see Figure 5.5) can be used to check for any wiring
problems or any setup mistakes in the data acquisition
system. A thermocouple that has been wired and con-
nected to the data acquisition system can be inserted
into the calibrator to check for problems. For example,
if a Type R thermocouple is actually being used but it
Flue gas
temperature
thermocouple
O
2
probe
Figure 5.3
Example of a thermocouple at the top of the radiant section in a pro-
cess heater. (Courtesy of John Zink Co. LLC, Tulsa, OK.)
Figure 5.4
Example of measuring the outlet temperature from calorimeters
used to measure the heat fux in fame impingement heating. (From
Baukal, C. E., Heat Transfer from Flame Impingement Normal to a Plane
Surface, Saarbrücken, Germany: VDM Verlag, 2009.)
Figure 5.5
Portable thermocouple calibrator set to 150°F with a thermocouple
inserted into it.
© 2011 by Taylor and Francis Group, LLC
Temperature 101
was set up as a Type K thermocouple in the data acquisi-
tion system, then systematic error will result as all read-
ings will at least be in error by the difference between
the calibration curves for the two different types of
thermocouples.
5.4 Gas Temperature
Many techniques have been used for measuring the
spatial gas temperature distribution. They may be
generally classifed as contact (thermocouples) and
optical, such as Rayleigh and Raman scattering [18–21].
At this time, the light-scattering methods are very
expensive and diffcult to use [22]. Their primary use
is in turbulent fows, and where probe survivability is
a problem. Thermocouples are preferable because of
their well-established properties, low cost, and ease
of use. Some optical techniques, such as line reversal,
may be diffcult to use because of the need to seed the
fame with particles. Those seeds could plug burners
with small passage sizes.
5.4.1 Suction Pyrometer
A suction pyrometer is designed to measure closer to
the true gas temperature than a bare thermocouple
measures in a hot environment such as a combustion
chamber. There are several types of heat transfer to and
from the junction of a thermocouple that cause it to read
lower than the actual temperature. This is discussed
in more detail later, including how to calculate a cor-
rection for a bare thermocouple. A properly designed
and operated suction pyrometer needs very little if any
correction.
At the desired measurement location, gases are
extracted from the fame through a sampling tube in
the suction pyrometer. A thermocouple is positioned
just inside this tube, typically made of a ceramic or
high temperature metal, which acts as a radiation
shield. In some designs, multiple shields are used (see
Figure 5.6). Examples of suction pyrometers are shown
in Figure 5.7.
This technique is particularly useful when measuring
gas temperatures in hot-surfaced enclosures. Surface
radiation from the walls to an unshielded thermocouple
Thermocouple wires Thermocouple junction
Hot
gases
Outer tube
Middle tube
Center tube
Hot gas
eductor
Air or
steam
Figure 5.6
Drawing of a multishield suction pyrometer.
(a) (b) (c)
Figure 5.7
Examples of suction pyrometers: (a) new, (b) in the sidewall of a furnace, and (c) in the roof of a furnace. (From Baukal, C. E., Heat Transfer in
Industrial Combustion, Boca Raton, FL: CRC Press, 2000.)
© 2011 by Taylor and Francis Group, LLC
102 Industrial Combustion Testing
may introduce large errors in the gas temperature mea-
surement. Chedaille and Braud [23] and Goldman [24]
specifcally discuss the use of a suction pyrometer to
measure gas temperatures in combustors. This device
has been successfully used to measure gas tempera-
tures in excess of 2200 K (3500°F). In fame impinge-
ment studies, Beér and Chigier [25], Vizioz and Lowes
[26], Smith and Lowes [27], and Rajani, Payne, and
Michelfelder [28] measured gas temperatures with a
suction pyrometer.
Grey [29] describes a technique related to suc-
tion pyrometry known as a twin sonic orifce probe.
In this technique, the gases are extracted at a very
high velocity, through two converging–diverging
nozzles inside the probe. One limitation of the suc-
tion pyrometer is that active chemical reactions may
also occur inside the probe passages for highly disso-
ciated gases. With proper corrections, this effect may
be overcome by the dual sonic orifce probe. Rajani
[28] and Ivernel and Vernotte [30] used a twin sonic
orifce probe to measure gas temperatures in oxy-
gen/fuel fames up to 3060 K and 2800 K (5050°F and
4600°F), respectively.
5.4.2 Optical Techniques
A variety of optical techniques have been used to mea-
sure gas temperatures in combustion applications,
particularly in fames. There are potentially some
important advantages of optical techniques compared
to contact techniques such as suction pyrometers (see
Figure 5.7). Optical measurement techniques do not
disturb the fow, where thermocouples may have a
signifcant impact on the fuid dynamics. Optical tech-
niques can potentially measure higher temperatures as
there are not the materials issues compared to thermo-
couples. For some optical techniques, temperature pro-
fles can be measured at one point in time without the
need to make multiple individual measurements over
some length of time. Optical techniques often have a
much faster response time compared to contact meth-
ods. This is particularly important in turbulent and
transient fows.
The techniques are categorized here as light scatter-
ing (where a light source is needed) and emission (no
light source needed).
5.4.2.1 Light Scattering
Laurendeau [31] gives a brief review of temperature
measurements using light-scattering techniques. The
techniques discussed included spontaneous vibrational
and rotational Raman scattering, Rayleigh scattering
and laser-induced fuorescence. Fluorescence methods
were recommended for temporally resolved planar
thermometry. See Chapter 12 for a discussion of some
of these techniques.
Kondic [32] developed a method to measure the den-
sity and temperature felds in fuids along a light path
where the light source and detector were located exter-
nal to the system. Although a laser is the preferred light
source, a conventional light source can also be used.
The gas temperature is related to the density feld and
the index of refraction of the fuid. Posillico [33] made
detailed spatial gas temperature measurements, using
Raman spectroscopy [20], for air/CH
4
fames imping-
ing on a water-cooled fat plate. The highest reported
gas temperature was 3200 K (5300°F), even though
the adiabatic fame temperature for a stoichiometric
air/CH
4
fame is 2223 K (3542°F). No explanation was
given for this apparent discrepancy. Hughes, Lacelle,
and Parameswaran (1995) compared a suction pyrom-
eter and CARS (coherent anti-Stokes Raman spectros-
copy) in an industrial-scale fame [34] (See Chapter 13).
Hofmann and Leipertz [35] presented experimental
measurements of the temperature feld in a sooting,
methane/air fame using a new technique referred to
as fltered Rayleigh scattering (FRS) [35]. Rumminger
et al. [36] have used laser-induced fuorescence (LIF) to
measure gas temperatures at the exit of porous radiant
burners. The measurements were slightly higher than
those made with a thermocouple after all corrections
had been made to both techniques. The overall uncer-
tainty in the LIF measurements was estimated to be ±
110°F (60 K).
5.4.2.2 Emission
Chao and Goulard [37] presented a nonlinear itera-
tive inversion technique for calculating fame tem-
peratures from a single-line-of-sight multi-frequency
set of radiance measurements. Hommert, Viskanta,
and Mellor [38] measured fame temperatures using
spectral remote sensing (SRS). A detector measures
the line-of-sight spectral radiation coming from hot
combustion products. Some advantages include: no
external radiation source needed, noncontact, remote,
and relatively simple equipment. Some disadvan-
tages include: strong infrared emitter needed, aver-
age rather than point measurement, single data point
rather than a profle, and accurate model needed to
convert the data. The technique was tested on a fat
air/CH
4
diffusion fame, which had relatively high
concentration and temperature gradients. The 4.3 µm
band for CO
2
was selected because it is strong with
relatively little overlap with other species. An inver-
sion algorithm was used to convert the spectral radia-
tion measurements to temperatures. Temperature
© 2011 by Taylor and Francis Group, LLC
Temperature 103
measurements were made with a Type S thermocou-
ple, which varied from about 600 to 1800 K (600 to
2800°F) to compare against the SRS measurements.
CO
2
was added to some of the fames to vary the con-
centration. Kim and Song used SRS to make gas tem-
perature measurements in furnaces [39–41]. In the frst
study [39], they investigated SRS in a 2 m (6.6 ft) long
by 0.12 m (4.7 in.) inner diameter stainless steel tube
where kerosene was the fuel and where the measure-
ment was made along the length of the tube. The SRS
measurements ranged from about 900 to 1400 K (1200
to 2100°F) and were within 4% of thermocouple read-
ings. In the second study [40], a longer 3.4 m (11 ft)
by 0.12 m (4.7 in.) inner diameter stainless steel tube,
kerosene was again the fuel, and temperatures were
similar to the frst study. In the third study [41], mea-
surements were made in three furnaces: 0.8 m long,
2.0 m long and 3.4 m long.
Reuss [42] used holographic interferometry in fuel
lean propane–air fames and compared the measure-
ments with those made with a thermocouple [42]. The
interferometry readings were within 50 K of those made
with the thermocouple. This technique was developed
to measure gas temperatures inside engine cylinders.
Char and Yeh [43] describe an optical technique of
measuring the fame temperature using a combination
of measuring the infrared radiation at a selected wave-
length (4.3 µm, which has the highest intensity for CO
2

and H
2
O) and making some iterative calculations to
determine the fame emissivity. The measurements were
made on a fuel lean (equivalent ratio = ϕ = 0.51) open
propane fame. The results compared favorably with
thermocouple measurements. A correlation was devel-
oped for the gas emissivity for this fame as a function
of the gas temperature position from the burner outlet.
Correia, Ferrao, and Caldeira-Pires [44] discuss three-
dimensional emission tomography for use inside com-
bustors with sooty fames. Previous applications of
this technique did not account for radiation absorption
(extinction) inside the fame. Correia et al. developed a
reconstruction algorithm to account for that effect. The
basic advantage of emission tomography compared to
transmission tomography is that no background radia-
tion source is needed, which makes it more amenable
to industrial combustion environments. This technique
was applied to both axisymmetric and nonaxisymmetric
semi-industrial fames with uncertainties below ±10%
where soot was available in the fame. Temperatures
in the range of approximately 700 to 1200°C (1300°F–
2200°F) were reported and compared favorably with
thermocouple measurements in laboratory propane
fames. Temperatures of approximately 1000 to 1700°C
(1800°F–3100°F) were measured in a large pilot-scale,
oil-fred test furnace.
Zhou et al. [45] used a fame image processing tech-
nique to measure the gas temperatures in a 200 MW
pulverized-coal corner-fred boiler. Eight fame image
CCD detectors were used to capture monochromatic
radiation intensity images under the visible wave-
lengths of red, green, and blue. A blackbody furnace
was used to calibrate the color images. A Monte Carlo
method was used to calculate the radiation intensity to
establish the relationship between the images and the
temperature in the furnace. Three dimensional profles
were constructed from the measurements. The com-
puted temperatures were within 5% of suction pyrom-
eter measurements.
5.4.3 Fine Wire Thermocouples
Fine wire thermocouples have a small diameter junction
that minimizes both response time and the heat transfer
effects that must be corrected for to get the actual tem-
perature that is discussed later. In some cases, the junc-
tion may be coated to minimize chemical reactions at
the junction that could be catalyzed by the bare metals
and which could impact the measurements. Coatings
are discussed frst in this section. Measurements with
fne wire thermocouples typically need to be corrected.
Several methods for making these calculations are dis-
cussed in the second part of this section. Fine wire ther-
mocouples have been used in numerous combustion
applications. Some examples are given at the end of this
section.
5.4.3.1 Coatings
Heitor and Moreira [22] provided an extensive review
of using thermocouples in combustion environments.
They stated that fow disturbances and measurement
errors can be made negligible in laminar fows. They
suggested a coating consisting of 90% Al
2
O
3
and 10%
MgO, which minimizes catalytic effects and is not toxic.
According to Cookson, Dunham, and Kilham (1964) [46],
thermocouple junctions should be coated. The junctions
should also be as small as possible, to prevent surface
recombination effects. Kent [47] lists the following gen-
eral requirements for a proper thermocouple coating:
noncatalytic to the fame •
inert to the thermocouple material •
impermeable to gases to protect the thermo- •
couple
poor electrical conductor to prevent electrical •
shorts
stable within the desired temperature range •
capable of being evenly applied •
© 2011 by Taylor and Francis Group, LLC
104 Industrial Combustion Testing
A coating made from a mixture of approximately 10%
BeO/90%Y
2
O
3
was recommended.
Most high temperature applications require some
type of protective coating on the thermocouple,
unless the atmosphere is inert. Kaskan [48] described
a procedure for applying a silica coating to thermo-
couples. The junction was immersed in a fame con-
taining a small amount of hexamethyldisiloxane. This
produced a uniform coating of fused quartz. It was
found that the coating failed at about 1900 K (3000°F),
due to softening of the silica. Cookson and Kilham
[49], Kilham and Dunham [50], and Milson and
Chigier [51] used silica coatings to prevent catalyic
hydrogen atom recombination on the platinum in the
thermocouple [64]. Madson and Theby [52] showed
that uncoated platinum alloy thermocouples signif-
cantly overpredicted temperature, compared to silica-
coated thermocouples in an O
2
/Ar/CH
4
fame. Pollock
[53] has shown that silica coatings may contaminate
platinum-alloy thermocouple wires, in chemically
reducing environments. Kremer, Buhr, and Haupt
[61] used Al
2
O
3
to prevent radical recombination.
Hoogendoorn, Popiel, and van der Meer [54], Popiel,
van der Meer, and Hoogendoorn [55], and Van der
Meer [56] used an yttrium–beryllium oxide (Yt
2
O
3
/
BeO), as recommended by Kent [47]. This eliminates
silica contamination. Unfortunately, beryllia is toxic
and extremely poisonous.
5.4.3.2 Corrections
When using thermocouples to measure high tempera-
tures, the measurements must be corrected for the
errors due to radiation, convection, and wire conduc-
tion. Sato et al. [57] present a calculation procedure for
making these corrections. Moffat [58] notes the general
equations for estimating these types of errors:

E
v
g Jc
c p
conv
= − ( ) 1
2
2
α ,
(5.1)

E
h
T T
rad probe
4
surr
4
= − ( )
σε
,
(5.2)

E
t t
l
hA
kA
c
cond
gas probe mount
cond
=







cosh
,,
(5.3)
where
E
conv
 = convection error
E
rad
 = radiation error
E
cond
 = conduction error
α = recovery factor
v = gas velocity
g
c
 = universal gravitational constant
J = Joules constant
c
p
 = gas specifc heat
σ = Stefan-Boltzmann constant
ε = probe emissivity
h =  convection coeffcient for gas fowing over probe
T
probe
 = absolute temperature of the probe
T
surr
 = absolute temperature of the surroundings
t
gas
 =  temperature of the gas fowing past the probe
(usually what is being measured)
t
probe mount
 = temperature of the probe mount
l = length of the exposed junction
A
c
 = area for convection heat transfer
k = probe thermal conductivity
A
cond
 = area for conduction heat transfer.
Note that these are iterative calculations as the temper-
ature of the gas is the quantity being measured. Also,
some of the quantities may not be known with a high
degree of precision. Some of these include the convec-
tion coeffcient, the gas velocity, the probe emissiv-
ity, the mount temperature, and the exposed junction
length. Therefore, there will still be some uncertainty
even after the measurements have been corrected for
these errors. In general, the errors increase as the probe
size increases.
Several methods are commonly used to compensate
for these errors. One is electrical compensation [59].
However, this method cannot be used at very high tem-
peratures, where the thermocouple may already be near
its melting point. Any additional heating caused by the
electrical heating may cause the thermocouple to fail.
As examples, Buhr, Haupt, and Kremer [60] and Kremer,
Buhr, and Haupt [61] used electrical compensation for
radiation and conduction heat losses.
Another compensation method is to make measure-
ments using diminishingly smaller diameter ther-
mocouples [62]. The thermocouple measurements are
plotted against the cross-sectional area of the ther-
mocouple junction. The temperature curves are then
extrapolated to a zero diameter thermocouple. This
eliminates the need for correcting the measurements
for the losses from the junction. In the study by Baukal
[17], two different types of thermocouples were used
to measure fame temperatures. Diminishing-diameter
uncoated Type B thermocouples were used to measure
the gas temperatures in oxygen-enhanced-air/natural
gas fames. A coated Type L thermocouple was used to
measure the gas temperatures in oxygen-enhanced/
natural gas fames (see Figure 5.8). The raw, uncor-
rected measurements for the Type B thermocouples
are shown in Figure 5.9. Three different wire diam-
eters were tested. These are referred to by their wire
gauge numbers. The 20, 24, and 30 gauge wires had
© 2011 by Taylor and Francis Group, LLC
Temperature 105
diameters of 0.79, 0.48, and 0.23 mm, respectively.
Using the method recommended by Nichols (1900),
the measured temperatures, at a given axial location
L, were plotted as a function of the square of the wire
diameter. This is shown in Figure 5.10. The data was
then extrapolated to a zero wire diameter. The extrap-
olations have been curve-ftted with a second-order
polynomial. No extrapolation was possible at L = 0.5.
There, a hollow inner core could be seen in the fame.
The temperature measurements there were very
unsteady. Also, the thermocouple wires were hotter at
a distance from the junction, than at the junction. This
confrmed the hollow inner core.
Figure 5.11 shows a comparison of the extrapolated
gas temperatures and the corrected gas temperatures,
at L = 1, 2, 4, and 6. The corrected values were consis-
tently higher than the extrapolated values. One pos-
sible explanation is the danger in extrapolating only
three data points to zero diameter. Nichols [63] used
four different wire diameters, to determine an extrap-
olation to zero diameter. In that study, extrapolated
gas temperature curves increased much more rapidly,
as the data was extended to zero, than is shown in
Figure 5.10. In Nichols’s study, the wires ranged from
0.082 to 0.20 mm in diameter. In Baukal’s study, the
wire diameters ranged from 0.23 to 0.79 mm in diam-
eter. The rapid increase in Nichols’s data was for ther-
mocouples that were much smaller than those used
here. This rapid increase may not have been detected
in Baukal’s study, since the thermocouples were much
larger. Therefore, due to the uncertainty in the extrap-
olation, and due to the consistency of the corrected
temperatures for all three wire sizes, the corrected
temperatures (discussed next) were subsequently used
in Baukal’s study.
Figure 5.8
Type L fne wire thermocouple measuring the gas temperature in a
natural gas fame.
1800
1600
1400
1200
1000
20 gauge
24 gauge
t
j

(
K
)
30 gauge
800
600
0 1 2 3 4 5 6
L
p
7
Figure 5.9
Uncorrected gas temperature (t
j
) measurements, along the centerline
(R
p
 = 0) of an air/natural gas fame (q
f
 = 5.00 kW), with 20, 24, and 30
gauge Type B thermocouples. (From Baukal, C. E., Heat Transfer from
Flame Impingement Normal to a Plane Surface, Saarbrücken, Germany:
VDM Verlag, 2009.)
1800
1600
1400
1200
1000
t
j

(
K
)
d
T/C
2
(mm
2
)
800
0.0 0.5
L = 0.5
L = 1.0
L = 2.0
L = 4.0
L = 6.0
1.0 1.5 2.0 2.5 3.0 3.5 4.0
Figure 5.10
Gas temperature (t
j
) measurements, along the centerline (R
p
 = 0) of
an air/natural gas fame (q
f
 = 5.00 kW), with 20, 24, and 30 gauge
Type B thermocouples, extrapolated to zero diameter. (From Baukal,
C. E., Heat Transfer from Flame Impingement Normal to a Plane Surface,
Saarbrücken, Germany: VDM Verlag, 2009.)
© 2011 by Taylor and Francis Group, LLC
106 Industrial Combustion Testing
A third method to correct thermocouple measure-
ments is to compute the correction [64]. An energy bal-
ance at the thermocouple junction may be given as:
q
catalytic
 + q
conduction
 + q
radiation
 + q
convection
 = 0. (5.4)
Hayhurst and Kittelson [59] have shown that only sur-
face reactions contribute to catalytic heating. This can
generally be neglected when the thermocouple is coated.
Bradley and Matthews [62] have shown that conduction
may be neglected for wires over approximately 1 mm
(0.04 in.) long. When conduction effects have been
neglected, the energy balance then simplifes to:
t t t h
j
= +
T/C T/C T/C
σε
4
/ , (5.5)
where t
T/C
is the uncorrected temperature as measured
with the thermocouple and t
j
is the corrected tempera-
ture. The radiation from the environment to the junc-
tion has been neglected. For example, an emissivity of
0.14 has been recommended for uncoated platinum Type
B thermocouple [65]. An emissivity of 0.60 is recom-
mended for a coated Type L thermocouple [66]. A con-
vection coeffcient can be calculated using the equation
recommended by Hinze [67]:
Nu = + 0 42 0 57
0 2 0 33 0 5
. Pr . Pr Re .
. . .
(5.6)
An iterative procedure is required to calculate t
j

because the gas properties used to calculate the convec-
tion coeffcient are dependent on t
j
. The gas tempera-
ture t
j
may also be required to calculate the gas velocity
when, for example, Pitot-static probe measurements are
used to calculate the velocity. Son, Queiroz, and Wood
[68] presented a mathematical method for compensat-
ing for the thermal inertia effects of thermocouples.
These heat transfer effects can cause errors if not prop-
erly corrected.
A Type L (iridium) thermocouple was used to mea-
sure the gas temperatures in oxygen-enhanced/natural
gas fames [17]. The raw, uncorrected, measurements are
shown in Figure 5.12. These have been corrected using
the calculation method discussed above. At q
f
 = 15.0 kW,
Ω = 0.30 (30 vol% O
2
in the oxidizer) was the lowest oxi-
dizer composition that produced an acceptable fame.
Several measurements were made at q
f
 = 15.0 kW and
Ω = 0.40 (40 vol% O
2
in the oxidizer), before the Type L
thermocouple failed.
Table 5.2 shows a comparison of the maximum mea-
sured fame temperatures from Baukal’s study, after
correction, with the adiabatic fame temperatures. The
maximum measured temperatures, after corrections,
were 340–488 K less than the adiabatic temperatures.
This was due to heat losses from the fames, which were
not adiabatic. These measured temperatures compare
favorably with the gas temperatures reported in previ-
ous air and air/O
2
fame jet studies [69].
Using calculations to correct for fne wire thermo-
couple errors has been the most popular method for
correction. For example, Horsley, Purvis, and Tariq [70],
Kataoka, Shundoh, and Matsuo [71], You [72], Hargrave,
1900
1800
1700
1600
1500
30 gauge, corrected
24 gauge, corrected
20 gauge, corrected
Extrapolated
t
j

(
K
)
1400
1300
0 1 2 3 4 5 6
L
Figure 5.11
Comparison of the extrapolated and the corrected Type B thermocou-
ple gas temperature (t
j
) measurements along the centerline (R
p
 = 0) of
an air/natural gas fame (q
f
 = 5 kW). (From Baukal, C. E., Heat Transfer
from Flame Impingement Normal to a Plane Surface, Saarbrücken,
Germany: VDM Verlag, 2009.)
Table 5.2
Comparison of the Peak Measured (after Correction) Flame Temperatures with Adiabatic Flame
Temperatures
q
f
(Kw) Ω T/C Type t
j,max
(K) Adiabatic Flame Temperature (K)
5 0.21 B 1880 2220
15 0.30 L 2170 2546
15 0.35 L 2170 2658
Source: From Baukal, C. E., Heat Transfer from Flame Impingement Normal to a Plane Surface, Saarbrücken, Germany:
VDM Verlag, 2009.
© 2011 by Taylor and Francis Group, LLC
Temperature 107
Fairweather, and Kilham [73], Hustad, Røkke, and Sønju
[74], Hustad, and Sønju [75], and Hustad, Jacobsen, and
Sønju [76] calculated corrections for the heat losses from
the thermocouple junction.
5.4.3.3 Example Uses
Many researchers have used fne wire thermocouples
(T/Cs), with and without coatings. Costa et al. [77] used
a bare 300 µm diameter wire uncoated Pt/Pt13%Rh
thermocouple to measure gas temperatures inside an
industrial glass-melting furnace. Mital et al. [78] used
three different diameter (76, 125, and 200 µm) fne wire
Type R thermocouples to measure the gas temperature
inside porous ceramic radiant burners by extrapolating
the measurements to zero diameter.
Rumminger et al. [36] used fne wire thermocouples
to measure the gas temperature above the surface of a
porous radiant burner. At 5 mm (0.2 in.) above the sur-
face, they estimated a temperature correction of 270 K
(490°F) including the effects of re-radiation from the
burner surface to the thermocouple bead. Such a large
correction resulted from a relatively large bead size (460
µm), high bead emissivity (0.6), and relatively slow gas
fow (Re
d
 = 0.9, Nu = 0.7).
Fontes, Costa, and Azevedo [79] used 300 µm diam-
eter Type R thermocouple in an alumina sheath with
an external diameter of 4 mm and placed inside a 3 m
or 5 m long water-cooled stainless steel probe to mea-
sure gas temperatures in a black liquor recovery boiler.
Calculations showed the “true” temperature in the
regions of highest temperature did not exceed measure-
ments by more than 8%. Measurements were made at
multiple elevations, multiple locations at each elevation,
and at multiple insertion depths into the boiler. A tra-
versing mechanism was used to insert the probe, which
was checked frequently for deposition of black liquor
particles and sulfate deposits. The temperatures ranged
from approximately 700°C to 1000°C.
In a fame impingement study, two types of thermo-
couples were used to measure the gas temperatures [17].
For the lower temperature fames, uncoated bare wire
Type B thermocouples (Pt-60% Rh/Pt-6% Rh) were used.
The probes were designed according to the method rec-
ommended by Peterson [80]. However, those thermo-
couples melted at Ω > 0.25. For the higher temperature
fames, an Ir / 60% Ir-40% Rh thermocouple was used.
Candler [81] has referred to this thermocouple as a Type
L. According to Blackburn and Caldwell [82], this ther-
mocouple may be used to measure temperatures up to
2400 K (3900°F), within ±20 K (±40°F), using the appro-
priate reference table. This accuracy only refers to the
measured temperature. The actual gas temperature still
requires correction for the losses from the thermocouple
junction. The Type L thermocouple wires were 0.25 mm
(0.01 in.) in diameter. Figure 5.8 shows this type L
thermocouple measuring the temperature in an oxygen-
enhanced natural gas fame.
Fine wire thermocouples are of low cost and easy to
use. They have relatively short response times and good
spatial resolution. However, they are somewhat intru-
sive to the fame. Corrections are also required due to
heat transfer effects at the junction bead. These include
radiation from the bead to the environment and the heat
conduction along the thermocouple wires. Often the
junction must be coated to prevent catalytic reactions.
There are material limitations, due to high temperature
oxidizing and reducing conditions. They can only make
so many point measurements over a span of time to get
a temperature profle.
5.4.4 line reversal
Many researchers have used this technique, originally
suggested by Féry [83], to measure fame temperatures.
The fame is commonly seeded with sodium salt. The
salt vaporizes in the fame and dissociates into sodium
atoms and other products. These excited atoms emit at
a specifc wavelength. This radiation is then compared
to a calibrated reference source, such as a tungsten fla-
ment lamp. The temperature of this source is adjusted,
until it matches the seed radiation. To the naked eye, the
background source appears to disappear when it is at
2200
2000
1800
1600
Ω = 0.30, Uncorrected
Ω = 0.35, Uncorrected
Ω = 0.30, Corrected
Ω = 0.35, Corrected
t
j

(
K
)
1400
1200
0 1 2 3 4 5 6
L
p
7
Figure 5.12
Gas temperature measurements (t
j
) along the centerline (R
p
 = 0)
of fames (q
f
 = 15.0 kW), as a function of the oxidizer composition
(Ω = 0.30 and 0.35), using a Type L thermocouple. (From Baukal, C. E.,
Heat Transfer from Flame Impingement Normal to a Plane Surface,
Saarbrücken, Germany: VDM Verlag, 2009.)
© 2011 by Taylor and Francis Group, LLC
108 Industrial Combustion Testing
the same temperature as the fame. Strong et al. [84] give
a general discussion of this technique to measure the
temperatures in fames [84]. Thomas [85,86] discusses
some of the problems associated with this technique as
well as a method for automatically remotely operating a
system for using the method.
Kilham [87] used a spectrometer to compare the
fame with the flament. Fells and Harker [88] estimated
the accuracy of their line reversal measurements to be
20K. Nawaz [89] and Fairweather and colleagues [90,91]
introduced sodium chloride smoke, by passing N
2
over
molten salt, through a tube in the center of their burner.
Therefore, only the temperatures along the central axis
of the fame were measured. This reduced the infuence
of ambient air entrainment cooling effects, at the fame
periphery. Nawaz used a high pressure xenon arc lamp
(maximum temperature of 5000 K or 8500°F) as the com-
parison radiator, since the temperature levels were out
of the range of the conventional tungsten lamps that
were available at that time. Measured temperatures
were 50–100 K (90°F–180°F) below the calculated adia-
batic fame temperature. Fairweather [91] measured a 33
K (59°F) and 40 K (72°F) reduction, across the equilib-
rium zone between the reaction zone and the stagnation
point, for laminar and turbulent fames, respectively.
Atmospheric air entrainment into the fame was thought
to be a possible explanation of this effect.
This method is only slightly intrusive, due to the seed-
ing. It is of relatively low cost. It may be effectively used for
very high temperatures. However, the seeded feld must
be uniform. This may require the addition of a carrier gas
to transport the seeds into the fame. Only an average gas
temperature, across the fame, is measured. A calibrated
reference light source is required and the adjustment for
the source to vanish is somewhat subjective.
5.4.5 acoustic Pyrometry
The speed of sound in a gas is dependent on tempera-
ture. For an ideal gas, the speed of sound is:

c RT
ideal
= γ ,
(5.7)
where γ is the ratio of specifc heats at constant pressure
and volume (C
p
/C
v
), R is the specifc gas constant, and T
is the absolute temperature. The product γR is sometimes
referred to as the acoustic constant, which is dependent
on the gas composition [92]. If the speed is known, then
the temperature can be calculated from the equation,
assuming the gas composition is known. The error in
not knowing the exact gas composition of combustion
products can be signifcant, but can usually be mini-
mized with proper knowledge of the fuel and operating
conditions [92]. This principle can be used to measure
temperature in a gas by measuring the fight time for
sound to travel between a source and a receiver that are a
known distance apart. The distance traveled divided by
the fight time gives the speed. This technique is referred
to as acoustic pyrometry and has been applied in indus-
trial combustors, particularly boilers [93–95] and process
heaters [96,97]. The average temperature is measured
across the fight path. By using an array of sources and
receivers and a deconvolution algorithm [98], a tempera-
ture profle in a combustor can be developed.
A common method used to generate the sound waves
is using blasts of air through a nozzle. One source can
be used with multiple receivers. In some cases, Pitot
tubes and thermocouples are used in conjunction with
the acoustical pyrometry system to improve accuracy.
The practical frequency range for these pyrometers
is between 500 and 2000 Hz. A study funded by the
Electric Power Research Institute (EPRI) showed acous-
tic pyrometry is easy to use, accurate, nonintrusive, and
real-time, which makes it useful for monitoring and
control [93]. It can be used as a diagnostic tool to identify
burner and furnace problems.
5.5 Surface Temperature
There are many surface temperatures that may be
important in industrial combustion applications. The
product temperature (usually measured at the surface)
is typically an important variable that often must be
maintained within a relatively narrow band. In some
cases, the product temperature is part of the control
system, where the fring rate is adjusted to maintain
a target product temperature. The surface tempera-
ture of equipment inside the combustion chamber,
such as process tubes, is often important. Figure 5.13
shows an example of an infrared scan of the process
tubes inside a process heater. This particular image
was taken while the heater tubes were being de-coked,
where coke build-up inside the process tubes is burned
out. The location of the de-coking can be seen by the
elevated temperatures near the middle of the photo.
Furnace refractory temperatures may be of interest to
ensure they do not exceed the material limits. Burner
temperatures may be important to ensure there are no
hot spots that could reduce the operating life. There are
two common ways to measure the surface temperature
of an object that are discussed next.
5.5.1 embedded Thermocouple
In the fame impingement study by Baukal [17], a
Type K thermocouple was imbedded in the impinge-
ment rings of copper calorimeters (see Figure 5.14). The
© 2011 by Taylor and Francis Group, LLC
Temperature 109
thermocouples had a 0.5 mm (0.02 in.) o.d. inconel sheath
and the clearance hole in the side of each impingement
ring was 0.64 mm (0.025 in.) i.d. A high conductivity
antiseize sealant was applied to the end of each thermo-
couple, before it was inserted into a ring, to ensure good
thermal contact. These thermocouples measured the
temperatures at the middle of each impingement ring.
The hot face temperatures were then calculated using
the one-dimensional, steady-state conduction equation.
Using the calculated heat fux to each ring, the hot face
temperature, t
w
, was calculated using:

t t
q l
k
w
= +
′′
middle
1
1
2
,
(5.8)
where t
middle
was the measured temperature at the mid-
dle of the ring, q” was the calculated heat fux to the
ring, l
1
was the thickness of the impingement ring, and
k
1
was the thermal conductivity of the impingement
ring. A typical result is shown in Figure 5.15. This graph
shows that the wall temperatures decreased as the axial
distance between the burner and the target decreased.
However, for L > 2, the differences became smaller.
5.5.2 infrared Detectors
Thermal imaging systems have been used for years in
a variety of applications including military and intel-
ligence surveillance activities. Within about the last
decade, the cost of these systems has declined enough
that they are now being routinely used to measure tem-
peratures in combustion applications. In combustion
systems, these IR imaging systems are typically used
to measure the temperature of the furnace or heater
walls, the heat load which might be a material like scrap
metal or a tube used for process fuid heating, or the
1267.5°F
947.8°F
1200
1100
1000
Figure 5.13
Infrared scan of tube temperatures inside a process heater during a
de-coking cycle. (Courtesy of Quest Tru-Tec.)
Figure 5.14
Thermocouple embedded in a ring calorimeter to measure surface
temperature. (From Baukal, C. E., Heat Transfer from Flame Impingement
Normal to a Plane Surface, Saarbrücken, Germany: VDM Verlag, 2009.)
1000
900
800
700
600
R
eff
= 0.16
R
eff
= 0.37
R
eff
= 0.59
R
eff
= 0.82
R
eff
= 1.04
t
w

(
K
)
500
400
0 1 2 3 4 5 6
L
Figure 5.15
Surface temperature (t
w
) for O
2
/natural gas fames (q
f
 = 15.0 kW)
impinging on an untreated stainless surface. (From Baukal, C. E., Heat
Transfer from Flame Impingement Normal to a Plane Surface, Saarbrücken,
Germany: VDM Verlag, 2009.)
© 2011 by Taylor and Francis Group, LLC
110 Industrial Combustion Testing
burner itself. Imaging systems are primarily used as
a diagnostic to detect hot spots which could indicate
potential problem areas.
Infrared detectors have a number of potential advan-
tages over thermocouples [99]:
faster response times •
very high accuracies •
can measure temperatures on moving objects •
can measure higher temperatures than conven- •
tional thermocouples
can give two-dimensional temperature profles •
do not disturb the surface being measured •
are not dependent on how they are attached •
to the surface, as thermocouples are, to get an
accurate surface temperature measurement
There are also some potential disadvantages com-
pared to thermocouples:
the surface emissivity is often required, except •
for two-color pyrometers
measurements can be impaired by anything, •
including combustion products, in between the
pyrometer and the surface, which could absorb
any of the radiation emitted by the surface
the front lens of the pyrometer must be kept •
clean
the electronics in the pyrometer must be kept •
cool
calibration is typically more involved than for a •
thermocouple
The theory of radiation pyrometry is available else-
where [100].
The original devices used for combustion applica-
tions were infrared pyrometers that give a temperature
based on the infrared energy emitted by the object and
an input emissivity for the object. An example of an
infrared pyrometer being used to measure burner sur-
face temperature is shown in Figure 5.16. Newer devices
use imaging technology to produce thermal images of
the temperature profle of an object (see Figure 5.13)
[101]. These devices are capable of producing real-time
images of dynamically changing temperature profles.
A typical system consists of a high resolution infrared
camera, a computer, and thermal image analysis soft-
ware. Besides displaying the temperature profle for an
object, the software provides capabilities for analyzing
the data. Due to the fast response of the new systems, it
is also possible to use them for real-time control of the
heating system to produce specifc temperature profles
in the load.
Lappe [102] gave some recommendations for the instal-
lation and use of single-color IR thermometers. Single
color means the sensor is tuned to operate at one specifc
wavelength. The most important factor in measuring the
correct surface temperature is to know the emissivity of
the body being measured. The emissivity of a graybody
is constant for all wavelengths and usually does not
change signifcantly with temperature unless there is a
change in the material or in its surface characteristics.
For example, if a solid material melts and becomes liq-
uid there may be a signifcant change in its emissivity.
Another example is if a metal oxidizes or becomes rough-
ened from wear, its emissivity also may change dramati-
cally. Lappe recommends using the shortest wavelength
instrument possible to measure the surface temperature,
for two reasons. The frst is that the difference in the
emissivity of common surfaces such as metals is usually
signifcantly less at shorter wavelengths, with the differ-
ence increasing with wavelength. Therefore, there will
be less error if the emissivity used on the pyrometer is
not properly adjusted for changes in the surface condi-
tions of the body whose temperature is being measured.
The second reason is that lower wavelength instruments
typically have less error in the temperature reading com-
pared to higher wavelength units. Lappe also gave some
recommendations for the installation of these units:
make sure the unit is focused properly on the •
object of interest and that the spot size is the
correct size
make sure there are no obstructions (e.g., sight- •
ports walls, smoke, steam, etc.) between the
pyrometer and the surface being measured
Figure 5.16
Example of an infrared pyrometer measuring the surface tempera-
ture of a porous refractory matrix burner. (From Baukal, C.E., Heat
Transfer in Industrial Combustion, Boca Raton, FL; CRC Press, 2000.)
© 2011 by Taylor and Francis Group, LLC
Temperature 111
fber optic systems may be appropriate for cer- •
tain applications
ensure that the interconnecting cable is not •
overheated or installed with other wires that
may carry high voltages that could adversely
affect the signal
make sure the sensing head is kept clean and •
cool but not overcooled, which could cause con-
densation in the unit
keep the lens clean with a purge of clean dry •
air
have the instrument calibrated at least annually •
Two-color pyrometers avoid the problem of determin-
ing the emissivity of the surface. Those pyrometers
make measurements at two different wavelengths and
with appropriate computations determine temperature
without knowing the emissivity.
Figure 5.17 shows infrared temperature measure-
ments made on a burner [103]. By selecting an appro-
priate wavelength for the intensity measurement (in
this case, the wavelength was 3.9 µm), the effects of CO
2

and H
2
O absorption between the emitting surface and
the infrared camera were negated because those gases
are essentially transparent at that wavelength. Surface
temperatures can then be readily measured “through”
a fame. It is very diffcult to make reliable gas tempera-
ture measurements by measuring the infrared emis-
sion because the emission from gases depends on the
temperature of the gas volume, the composition of the
gas volume, as well as the dimension of the gas volume.
Since most real applications involve nonisothermal gas
volumes (such as a fame in a furnace), IR measurements
are not feasible.
Hernberg, Stenberg, and Zethraeus [104] used a two-
color infrared pyrometer coupled to a fberoptic probe
to measure burning fuel particle temperatures in a
fuidized bed furnace. The probe consisted of a 1 mm
diameter quartz fberguide encapsulated in an uncooled
stainless steel tube. The probe was operated in tempera-
tures of about 850°C and pressures up to 20 bar. Two-
color (wavelengths of 0.65 µm and 1.05 µm) allowed
determination of both temperature and a parameter
related to particle emissivity, particle size, and geomet-
rical confguration factor. The instrument was calibrated
against a blackbody standard. A statistical method was
developed to determine particle size from these mea-
surements in addition to temperature.
5.5.3 Other Techniques
Stephan et al. [105,106] used a technique referred to as
microwave radiometry where low cost X-band micro-
waves are used to measure surface temperatures in
industrial combustion environments with suspended
particles such as water vapor and dust. In most applica-
tions, IR pyrometry is used to inexpensively measure
product temperatures. However, in heavily particle-laden
environments, an IR pyrometer’s view is obscured mak-
ing it diffcult to get an accurate temperature measure-
ment. In microwave radiometry, the longer wavelengths
are not scattered by the particulates. The temperature
range of one version of this device was from room tem-
perature up to about 500°C (900°F). This device was
used in a demonstration that showed the device could
accurately measure the surface temperature of a layer
of carbon adsorbent through an artifcially created,
water-based fog [105]. A higher temperature version of
the device was used to measure the wall temperatures
inside a dusty cement kiln that ranged up to approxi-
mately 1500°C (2700°F).
5.6 Future
Measuring temperatures inside industrial combustors
and in fames continues to be challenging. More tools
are being developed and applied to these conditions.
Some of those tools are not yet commercially viable,
which means they may be too expensive, require spe-
cialized knowledge to be properly used, or have not
yet been ruggedized enough to handle the challenges
of an industrial environment. This is likely to change
in the future as the equipment costs decline and more
*> 900.0°C
895.2
875.4
855.1
834.3
812.5
789.0
762.6
730.8
691.3
648.9
610.4
573.2
530.5
*< 530.1°C
Figure 5.17
(See color insert following page 424.) Infrared thermal image of
a fame in a furnace. (From Singh, P., Henneke, M., Jayakaran, J. D.,
Hayes, R., and Baukal, C. E., The John Zink Combustion Handbook, Boca
Raton, FL: CRC Press, 2001.)
© 2011 by Taylor and Francis Group, LLC
112 Industrial Combustion Testing
experience is gained using the techniques under indus-
trial operating conditions.
One might argue that the rapid expansion in the use of
computational fuid dynamics [107], may cause a decline
in experimental measurements. However, the old adage
“garbage in = garbage out” is still relevant because a
computer model that has not been validated is problem-
atic. Given the extreme operating conditions, environ-
mental regulations, and safety considerations associated
with industrial combustion applications, relying solely
on computer models in the absence of validating mea-
surements seems to be imprudent. Therefore, making
temperature measurements in industrial combustion
applications is expected to continue into the foreseeable
future.
References
1. Baukal, C. E. Heat Transfer in Industrial Combustion. Boca
Raton, FL: CRC Press, 2000.
2. Baukal, C. E. Industrial Combustion Pollution and Control.
New York: Marcel Dekker, 2004.
3. Reed, R. J. North American Combustion Handbook. Vol. 1,
3rd ed. Cleveland, OH: North American Mfg. Co., 1986.
4. Baukal, C., Hong, J., Bussman, W., and Waibel, R. “Cold
Furnace Startup Emissions.” Chemical Engineering
Progress 103, no. 2 (2007): 42–46.
5. Bussman, W., and Baukal, C. “Ambient Condition Effects
on Process Heater Effciency.” Energy, 34, no. 10 (2009).
1624–35.
6. Bussman, W., and Baukal, C. “Ambient Condition Effects
on Process Heater Emissions.” Paper IMECE2008-68284,
Proceedings of the International Mechanical Engineering
Congress & Exhibition, Boston, MA, November 2008.
7. Hayes, R. R., Wang, J., McQuay, M. Q., and Webb, B. W.
“Predicted and Measured Glass Surface Temperatures
in an Industrial, Regeneratively Gas-Fired Flat Glass
Furnace.” Glass Science and Technology-Glastechnische
Berichte 72, no. 12 (1999): 367–77.
8. Honner, M., Vesely, Z., and Svantner, M. “Temperature
and Heat Transfer Measurement in Continuous Reheating
Furnaces.” Scandinavian Journal of Metallurgy 32, no. 5
(2003): 225–32.
9. Rafdi, N., and Blasiak, W. “Heat Transfer Characteristics
of HiTAC Heating Furnace Using Regenerative Burners.”
Applied Thermal Engineering 26, no. 16 (2006): 2027–34.
10. Eckert, E. R. G., and Goldstein, R. J. Measurements in Heat
Transfer. 2nd ed. Washington: Hemisphere, 1976.
11. Goldstein, R. J., and Chiang, H. D. “Measurement of
Temperature and Heat Transfer.” In Handbook of Heat
Transfer Applications, edited by W. M. Rohsenow, J. P.
Hartnett, and E. N. Ganic´, 2nd ed. New York: McGraw-
Hill, 1985.
12. Holman, J. P. Experimental Methods for Engineers. 5th ed.
New York: McGraw-Hill, 1989.
13. Deshmukh, Y. V. Industrial Heating. Boca Raton, FL: CRC
Press, 2005.
14. Bowman, C. T. “Probe Measurements in Flames.” Progress
in Astronautics and Aeronautics 53 (1977): 1–24.
15. Cornforth, J. R. Combustion Engineering and Gas Utilisation
(sic). 3rd ed. London: E&FN Spon, 1992.
16. Okoh, C. I., and Brown, R. A. Combustion Experimentation
Handbook. Gas Research Institute report GRI-88/0143,
Chicago, IL, 1988.
17. Baukal, C. E. Heat Transfer from Flame Impingement Normal
to a Plane Surface. Saarbrücken, Germany: VDM Verlag,
2009.
18. Durão, D. F. G., Heitor, M. V., Whitelaw, J. H., and Witze,
P. O. Combusting Flow Diagnostics. Dordrecht: Kluwer
Academic Publishers, 1992.
19. Taylor, A.M.K.P. Instrumentation for Flows with Combustion.
London: Academic Press, 1993.
20. Eckbreth, A. C. Laser Diagnostics for Combustion
Temperature and Species. Cambridge, MA: Abacus Press,
1988.
21. Chigier, N. Combustion Measurements. New York:
Hemisphere Publishing Corp., 1991.
22. Heitor, M. V., and Moreira, A. L. N. “Probe Measurements
of Scalar Properties in Reacting Flows.” In Combusting
Flow Diagnostics, edited by D. F. G. Durão, M. V. Heitor,
J. H. Whitelaw, and P. O. Witze, 79–136. Netherlands:
Kluwer Academic Publishers, 1992.
23. Chedaille, J., and Braud, Y. Vol. 1: Measurements in Flames.
New York: Crane, Russak & Co., 1972.
24. Goldman, Y. “Gas Temperature Measurement in
Combustors by Use of Suction Pyrometry.” in Heat
Transfer in Furnaces, edited by C. Presser and D. G. Lilley,
19–22. New York: ASME HTD-Vol. 74, 1987.
25. Beér, J. M., and Chigier, N. A. Combustion Aerodynamics
London: Applied Science Publishers, 1972.
26. Vizioz, J.-P., and Lowes, T. M. “Convective Heat Transfer
from Impinging Flame Jets.” International Flame
Research Foundation report F 35/a/6, IJmuiden, the
Netherlands, 1971.
27. Smith, R. B., and Lowes, T. M. “Convective Heat Transfer
from Impinging Tunnel Burner Flames: A Short Report on
the NG-4 Trials.” International Flame Research Foundation
report F 35/a/9, IJmuiden, the Netherlands, 1974.
28. Rajani, J. B., Payne, R., and Michelfelder, S. “Convective
Heat Transfer from Impinging Oxygen-Natural Gas
Flames: Experimental Results from the NG5 Trials.”
International Flame Research Foundation report F
35/a/12, IJmuiden, the Netherlands, 1978.
29. Grey, J. “Thermodynamic Methods of High Temperature
Measurement.” ISA Transactions 4 (1965): 102–15.
30. Ivernel, A., and Vernotte, P. “Etude expérimentale de
l’amélioration des transferts convectis dans les fours
par suroxygénation du comburant.” Revue Generale de
Thermique 18, nos. 210–211 (1979): 375–91.
31. Laurendeau, N. M. “Temperature Measurements
by Light-Scattering Methods.” In Developments in
Experimental Techniques in Heat Transfer and Combustion,
edited by R. O. Warrington, M. M. Chen, J. D. Felske, and
W. L. Grosshandler, 45–65. New York: ASME HTD-Vol.
71, 1987.
© 2011 by Taylor and Francis Group, LLC
Temperature 113
32. Kondic, N. N. “Temperature Field Measurement in
Flames by External Means.” In Heat Transfer in Flames,
edited by N. H. Afgan and J. M. Beer, 353–63. Washington,
DC: Scripta Book Company, 1974.
33. Posillico, C. J. “Raman Spectroscopic and LDV
Measurements of a Methane Jet Impinging Normally on
a Flat Water-Cooled Boundary.” PhD thesis, Polytechnic
Institute of New York, 1986.
34. Hughes, P. M. J., Lacelle, R. J., and Parameswaran, T. “A
Comparison of Suction Pyrometer and CARS Derived
Temperatures in an Industrial Scale Flame.” Combustion
Science and Technology 105 (1995): 131–45.
35. Hofmann, D., and Leipertz, A. “Temperature Field
Measurements in a Sooting Flame by Filtered Rayleigh
Scattering (FRS).” 26th Symposium (International) on
Combustion, The Combustion Institute, Pittsburgh, PA,
945–50, 1996.
36. Rumminger, M. D., Dibble, R. W., Heberle, N. H., and
Crosley, D. R. “Gas Temperature Above a Porous
Radiant Burner: Comparison of Measurements and
Model Predictions.” 26th Symposium (International) on
Combustion, The Combustion Institute, Pittsburgh, PA,
1755–62, 1996.
37. Chao, C. M., and Goulard, R. “Nonlinear Inversion
Techniques in Flame Temperature Measurements.” In
Heat Transfer in Flames, edited by N. H. Afgan and J. M.
Beer, 295–337 Washington, DC: Scripta Book Company,
1974.
38. Hommert, P. J., Viskanta, R., and Mellor, A. M. “Flame
Temperature Measurements by Spectral Remote
Sensing.” Combustion and Flame 30 (1977): 295–308.
39. Kim, H. K., and Song, T.-H. “Measurement of Gas
Temperature Distributions in a Test Furnace Using
Spectral Remote Sensing.” Journal of Quantitative
Spectroscopy & Radiative Transfer 73 (2002): 517–28.
40. Kim, H. K., and Song, T.-H. “Determination of the Gas
Temperature Profle in a Large-Scale Furnace Using a
Fast/Effcient Inversion Scheme for the SRS Technique.”
Journal of Quantitative Spectroscopy & Radiative Transfer 93
(2005): 369–81.
41. Song, T.-H. “Spectral Remote Sensing for Furnaces
and Flames,” Heat Transfer Engineering 29, no. 4 (2008):
417–28.
42. Reuss, D. L. “Temperature Measurements in a Radially
Symmetric Flame Using Holographic Interferometry.”
Combustion and Flame 49 (1983): 207–19.
43. Char, J.-M., and Yeh, J.-H. “The Measurement of Open
Propane Flame Temperature Using Infrared Technique,”
Journal of Quantitative Spectroscopy & Radiative Transfer 56,
no. 1 (1996): 133–44.
44. Correia, D. P., Ferrao, P., and Caldeira-Pires, A. “Flame
Three-Dimensional Tomography Sensor for In-Furnace
Diagnostics.” Proceedings of the Combustion Institute 28
(2000): 431–38.
45. Zhou, H.-C., Lou, C., Cheng, Q., Jiang, Z., He, J., Huang,
B., Pei, Z., and Lu, C. “Experimental Investigations
on Visualization of Three-Dimensional Temperature
Distributions in a Large-Scale Pulverized-Coal-Fired
Boiler Furnace.” Proceedings of the Combustion Institute 30
(2005): 1699–1706.
46. Cookson, R. A., Dunham, P. G., and Kilham, J. K. “Non-
Catalytic Coatings for Thermocouples.” Combustion and
Flame 8 (1964): 168–70.
47. Kent, J. H. “A Noncatalytic Coating for Platinum-
Rhodium Thermocouples.” Combustion and Flame 14
(1970): 279–82.
48. Kaskan, W. E. “The Dependence of Flame Temperatures
on Mass Burning Velocity.” Sixth Symposium (International)
on Combustion, 134–43. Reinbold, NY, 1957.
49. Cookson, R. A., and Kilham, J. K. “Energy Transfer from
Hydrogen-Air Flames.” Ninth Symposium (International)
on Combustion, 257–63. New York: Academic Press, 1963.
50. Kilham, J. K., and Dunham, P. G. “Energy Transfer from
Carbon Monoxide Flames.” 11th Symposium (International)
on Combustion, The Combustion Institute, Pittsburgh, PA,
899–905, 1967.
51. Milson, A., and Chigier, N. A. “Studies of Methane-Air
Flames Impinging on a Cold Plate.” Combustion and
Flame 21 (1973): 295–305.
52. Madson, J. M., and Theby, E. A. “SiO
2
Coated
Thermocouples.” Combustion Science and Technology 36
(1984): 205–9.
53. Pollock, D. D. “Thermocouples in High-Temperature
Reactive Atmospheres.” Combustion Science and Technology
42 (1984): 111–13.
54. Hoogendoorn, C. J., Popiel, C. O., and van der Meer,
T. H. “Turbulent Heat Transfer on a Plane Surface in
Impingement Round Premixed Flame Jets.” Proceedings
of 6th International Heat Transfer Conference, Toronto,
Vol. 4, pp. 107–12, 1978.
55. Popiel, C. O., van der Meer, T. H., and Hoogendoorn, C. J.
“Convective Heat Transfer on a Plate in an Impinging
Round Hot Gas Jet of Low Reynolds Number.”
International Journal of Heat and Mass Transfer 23 (1980):
1055–68.
56. van der Meer, T. H. “Stagnation Point Heat Transfer from
Turbulent Low Reynolds Number Jets and Flame Jets.”
Experimental Thermal and Fluid Science 4 (1991): 115–26.
57. Sato, A., Hashiba, K., Hasatani, M., Sugiyama, S.,
and Kimura, J. “A Correctional Calculation Method
for Thermocouple Measurements of Temperatures in
Flames.” Combustion and Flame 24 (1975): 35–41.
58. Moffat, R. J. “Temperature and Heat Transfer
Measurements.” In The CRC Handbook of Mechanical
Engineering, edited by F. Kreith, 4-182–4-205. Boca Raton,
FL: CRC Press, 1998.
59. Hayhurst, A. N., and Kittelson, D. B. “Heat and Mass
Transfer Considerations in the Use of Electrically Heated
Thermocouples of Iridium versus an Iridium/Rhodium
Alloy in Atmospheric Pressure Flames.” Combustion and
Flame 28 (1977): 301–17.
60. Buhr, E., Haupt, G., and Kremer, H. “Heat Transfer from
Impinging Turbulent Jet Flames to Plane Surfaces.” In
Combustion Institute European Symposium 1973, edited
by F. J. Weinberg, 607–12. New York: Academic Press,
1973.
61. Kremer, H., Buhr, E., and Haupt, R. “Heat Transfer from
Turbulent Free-Jet Flames to Plane Surfaces.” In Heat
Transfer in Flames, edited by N. H. Afgan and J. M. Beér,
463–72. Washington, DC: Scripta Book Company, 1974.
© 2011 by Taylor and Francis Group, LLC
114 Industrial Combustion Testing
62. Bradley, D., and Matthews, K. J. “Measurements of High
Gas Temperatures with Fine Wire Thermocouples.”
Journal of Mechanical Engineering Science 10, no. 4 (1968):
299–305.
63. Nichols, E. L. “On the Temperature of the Acetylene
Flame.” Physical Review 10 (1900): 234–52.
64. Fristrom, R. M., and Westenberg, A. A. Flame Structure.
New York: McGraw-Hill, 1965.
65. Sparrow, E. M., and Cess, R. D. Radiation Heat Transfer.
New York: McGraw-Hill, 1978.
66. Peterson, R. C., and Laurendeau, N. M. “The Emittance
of Yttrium-Beryllium Oxide Thermocouple Coating.”
Combustion and Flame 60 (1985): 279–84.
67. Hinze, J. O. Turbulence. New York: McGraw-Hill, 1959.
68. Son, S. F., Queiroz, M., and Wood, C. G. “Compensation
of Thermocouples for Thermal Inertia Effects Using a
Digital Deconvolution.” In Heat Transfer Phenomena in
Radiation, Combustion, and Fires, edited by R. K. Shah,
515–22. New York: ASME HTD-Vol. 106, 1989.
69. Baukal, C. E., and Gebhart, B. “A Review of Flame
Impingement Heat Transfer Studies—Part 2:
Measurements.” Combustion Science & Technology 104,
nos. 4–6 (1995): 359–85.
70. Horsley, M. E., Purvis, M. R. I., and Tariq, A. S.
“Convective Heat Transfer From Laminar and Turbulent
Premixed Flames.” Heat Transfer 1982, vol. 3, edited by
U. Grigull, E. Hahne, K. Stephan, and J. Straub, 409–15.
Washington, DC: Hemisphere 1982.
71. Kataoka, K., Shundoh, H., and Matsuo, H. “Convective
Heat Transfer Between a Flat Plate and a Jet of Hot
Gas Impinging On It.” In Drying ’84, edited by A. S.
Mujumdar, 218–27. New York: Hemisphere/Springer-
Verlag 1984.
72. You, H.-Z. “An Investigation of Fire-Plume Impingement
on a Horizontal Ceiling: 2—Impingement and Ceiling-
Jet Regions.” Fire & Materials 9, no. L (1985): 46–56.
73. Hargrave, G. K., Fairweather, M., and Kilham, J. K.
“Forced Convective Heat Transfer from Premixed
Flames—Part 1: Flame Structure.” International Journal of
Heat and Fluid Flow 8, no. 1 (1987): 55–63.
74. Hustad, J. E., Røkke, N. A., and Sønju, O. K. “Heat
Transfer to Pipes Submerged in Lifted Buoyant Diffusion
Flames.” In Experimental Heat Transfer, Fluid Mechanics,
and Thermodynamics, 1991, edited by J. F. Keffer, 567–74.
New York: Elsevier, 1991.
75. Hustad, J. E., and Sønju, O. K. “Heat Transfer to Pipes
Submerged in Turbulent Jet Diffusion Flames.” In Heat
Transfer in Radiating and Combusting Systems, 474–90.
Berlin: Springer-Verlag, 1991.
76. Hustad, J. E., Jacobsen, M., and Sønju, O. K. Radiation
and Heat Transfer in Oil/Propane Jet Diffusion Flames,
Institution of Chemical Engineers Symposium Series 10, no.
129 (1992): 657–63.
77. Costa, M., Mourão, M., Baltasar, J., and Carvalho, M.
G. “Combustion Measurements in an Industrial Glass-
Melting Furnace.” Journal of the Institute of Energy 69, no.
479 (1996): 80–86.
78. Mital, R., Gore, J. P., Viskanta, R., and Singh, S. “Radiation
Effciency and Structure of Flames Stabilized Inside
Radiant Porous Ceramic Burners.” In Combustion and
Fire, edited by M. C. McQuay, W. Schreiber, E. Bigzadeh,
K. Annamalai, D. Choudhury, and A. Runchal, 131–37.
New York: ASME Proceedings of the 31st National Heat
Transfer Conference, Vol. 6, ASME HTD-Vol. 328, 1996.
79. Fontes, P., Costa, M., and Azevedo, J. L. T. “Measurements
in and Modeling of a Black Liquor Recovery Boiler.”
Combustion Science and Technology 180, no. 3 (2008):
494–508.
80. Peterson, R. C. “Kinetics of Hydrogen-Oxygen-Argon and
Hydrogen-Oxygen-Argon-Pyridine Combustion Using
a Flat Flame Burner.” PhD thesis, Purdue University,
1981.
81. Candler, E. M. Thermocouple Reference Tables, Air Force
Flight Dynamics Laboratory report no. AFFDL-TR-66-
178, Wright-Patterson Air Force Base, Ohio, 1967.
82. Blackburn, G. F., and Caldwell, F. R. Reference Tables
for Thermocouples of Iridium-Rhodium Alloys Versus
Iridium, J Research of National Bureau of Standards—C.
Engineering and Instrumentation 68C, no. 1 (1964):
41–59.
83. Féry, C. “Sur la température des fammes.” Comptes
Rendus de l’Académie des Science, no. 22 (1903): 909–12.
84. Strong, H. M., Bundy, F. P., and Larson, D. A. “Temperature
Measurement on Complex Flames by Sodium Line
Reversal and Sodium D Line Intensity Contour Studies”
Third Symposium (International) on Combustion, 641–47.
Baltimore, MD: The Williams & Wilkins Co., 1949.
85. Thomas, D. L. “Problems in Applying the Line Reversal
Method of Temperature Measurement to Flames.”
Combustion and Flame 12 (1968): 541–49.
86. Thomas, D. L. “An Automatic Remotely Operated Sodium
D-Line Reversal Temperature Sensing Technique.”
Combustion and Flame 12 (1968): 569–74.
87. Kilham, J. K. “Energy Transfer from Flame Gases to
Solids.” Third Symposium on Combustion and Flame
and Explosion Phenomena, 733–40. Baltimore, MD: The
Williams and Wilkins Co., 1949.
88. Fells, I., and Harker, J. H. “An Investigation Into Heat
Transfer from Unseeded Propane-Air Flames Augmented
with D.C. Electrical Power.” Combustion and Flame 12
(1968): 587–96.
89. Nawaz, S. “Heat Transfer from Oxygen Enriched
Methane Flames.” PhD thesis, The University of Leeds,
Leeds, U.K., 1973.
90. Fairweather, M., Kilham, J. K., and Nawaz, S. “Stagnation
Point Heat Transfer from Laminar, High Temperature
Methane Flames.” International Journal of Heat and Fluid
Flow 5, no. 1 (1984): 21–27.
91. Fairweather, M., Kilham, J. K., and Mohebi-Ashtiani, A.
“Stagnation Point Heat Transfer from Turbulent Methane-
Air Flames.” Combustion Science Technology 35 (1984):
225–38.
92. Young, K. J., Ireland, S. N., Melendez-Cervates, M. C., and
Stones, R. “On the Systematic Error Associated with the
Measurement of Temperature Using Acoustic Pyrometry
in Combustion Products of Unknown Mixture.”
Measurement Science and Technology 9 (1998): 1–5.
93. Muzio, L. J., Eskinazi, D., and Green, S. F. “Acoustic
Pyrometry: New Boiler Diagnostic Tool.” Power
Engineering 93, no. 11 (1989): 49–52.
© 2011 by Taylor and Francis Group, LLC
Temperature 115
94. Kleppe, J. A. “Measurement of Combustion Gas
Temperature Using Acoustic Pyrometry.” Proceedings
of Instrumentation, Control, and Automation in the Power
Industry 33 (1990): 25–40.
95. Bramanti, M., Salerno, E. A., Tonazzini, A., Pasini,
S., and Gray, A. “An Acoustic Pyrometer System for
Tomographic Thermal Imaging in Power Plant Boilers.”
IEEE Transactions on Instrumentation and Measurement 45,
no. 1 (1996): 159–67.
96. DiSimone, G. “Combustion Control.” Hydrocarbon
Engineering 11, no. 10 (2006): 53–55.
97. DiSimone, G. “Furnace Combustion Management.”
Petroleum Technology Quarterly 11, no. 4 (2006): 103–11.
98. Lu, J., Wakai, K., Takahashi, S., and Shimizu, S. “Acoustic
Computer Tomographic Pyrometry for Two-Dimensional
Measurement of Gases Taking into Account the Effect of
Refraction of Sound Wave Paths.” Measurement Science
and Technology 11 (2000): 692–97.
99. Young, A. M. “Infrared Temperature Measurement
Essentials: Performance Characteristics, Calibration
and Testing.” Industrial Heating LXV, no. 8 (1998):
45–60.
100. Dewitt, D. P., and Nutter, G. D., eds. Theory & Practice
of Radiation Thermometry. New York: John Wiley & Sons,
1988.
101. Patrick, D. K. “Infrared Thermal Image Recording and
Analysis for Thermal Processes.” Industrial Heating LXV,
no. 1 (1998): 33–35.
102. Lappe, V. “Installation and Maintenance of Infrared Ther-
mometers.” Industrial Heating LXIV, no. 11 (1997): 45–49.
103. Singh, P., Henneke, M., Jayakaran, J. D., Hayes, R., and
Baukal, C. E. “Heat Transfer.” In The John Zink Combustion
Handbook, edited by C. E. Baukal. Boca Raton, FL: CRC
Press, 2001.
104. Hernberg, R., Stenberg, J., and Zethraeus, B.
“Simultaneous In Situ Measurement of Temperature
and Size of Burning Char Particles in a Fluidized Bed
Furnace by Means of Fiberoptic Pyrometry.” Combustion
and Flame 95, nos. 1–2 (1993): 191–205.
105. Stephan, K. D., Pearce, J. A., Wang, L., and Ryza, E. “Pros-
pects for Industrial Remote Temperature Sensing Using
Microwave Radiometry.” Microwave Symposium Digest,
2004 IEEE MTT-S International, Vol. 2, pp. 651–54, 2004.
106. Stephan, K. D., Pearce, J. A., Wang, L., and Ryza, E.
“Cement Kiln Temperature Measurements Using
Microwave Radiometry.” IEEE MTT-S International
Symposium, pp. 151–54, 2005.
107. Baukal, C. E., Gershtein, V. Y., and Li, X., eds. Computational
Fluid Dynamics in Industrial Combustion. Boca Raton, FL:
CRC Press, 2001.
© 2011 by Taylor and Francis Group, LLC
117
6.1 Introduction
For most industrial heating applications, heat fux is one
of the most important parameters in the system design
[1]. The heat fux distribution in many furnaces is not uni-
form because of the combustion chamber geometry or
asymmetrical fring of the burners [2]. The temperatures
are generally over 1000°C (1800°F), which means that
radiation is often an important if not dominant mode of
heat transfer. The various types of fuels may impact the
heat fux in the furnace. For example, the combustion
of H
2
produces very little gaseous radiation, while the
combustion of oil or coal produces high levels of fame
radiation due to soot formation. This means that in pet-
rochemical applications where the fuel composition is
usually highly variable, fame radiation may also be
highly variable [3]. Some industrial applications, such as
regeneratively fred glass furnaces, are highly transient
in nature, which impacts the temporal heat fux in the
furnace. While many analytical and numerical models
are available to predict heat fux, actual measurements
are needed to validate those predictions. Measuring the
heat fux in an industrial combustion process is both
important and necessary.
Heat fux to the load is closely related to the material
processing rate, which is directly related to the fring rate
(see Figure 6.1) [4]. Colannino [5] showed that heat fux
can be correlated with furnace elevation (see Figure 6.2)
using appropriate normalization techniques. An empiri-
cal model was developed that correlates heat fux as a
function of fuel pressure, air preheat temperature, fur-
nace bridgewall temperature, excess O
2
in the exhaust,
fring rate, H
2
content in the fuel, furnace height, furnace
width, heat absorption surface area, fuel port geometry,
burner tile (quarl) geometry, amount of fue gas recir-
culation, number of fuel stages, and burner type. This
data is critical for designing ethylene cracking furnaces,
which illustrates the importance of measuring heat fux
in industrial combustion applications.
6
Heat Flux
Charles E. Baukal, Jr.
CONTENTS
6.1 Introduction ....................................................................................................................................................................117
6.2 Total Heat Flux .............................................................................................................................................................. 120
6.2.1 Steady-State Uncooled Solids .......................................................................................................................... 120
6.2.2 Steady-State Cooled Solids .............................................................................................................................. 120
6.2.2.1 Single Cooling Circuit ....................................................................................................................... 121
6.2.2.2 Multiple Cooling Circuits ................................................................................................................. 121
6.2.2.3 Surface Probe ...................................................................................................................................... 121
6.2.3 Steady-State Cooled Gauges ............................................................................................................................ 122
6.2.3.1 Gradient through Thin Solid Rod.................................................................................................... 122
6.2.3.2 Thin Disk Calorimeter ...................................................................................................................... 123
6.2.3.3 Heat Flux Transducer ........................................................................................................................ 124
6.2.4 Transient Uncooled Targets ............................................................................................................................. 124
6.2.5 Transient Uncooled Gauges ............................................................................................................................. 124
6.2.5.1 Slug Calorimeter................................................................................................................................. 125
6.2.5.2 Heat Flux Transducer ........................................................................................................................ 125
6.3 Radiant Heat Flux ......................................................................................................................................................... 126
6.3.1 Heat Flux Gauge ................................................................................................................................................ 126
6.3.2 Ellipsoidal Radiometer ..................................................................................................................................... 130
6.3.3 Spectral Radiometer ......................................................................................................................................... 131
6.3.4 Other Techniques .............................................................................................................................................. 132
6.4 Convective Heat Flux ................................................................................................................................................... 136
References ................................................................................................................................................................................ 136
© 2011 by Taylor and Francis Group, LLC
118 Industrial Combustion Testing
Some examples of heat fux measurements in indus-
trial combustion processes will serve to illustrate the
various methods that have been used that are described
in some detail in this chapter. Hayes et al. [6] describe
heat fux measurements for full-scale burners fred
vertically upward along a radiant wall in pilot scale
furnaces that simulate ethylene cracking furnaces.
A water-cooled heat fux probe was inserted into the
furnace at different vertical locations, at the centerline
of the wall closest to the water-cooled tubes, between
the central pair of tubes. The probe had an air-cooled
window on the front to both protect the heat fux sensor
from contamination and to exclude convective heat fux.
Figure 6.3 shows heat fux profles as a function of fuel
composition for a natural draft burner mounted in the
foor of the furnace and fring vertically upward along
a wall at 3.5 × 10
6
(1.0 MW) Btu/hr with 2% O
2
(dry) in
the exhaust products. Analysis shows that the effect on
the measured radiant heat fux profle is relatively small
due to small variances in fuel compositions tested.
Figure 6.4 shows the effect of the fring rate on heat fux
profles for a fuel blend with 78% natural gas (Tulsa) and
22% H
2
, with 2% O
2
(dry) in the exhaust. All of the heat
fuxes have been normalized to the maximum heat fux
that occurred at 8.1 × 10
6
Btu/hr (2.4 MW) at an eleva-
tion slightly over 15 ft (4.6 m). As expected, there was
a signifcant increase in the overall measured incident
radiant heat fux at higher fring rates. Figure 6.5 shows
the heat fux profles as a function of excess O
2
for a fuel
blend with 83% H
2
and 17% natural gas (Tulsa). Over the
entire furnace, the higher excess O
2
case produced lower
heat fux compared to the lower excess O
2
case. This is
as expected since higher levels of excess air absorb addi-
tional heat, which reduces temperatures that decrease
the overall radiant heat transfer from the hot gases and
surfaces in the furnace.
Heat release (MMBtuh)
P
e
a
k

h
e
a
t

f
l
u
x

(
B
t
u
/
h
r
-
f
t
2
)
0 5 10 15
9
8
7
6
5
4
3
2
1
0
Figure 6.1
Heat fux versus heat release for a typical burner used in ethylene fur-
naces. (From Colannino, J., Mathematical Models for Characterizing
and Predicting Heat Flux Profles from Ethylene Cracking Units,
Proceedings of the 2007 AIChE Spring National Meeting, Houston,
TX, April 23–26, 2007.)
1.0
0.5
0.0
–0.5
–1.0
–1.5
–1.5 –1.0 –0.5
Normalized elevation
N
o
r
m
a
l
i
z
e
d

h
e
a
t

f
l
u
x
0.0 0.5 1.0
Figure 6.2
Normalized heat fux versus elevation for nearly 1000 measure-
ments in a pilot-scale test furnace. (From Colannino, J., Modeling of
Combustion Systems: A Practical Approach, Boca Raton, FL, CRC Press,
2006.)
75% NG, 25% H
2
25% NG, 25% H
2
% of maximum heat flux
30
25
20
15
10
5
0
65 70 75 80 85 90 95 100
F
u
r
n
a
c
e

e
l
e
v
a
t
i
o
n

(
f
t
)
Figure 6.3
Effect of fuel composition on measured heat fux profles in a furnace
for a burner fring at 3.5 × 10
6
Btu/hr. (From Hayes, R., Singh, P., Foote,
D., and Baukal, C. E., Heat Flux from Process Burners, Proceedings of
the ASME IMECE, New York, November 11–16, 2001, Heat Transfer
Division, Vol. 369, No. 4, pp. 161–64, 2001.)
© 2011 by Taylor and Francis Group, LLC
Heat Flux 119
Rafdi and Blasiak [7] measured local heat fux in a
test furnace using four high temperature air combus-
tion (HiTAC) regenerative burners (see Chapter 21). An
ellipsoidal radiometer was used to measure irradiation
(radiation heat fux incident per unit area falling on a
surface). A plug type total heat fux meter was used to
measure the rate of total heat fow (convection plus radi-
ation). The convection heat fux was calculated by using
an energy balance on the total heat fux measurements:
q
T
 = q
r
 + q
c
 = (α
s
G
r
− εσT
s
4
) + h
c
(T
g
− T
s
), (6.1)
where q
T
is the total heat fux received at the meter, q
r

is the net radiative heat fux, q
c
is the net convective
heat fux, α
s
is the absorptivity of the plug surface, G
r

is the measured irradiance, ε is the emissivity of the
plug surface, σ is the Stefan–Boltzmann constant, T
g

is the measured gas temperature, h
c
is the unknown
convection coeffcient, and T
s
is the calculated surface
temperature of the plug. The convection coeffcient
is the only unknown, which can be calculated using
Equation 6.1.
Hayes et al. [8] describe making radiation heat fux
measurements in a production regenerative, fat-glass
furnace. The measurements were taken at the crown
(roof) during both the fring and nonfring cycles, using
both a hemispherical ellipsoidal radiometer and a cir-
cular foil heat fux gauge radiometer. Measured peak
heat fuxes increased from 425 up to 710 kW/m
2
dur-
ing the fring cycle. These measurements showed the
importance of transient conditions in certain applica-
tions. These measurements and some related stud-
ies [9,10] are discussed in more detail in Chapter 32 of
this book.
There are usually two types of heat fux that are
important in industrial furnaces: radiative and total,
which is the combination of radiation and convection.
In some high temperature, low gas velocity combustors,
the total and radiative fuxes may be nearly the same
because of the predominance of radiation over convec-
tion. In lower temperature, high gas velocity combus-
tors, the radiative fux may be much lower than the total
fux where radiation and convection may be compa-
rable in those applications. However, if any heat fux is
measured, it is rare that more than one type would be
measured. This type of information can be very valuable
in understanding the dynamics of a given combustion
system and is useful for making design improvements.
Total, radiative, and convective heat fux measure-
ments are discussed here. Hoogendoorn, Ballintijn, and
Dorresteijn [11] describe the development and applica-
tion of a heat fow meter capable of separately deter-
mining radiation and convection. The probe was used
in a large-scale furnace fred with either gas or oil. The
results showed a signifcant difference in the vertical
heat fux profle with these two types of fuels where the
profle was more uniform with gas and had a higher
peak closer to the burners with oil.
Goldstein and Chiang [12], Arai, Matsunami, and
Churchill [1], and Okoh and Brown [13] discuss the
theory of operation of the heat fux techniques discussed
here. Bachmann, Chambers, and Giedt [14] experimen-
tally studied how the installation of heat fux sensors
embedded in a surface affects the actual heat transfer
in the body. They concluded that proper sensor design
can minimize the heat fux measurement errors caused
by probe thermal resistance that can disturb the heat
55
40
35
30
25
20
15
10
F
u
r
n
a
c
e

e
l
e
v
a
t
i
o
n

(
f
t
)
5
0
60 65
10.2 × 10
6
Btu/hr
11.8 × 10
6
Btu/hr
70 75
% of maximum heat flux
80 85 90 95 100
Figure 6.4
Effect of fring rate on measured heat fux profles in a furnace.
(From Hayes, R., Singh, P., Foote, D., and Baukal, C. E., Heat Flux
from Process Burners, Proceedings of the ASME IMECE, New York,
November 11–16, 2001, Heat Transfer Division, Vol. 369, No. 4,
pp. 161–64, 2001.)
40
35
30
25
20
15
10
5
0
50 55 60
2% O
2
4% O
2
65 70 75 80 85 90 95 100
% of maximum heat flux
F
u
r
n
a
c
e

e
l
e
v
a
t
i
o
n

(
f
t
)
Figure 6.5
Effect of excess O
2
on measured heat fux profles in a furnace.
(From Hayes, R., Singh, P., Foote, D., and Baukal, C. E., Heat Flux
from Process Burners, Proceedings of the ASME IMECE, New York,
November 11–16, 2001, Heat Transfer Division, Vol. 369, No. 4,
pp. 161–64, 2001.)
© 2011 by Taylor and Francis Group, LLC
120 Industrial Combustion Testing
transfer through the surface. Beasley and Figliola [15]
mathematically analyzed heat fux probes embedded in
surfaces for a range of probe and wall materials to study
probe sensitivity and design. Ducharme, Frim, and
Tikuisis [16] analytically and experimentally studied the
measurement errors caused by the thermal resistance of
the heat fux sensors and showed the errors can be very
large in certain circumstances.
6.2 Total Heat Flux
While total heat fux includes conduction, convection,
and radiation, in most industrial applications the total
heat fux is primarily radiation plus convection. Total
heat fux has been measured using both steady-state
and transient methods. Three different steady-state
methods have been used. The frst is to calculate the fux
from an energy balance on an uncooled solid. The sec-
ond is to measure the sensible energy gain of the cool-
ant, for a cooled solid. The third is to directly measure
the fux with a calibrated heat fux gauge imbedded in
a cooled target.
Two different transient methods have been used. In
the frst method, an uncooled solid is inserted into the
fame for either a given amount of time or until a certain
temperature is reached. The fux is calculated from the
sensible energy gain of the solid. In the second method,
the fux is measured using a gauge imbedded in the sur-
face of an uncooled solid. Both of these methods have
very short testing times. Both steady-state and transient
methods are discussed next.
6.2.1 Steady-State uncooled Solids
In one version of this method, the heat fux is deter-
mined from an energy balance on the solid. It was
used in two early fame impingement studies [17,18].
Refractory cylinders were coated with different oxides.
These were heated until steady-state conditions were
reached. Radiation from the cylinders, along with the
gas and cylinder surface temperatures, were measured.
The total heat fux to the cylinder was then calculated
from an energy balance on the cylinder surface. The
heat gained by the cylinder was assumed to be by con-
vection and radiation from the fame. The energy lost
by the cylinder was assumed to be solely due to radia-
tion. A thermopile was used to measure radiation from
the fame, both with and without the cylinder present.
The surface radiation from the cylinder was calculated
by subtracting the frst measurement from the second.
This value was assumed to be equal to the total heat fux
from the fame to the cylinder.
Another version of this technique is to measure the
temperature gradient through a surface and use the
steady-state, one-dimensional conduction equation to
calculate the heat fux:
′′ =

q k
t t
l
h l
, (6.2)
where
q” = heat fux per unit area (kW/m
2
),
k = thermal conductivity (W/m-K),
t
h
 = high temperature (K),
t
l
 = low temperature (K),
and l = distance between t
h
and t
l
.
This equation assumes the thermal conductivity of the
wall is constant. By measuring the temperature at two
locations in the material and measuring the distance
between the locations, the total heat fux can be simply
calculated. It is recommended that a third temperature be
measured in between to ensure that steady-state condi-
tions have been achieved. Special thermocouple assem-
blies can be made that have multiple thermocouples in
a single sheath that can be used to measure multiple
temperatures at different locations through a material.
The thermocouples can also be of different types. For
example, a type R or S can be located in the hottest part
and a type K or T could be located in the coolest part
(see Chapter 5). One specifc use of this arrangement is
to measure the heat loss through a refractory wall in a
furnace, where a single hole can be drilled through the
wall, rather than multiple holes.
This technique is very simple and of low cost.
However, only the average heat fux over the entire
solid is determined. It also relies on an accurate value
of the target emissivity. It has only limited practical rel-
evance and it is limited by the permissible maximum
material temperature level of the solid. Also, the solid
should be uniformly heated to obtain an accurate aver-
age measurement. Nonuniform heating complicates the
energy balance calculations. In Kilham’s studies [17,18],
the cylinders were rotated to minimize surface temper-
ature gradients. However, only a single thermocouple,
mounted on the inside diameter of the hollow cylinder
to measure the inside cylinder temperature, was used to
calculate the surface temperature.
6.2.2 Steady-State Cooled Solids
In this method, a cooled solid is used as a calorimeter.
It is commonly internally cooled with water or glycol.
The fow rate and the temperature rise of the coolant
are measured after steady-state conditions have been
reached. The heat fux to the solid is calculated from
© 2011 by Taylor and Francis Group, LLC
Heat Flux 121
the rate of sensible energy gain of the coolant. The three
arrangements that have been used are: a single cooling
circuit, multiple cooling circuits, and a surface probe.
6.2.2.1 Single Cooling Circuit
In this technique, the average heat fux over the entire
solid surface is calculated from the sensible energy
gain of a single cooling circuit. Davies [19] and Fells
and Harker [20] studied the effects of electrical boost on
enhancing heat fux from impinging fames boosted by
an AC discharge that produced as much as 3.6 times
more heat fux than unboosted fames. The fames
impinged on water-cooled stainless steel and copper
pipes. Vizioz and Lowes [21] studied large-scale fames
impinging on a large, circular, water-cooled plate with
a refractory surface to simulate a hot target. The surface
temperature was maintained using an imbedded spiral
water-cooled channel. Davies [22] used water-cooled
pipes to study heat fux from impinging fames on a
cylinder as a function of equivalence ratio, fring rate,
oxidizer/fuel mixing, and oxygen enrichment ratio.
Posillico [23] studied fames impinging on a fat plate
with a single internal cooling passage.
6.2.2.2 Multiple Cooling Circuits
This technique has been used for fames impinging
normal to the surface of a large metal disk, located
inside a furnace. The surface temperature of the disk
was maintained with concentric cooling channels. The
heat fux to each annulus of the target was calculated
from the sensible energy gain for each of the indi-
vidual cooling circuits. Vizioz and Lowes [21] used a
steel plate to simulate a cold target. The surface tem-
perature was maintained below 373 K (212°F), with
eight equally spaced concentric cooling channels. The
temperature difference between each water inlet and
outlet was only about 20–25 K (36–45°F) to reduce the
radial heat transfer between the adjacent channels.
Smith and Lowes [24] used a similar plate design. Air
gaps between the channels were provided to further
reduce radial conduction. Experiments included both
air and water cooling. Rajani, Payne, and Michelfelder
[25] used a smaller version of this plate. Baukal and
Gebhart (1998) used a disk segmented into six con-
centric sections, as shown in Figure 6.6 [26]. The space
between sections was insulated to minimize radial heat
transfer between sections. Figure 6.7 shows contours of
heat fux for impinging oxygen-natural gas fames as
functions of fring rate and position (L = normalized
axial distance from the burner exit and R = normal-
ized radial distance from the centerline). As expected,
the highest measured fuxes were at the highest fring
rate, at the centerline, and close to the burner outlet.
Figure 6.8 shows heat fux as a function of fring rate
and axial distance from the burner. As expected, the
highest fuxes were measured at the highest fring rate
for the shortest distance from the burner.
6.2.2.3 Surface Probe
The American Society for Testing and Materials (ASTM)
standard E422 provides details for the construction of a
single circuit, water-cooled calorimeter probe to measure
heat fux [27]. Anderson and Stresino [28], Reed [29], and
Fay [30] studied fames impinging on a so-called sur-
face probe. In those studies, two identical, water-cooled
copper plates were connected in series and separated by
a thermal barrier. A single water-cooling circuit passed
across both sections. The fame was traversed from the
frst to the second section. The water temperature rise
through each plate was measured as a function of dis-
tance from the fame centerline. The data was inverted
to determine the spatial heat fux distribution. See Reed
[29] for details of the inversion calculation.
This method is simple and of relatively low cost. It
needs no calibration. The heat fux through the sur-
face is measured directly by the sensible energy gain
of the internal coolant. It has high accuracy. The spatial
resolution depends on the number of separate cooling
Figure 6.6
Circular target (before insulation) used in fame impingement stud-
ies. (From Baukal, C. E., and Gebhart, B., “Surface Condition Effects
on Flame Impingement Heat Transfer,” Thermal and Fluid Science 15
(1997): 323–35.)
© 2011 by Taylor and Francis Group, LLC
122 Industrial Combustion Testing
channels used. However, fner spatial resolution requires
more controls. The choices of target material and cool-
ant are limited at higher surface temperatures.
6.2.3 Steady-State Cooled gauges
In this method, the local heat fux is determined using
a small gauge imbedded in a much larger solid. The
heat fux gauges are commonly made of a high thermal
conductivity material, such as copper. The hot end of
the gauge is exposed to the fame, while the cold end is
water-cooled. Three different variations of this method
have been used: a gradient through a thin solid rod, a
thin disk calorimeter, and a heat fux transducer.
6.2.3.1 Gradient through Thin Solid Rod
The hot end of a thin solid rod is made fush with the
target surface. Thermocouples, located along the probe
axis, are used to determine the axial internal temper-
ature gradient that is assumed to be linear. The heat
fux is calculated using a one-dimensional conduc-
tion equation. In most cases, the probe has shields to
minimize heat fux from the sides. Figure 6.9 shows
an example of a portable, water-cooled total heat fux
probe used for inserting into high temperature furnace
environments.
This type of probe, used by both Cookson and
Kilham [31] and Kilham and Dunham [32] is described
200
200
200
200
200
5 kW
10 kW
15 kW
20 kW
25 kW
200
100
100
300
300
300
300
300
300
400
400
400
400
400
400
500
500
500
600
700
400
400
400
400
400
300
300
500
500
500
500
500
600
600 700
800
500
400
400
400
500
500
500
500
500
600
600
600
600
700
700
800
900
500
500
500
500
500
400
400
600
600
600
600
600
700
700
700
800
800
900
1.0
0.8
0.6
R
e
f
f
R
e
f
f
R
e
f
f
R
e
f
f
R
e
f
f
0.4
0.2
1.0
0.8
0.6
0.4
0.2
1.0
0.8
0.6
0.4
0.2
1.0
0.8
0.6
0.4
0.2
1.0
0.8
0.6
0.4
0.2
0 1 2 3
L
4 5 6
300
300
Figure 6.7
Contours of the total heat fux, q” (kW/m
2
), from oxygen-natural gas fames with various fring rates (q
f
 = 5 to 25 kW) impinging on the surface
of an untreated stainless surface. (From Baukal, C. E., and Gebhart, B., Experimental Thermal and Fluid Science, 16, no. 3, 247–59, 1998.)
© 2011 by Taylor and Francis Group, LLC
Heat Flux 123
in detail by Cookson, Dunham, and Kilham [33]. It was
a 4.8 mm (0.19 in.) o.d. copper rod, imbedded in a larger
diameter cooled hemi-nosed brass cylinder. Beér and
Chigier [34] used a 2.5 cm (1.0 in.) o.d. uncoated stainless
steel probe. Buhr, Haupt, and Kremer [35] and Kremer,
Buhr, and Haupt [36] used a 3 mm (0.12 in.) o.d. copper
rod imbedded in an internally water-cooled fat plate.
Hoogendoorn, Popiel, and van der Meer [37] and Popiel
van der Meer, and Hoogendoorn [38] used a 5.85 mm
(0.23 in.) o.d. copper rod imbedded in a water-cooled
fat plate. Hargrave and Kilham [39] and Hargrave,
Fairweather, and Kilham [40] used a 3.2 mm (0.125 in.)
o.d. copper rod internally cooled by glycol. The rod was
imbedded inside a 22 mm o.d. hemi-nosed cylinder.
Kataoka, Shundoh, and Matsuo [41] used a 10.21 mm
(0.402 in.) o.d. copper rod imbedded in a water-cooled
fat plate.
Cassiano, Heitor, and Silva [42] used a probe of this
type to measure the total heat fux in an end-port regen-
erative glass furnace. The probe consisted of cylindrical
block inside a water-cooled housing that was required
because of the high temperatures in the furnace. The
front surface was serrated and blackened to maximize
the radiant absorptivity. The probe was calibrated using
a blackbody furnace. The response time of the probe was
~10 minutes with an accuracy of approximately 5%.
6.2.3.2 Thin Disk Calorimeter
In this variation, a thin water-cooled disk, usually cop-
per, is imbedded in a much larger target body. The
heat fux to the calorimeter is calculated using the fow
rate and the temperature rise of the water. Shorin and
Pechurkin [43] used a 20 mm (0.8 in.) o.d. copper disk
imbedded in a larger fat plate. The gauge and target
were both independently water-cooled.
L = 0.5
1000
900
800
700
600
500
400
300
200
100
5 10 15 20 25
q
f
(kW)
q
"

(
k
W
/
m
2
)
L = 1.0
L = 2.0
L = 4.0
L = 6.0
Figure 6.8
Total heat fux (q”) from oxygen-natural gas fames of various fring
rates (q
f
) impinging on the stagnation point (R
eff
 = 0.16) of an untreated
stainless surface. (From Baukal, C. E., and Gebhart, B., Experimental
Thermal and Fluid Science, 16, no. 3, 247–59, 1998.)
Water
outlet
Water
inlet
Water jacket
Operating length
Cooling water
Cooling water
Thermocouples
Guard
rings
Stainless
steel plug
Figure 6.9
Water-cooled total heat fux probe. (From Baukal, C. E., The John Zink Combustion Handbook, Boca Raton, FL: CRC Press, 2001.)
© 2011 by Taylor and Francis Group, LLC
124 Industrial Combustion Testing
This technique is relatively low cost and needs no
calibration. Accuracy depends on at least two tempera-
ture measurements and a coolant fow measurement.
Response time may be slow depending on the design.
6.2.3.3 Heat Flux Transducer
In this technique, an electrical signal is generated that
is proportional to the heat fux. The transducer is often
imbedded fush with a solid surface, such as the target
in fame impingement heating or a furnace wall. One
type of transducer uses an array of thermocouples to
measure the heat fux across a thin distance. Schulte
and Kohl [44] describe the water-cooled heat fux trans-
ducer used in Schulte’s [45] study. It was designed for
surface fux rates up to 1000 kW/m
2
(300,000 Btu/hr-ft
2
),
using a single germanium crystal as the sensing ele-
ment. Smith and Lowes [24] used a stainless steel disk
with 15 spaced heat fux gauges, at eight equidistant
radial locations. You [46] used water-cooled transducers
whose surfaces were coated with either a gold or black
foil, to determine either the convective or total heat fux,
respectively. The transducers were imbedded fush
with the surface of a water-cooled plate. Hager et al. [47]
describe a thin flm heat fux probe that can be attached
to a surface, which minimizes the error caused by the
thermal resistance of the sensor disturbing the heat fux
through the surface.
Another type of transducer is known as the Gardon
[48] gauge. The basic confguration consists of a thin
metal disk. Thermocouples are connected on the back of
the disk, at the center, and at the perimeter. When heat
fux is received at the front of the disk, a linear voltage is
produced by the radial conduction from the center of the
disk to the perimeter. This voltage is proportional to the
heat fux. This gauge was originally developed for mea-
suring high radiant fuxes. Its use has been extended to
mixed convection and to radiation environments, using
proper corrections. Grey [49] gives details on the con-
struction of these probes. Ash [50] discusses the theory of
the fast response of Gardon gauges. Borell and Diller [51]
describe an apparatus for calibrating Gardon gauges in
convective air fows. Van der Meer [52] used this type of
gauge, imbedded fush with the surface of a water-cooled
fat plate.
The technique is simple and of relatively low cost. It has
good accuracy and spatial resolution. The response time
is fast enough that it can be used for feedback control
systems [53]. There are some potential concerns with heat
fux gauges. Calibration is required. This may be compli-
cated in a mixed radiation and convection environment,
since calibration typically requires a blackbody source.
The maximum allowable temperature and heat fux, for
some of the commercial transducers, may limit their use,
for example, in high intensity fame impingement.
6.2.4 Transient uncooled Targets
In this technique, the transient heat fux to a relatively
large uncooled body is determined indirectly. The tem-
perature profles in the target are measured by an array
of imbedded thermocouples, usually at or near the
impingement surface. The fux is then calculated from
the measured temperature responses, using inverse
conduction heat transfer computational techniques
[54]. Giedt, Cobb, and Russ [55] measured the transient
heating in an ingot iron plate, whose surface was par-
allel to the fame. Measurements were made until the
plate reached 1170 K (1650°F). This usually took about
21 seconds. Woodruff and Giedt [56] used a blunt-nosed
molybdenum plate to simulate an air foil. Test durations
were about 50 seconds. Milson and Chigier [57] used a
fat, uncooled plate. The back of the plate was coated with
lamp black, to create a surface boundary condition hav-
ing a well-defned emissivity. The heat fow through the
plate was equated to the natural convection and radia-
tion losses from the back side. Veldman, Kubota, and
Zukoski [58] used a thin steel disk target with an array of
thermocouples welded onto the back side. Matsuo et al.
[59] heated rectangular slabs of varying thicknesses,
located inside a furnace. Rauenzahn [60] used a thick
copper plate, which remained nearly isothermal dur-
ing the heating process. Gatowski, Smith, and Alkidas
[61] describe the use of a thermocouple embedded in the
wall of an internal combustion engine to measure the
heat fux to the wall by using the transient surface tem-
perature measurements.
This technique is very simple and of low cost. It does
not need calibration. It commonly suffers from poor
spatial resolution. It requires very rapid instrument
response times. This may cause diffculties if many mea-
surements must be frequently recorded in a very short
time period. This technique also relies on accurate ther-
mal conductivity values, as a function of temperature.
6.2.5 Transient uncooled gauges
In this technique, a small heat fux gauge is imbedded
fush with the target. The gauge is commonly made of a
high thermal conductivity material. It is assumed to be at
a uniform temperature level at all times. The sensible heat
gained by the gauge may be simply calculated. Hornbaker
and Rall [62] list the ideal characteristics for this kind of
gauge in high temperature applications as:
1. the measured fux should equal that transferred
to the target
2. the fux that would have been received by the
target, with no gauge, should not be changed
3. the gauge must have a rapid temporal response
© 2011 by Taylor and Francis Group, LLC
Heat Flux 125
4. the gauge output should be directly proportional
to the fux
5. the gauge should be small, simple, and ruggedly
constructed
Bachmann, Chambers, and Giedt [14] discuss the errors
using gauges caused by (1) conduction between the
gauge and target and (2) by perturbation of the boundary
layer resulting from the temperature discontinuity at the
gauge-target interface. Bachmann used copper and nickel
slug calorimeters to measure the heat fux from O
2
/C
2
H
2

fames fowing parallel to a fat plate. The gauge heat fux
was 40–140% higher than to a calibrated reference calo-
rimeter, as a result of the second error type above. Two
different variations of this method have been used: the
slug calorimeter and heat fux transducer.
6.2.5.1 Slug Calorimeter
In this technique, the slug thermal capacity is assumed
to be internally isothermal, at any instant in time.
This occurs when the thermal conductivity is large
compared to the convection heat transfer coeffcient.
The heat fux to the slug is calculated from the sen-
sible energy gain of the slug. The average fux is usu-
ally calculated over a temperature range during which
the energy gain was linear. Kilham and Purvis [63],
Nawaz [64], Fairweather, Kilham, and Nawaz [65], and
Fairweather, Kilham, and Mohebi-Ashtiani [66] used a
2.18 mm (0.086 in.) o.d. stainless steel conductivity plug,
located inside a hemi-nosed cylinder. The target was
inserted into the fame using a solenoid coil. The inser-
tion time was estimated by Nawaz to be 0.04 seconds.
This was compared to the residence time in the fame,
which was estimated to be 0.15–0.25 seconds. Nawaz
used a hollow stainless steel cylinder. The conductiv-
ity plug was a 0.26 mm thick plate, used to close one
end of the cylinder. A thermocouple was welded on the
inside surface of the plug, to measure the energy gain.
The experimental results of Fairweather, Kilham, and
Mohebi-Ashtiani [66] showed large heat fuxes near the
laminar fame reaction zone due to the large and non-
equilibrium concentration of reactive species. Lower
peaks in fux were measured for turbulent fames. This
was due to the more diffuse mixing in the reaction zone.
Conolly and Davies [67] used an assembly of a cylindri-
cal plug (made of a copper/1% beryllium alloy) and a
rear-mounted thermocouple. Kremer, Buhr, and Haupt

[36] used a 3 mm o.d. by 1.5 mm (0.059 in.) thick copper
disk. The disk was insulated from the rest of the target
with a ceramic, to reduce radial losses. It was mounted
fush to the surface of a fat plate. Kilham and Purvis
[68] used 12 mm (0.47 in.) o.d. high thermal conductiv-
ity disks, with thicknesses of 1, 2, or 3 mm (0.04, 0.08
or 0.12 in.). These were mounted fush with the surface
of a large fat plate. They were made of either silicon
carbide or platinum, to test the importance of hydro-
gen atom recombination. No difference in heat fux was
measured using the two materials. Ivernel and Vernotte
[69] used 15 mm (0.59 in.) o.d. by 2 mm (0.08 in.) thick
platinum disks. The temperature rise of the disk was
measured with a thermocouple attached to the back.
This rise was used to calculate the total heat fux, based
on the sensible energy gain of the disk. Horsley, Purvis,
and Tariq [70] and Hemeson et al. [71] used a calorim-
eter consisting of a small oxygen-free high conductiv-
ity (OFHC) copper disk with a thermocouple attached
on the back face. Horsley estimated that the exposure
time for the probe varied from 0.32 to 2.0 seconds. The
mean surface temperature was 360 K (190°F). The sen-
sor probe was located in a cylinder that was inserted
by a piston normal to the surface of a water-cooled
plate. Hemeson used a 4.44 mm (0.175 in.) o.d. by 1.57
mm thick (0.0618 in.) copper disk imbedded fush and
contoured to the surface of both cylinders and hemi-
spheres. Hargrave, Fairweather, and Kilham [40] used a
3.2 mm (0.125 in.) o.d. by 1.0 mm (0.04 in.) thick copper
slug, located at the stagnation point of a solid brass cyl-
inder. Hustad, Røkke, and Sønju [72], Hustad and Sønju
[73], and Hustad, Jacobsen, and Sønju [74] used a probe
consisting of a 25 mm (1 in.) o.d. by 3 mm (0.12 in.) thick
fat copper disk. It was imbedded fush with the surface
of a round steel pipe, at the forward stagnation point of
the fow.
This technique is relatively simple in principle and
needs no calibration. However, fast-acting insertion
equipment and accurate temporal temperature mea-
surements and material properties are needed for this
technique.
6.2.5.2 Heat Flux Transducer
In this method, an uncooled transducer is used to mea-
sure heat fux. Hornbaker and Rall [62] measured the heat
fux from O
2
/C
2
H
2
fames, parallel to a fat plate, using
both the transient uncooled target and the uncooled
gauge techniques. The frst technique was assumed to
indicate the actual heat fux. Comparison with the sec-
ond technique showed that the gauges overpredicted,
matched, and underpredicted the true fux, respectively,
at surface temperatures of about 300–550, 550–600, and
600 K–900 K (80–530, 530–620, 620°F –1200°F).
The transient uncooled gauge technique is relatively
simple. It simulates many industrial heating processes,
such as rapid metal heating. However, the instruments
must have very fast response times. Some mechanism
must be provided for rapidly inserting the target into the
fame. For high temperature fames, the response time
must be fractions of a second, to prevent the transducer
from melting.
© 2011 by Taylor and Francis Group, LLC
126 Industrial Combustion Testing
6.3 Radiant Heat Flux
Gray and Müller [75] note there are four components for
most instruments used to measure radiation: (1) a detec-
tor that converts the radiation to an electrical signal, (2)
an optical system that directs the radiation to the detec-
tor that may include some type of focusing element, (3) a
fltering system that ensures only the proper wavelength
band of radiation required for the detector is received by
the detector, and (4) an amplifer to enhance the signal and
some type of recorder/indicator to display the output. In
the following sections, only components one and three
are considered. It has been assumed the optical system
and amplifer can be properly designed for a given detec-
tor. Note that radiation from fares is covered in Chapter
30 of this book and is therefore not included here.
Many of the devices used to measure radiation have a
highly absorptive black coating. This coating is usually
very thin to minimize its effect on reducing the conduction
through the coating that would affect the measured heat
fux. One of the challenges in calibrating these absorptive
sensors is determining the thickness of the blackened
coatings. Stradomskiy, Maksimov, and Malyarov [76]
describe a method for determining the coating thickness
for these probes so they can be properly calibrated.
6.3.1 Heat Flux gauge
There are three general types of radiant emission that
may arise in combustion processes. One is nonluminous
gaseous radiation (see Figure 6.10). This results from the
well-characterized spectral emission characteristics of
both H
2
O (see Figure 6.11) and CO
2
(see Figure 6.12) [77].
These are two of the main constituents in the products
of combustion, for hydrocarbon fuels. Even at high fame
temperatures, nonluminous radiation is usually small
compared to convection, since the emissivity of the gases
is very low [78]. The second type of radiation involves
luminous emission from particles. Such sooting fames
(see Figure 6.13) often generate very high radiation fuxes.
This effect is usually only important for liquid and solid
fuels, but not for gases that normally produce very little
if any soot unless the fames are very fuel rich. The third
effect is radiation from solid surfaces, such as from refrac-
tory walls inside a furnace (see Figure 6.14). Such radia-
tion depends on the geometry, temperature, and radiant
properties of the walls. In some studies, several of these
types of radiation have been important. The techniques
used to measure thermal radiation are discussed below.
One method that has been used to measure nonlu-
minous radiation is to directly measure the radiation
from free jet fames with a radiometer. An elevation and
plan view of a radiometer aimed at a fame are shown in
Figure 6.15 and Figure 6.16, respectively. In one study, a
Medtherm Corp. (Huntsville, AL) model 40-5-18T heat
fux transducer was used to measure the radiation [79].
The transducer was a Gardon gauge [48] with a diam-
eter of 6.35 mm (0.25 in.). The sensor absorptance was
92%, in the spectral range of 0.6 to 15.0 µm. A model
40VRW-7C water-cooled view restrictor was used to
restrict the feld of view to θ
radm
 = 6.45°. A narrow angle
radiometer was used to ensure that only radiation from
the fame, not from the ambient environment, was
measured. This was representative of the case for an
impinging fame, where the target was engulfed by the
hot combustion gases. The radiometer did not have any
windows or gas purging, since the ambient environ-
ment was clean. Figure 6.17 shows the measured fame
radiation as a function of axial distance from the burner
outlet and fring rate. The graph shows that radiation
initially increases with distance and then decreases at
differing rates depending on the fring rate Figure 6.18
shows the radiation from a pure O
2
/natural gas fame
was approximately double that of an oxygen-enriched
air/natural gas fame.
Gore et al. [80] used a Medtherm wide angle (150°)
radiometer to measure luminous radiation from meth-
ane/heptane and methane/crude oil fames. The mea-
sured radiant fuxes for the methane/heptane fames
Figure 6.10
(See color insert following page 424.) Example of a nonluminous
fame. (Courtesy of John Zink Co. LLC.)
© 2011 by Taylor and Francis Group, LLC
Heat Flux 127
and for the methane/crude oil fames ranged from 0.34
to 0.64 and 0.53 to 0.68 W/cm
2
(1100–2000 and 1700–2200
Btu/hr-ft
2
). The modeling results signifcantly under-
predicted the experimental data due to lower estimates
of temperature, the complexity of predicting soot in tur-
bulent fow, and turbulent radiation interactions.
Shorin and Pechurkin [43] used a steady-state cooled
gauge to measure the total heat fux from impinging
fames. The gauge was imbedded fush with the surface
of a large plane surface. To determine the importance
of radiation, gauges with polished, oxidized, and black-
ened surfaces were tested. The lowest fux was expected
to arise for the polished gauge, which refects most of
the radiation. Higher fuxes were expected with the oxi-
dized and blackened gauges. These are better absorbers.
However, no difference in measured heat fux was found.
It was concluded that radiation was not important for
their experimental conditions. You [46] measured fame
radiation to the surroundings, using commercial heat
fux gauges. The radiant fux to a plane surface was also
indirectly measured. For these measurements, commer-
cial heat fux gauges were coated with either gold or black
foil, to measure the convective or total heat fux, respec-
tively. The radiant fux was determined by subtracting
the convective fux from the total. Thermal radiation was
13–26% of the total fux, near the stagnation point. It was
negligible at the edge of the plate (R = 7.3).
Heat fux gauges are often used as radiometers by put-
ting a glass window in front of them to prevent convec-
tive heat transfer to the gauge. A variety of materials may
be used for the infrared detectors, each having a specifc
spectral characteristic as shown in Figure 6.19 [81].
L = 2000 cm
L = 200 cm
L = 20 cm
L = 2 cm
L = 20 cm
L = 2 cm
L = 2000 cm
L = 200 cm
L = 2 cm
L = 2000 cm
L = 200 cm
L = 20 cm
600 K
1200 K
2400 K
10000 9000 8000 7000 6000 5000
Wavenumber (cm
–1
)
A
b
s
o
r
p
t
i
v
i
t
y
R
e
l
a
t
i
v
e

b
l
a
c
k
b
o
d
y

r
a
d
i
a
n
c
e
4000 3000 2000 1000
1
0
1
0
1
0
1
0
T = 600 K
p = 1 atm
p
H
2
O
= 0.5 atm
T = 1200 K
p = 1 atm
p
H
2
O
= 0.5 atm
T = 2400 K
p = 1 atm
p
H
2
O
= 0.5 atm
Figure 6.11
Calculated spectral emissivity of H
2
O as a function of wavelength and path length. (From Ludwig, C. B., Malkmus, W., Reardon, J. E., and
Thomson, J. A. L., Handbook of Infrared Radiation from Combustion Gases, Washington, DC: National Aeronautics and Space Administration
report SP-3080, 1973.)
© 2011 by Taylor and Francis Group, LLC
128 Industrial Combustion Testing
Matthews, Harris, and Garcia [82] describe a multi-
head transpiration radiometer used to make radiative
heat fux measurements in a sooty pool fre. The transpi-
ration radiometer [83] uses a transducer to measure the
radiation and the transpiration is used to eliminate the
effects of convection. The multihead version measured
radiation from fve different directions in the fame and
worked successfully in a sooty environment. The main
drawback was insensitivity at low fux rates.
Brajuskovic, Matovic, and Afgan [84] and Brajuskovic,
and Afgan [85] describe a blow-off heat fux sensor that
uses gas fowing through a porous sintered metal disc
to measure hemispherical radiation heat fux in a dirty
environment that includes high particulate loads. The
gas acts to blow off the boundary layer at the surface. It
also cools the instrument, prevents fouling of the porous
surface that can particularly be a problem when gaseous
or liquid fuels are used that can produce soot, and is
used to distinguish between convection and radiation
heat fux. Martins et al. [86,87] describe the calibration
of such a blow-off heat fux sensor.
Ploteau, Glouannec, and Noel (2007) developed a ther-
moelectric heat fux meter for use in measuring radiation
in industrial furnaces [88]. Three versions were devel-
oped: water-cooled for a constant reference temperature,
uncooled with a paraffn buffer to maintain a constant
reference temperature for lower temperature environ-
ments, and foating reference temperature with two
thermoelectric modules. The cooled version can handle
the highest temperatures.
Heat fux gauges need to be properly calibrated
to ensure the measurements are accurate. Gauges
L

=

2
0
0
0

c
m
L = 200 cm
L = 20 cm
L = 2 cm
L = 20 cm
L = 2 cm
L

=

2
0
0
0

c
m
L

=

2
0
0

c
m
L = 2 cm
L = 2 cm
L = 2000 cm
L = 200 cm
L = 20 cm
T = 2400 K
p = 1 atm
L = 2 cm
600 K
1200 K
2400 K
10000 9000 8000 7000 6000 5000
Wavenumber (cm
–1
)
A
b
s
o
r
p
t
i
v
i
t
y
R
e
l
a
t
i
v
e

b
l
a
c
k
b
o
d
y

r
a
d
i
a
n
c
e
4000 3000 2000 1000
1
0
1
0
1
0
1
0
T = 600 K
p = 1 atm
T = 1200 K
p = 1 atm
p
CO
2
= 0.5 atm
p
CO
2
= 0.5 atm
p
CO
2
= 0.5 atm
Figure 6.12
Calculated spectral emissivity of CO
2
as a function of wavelength and path length. (From Ludwig, C. B., Malkmus, W., Reardon, J. E., and
Thomson, J. A. L., Handbook of Infrared Radiation from Combustion Gases, Washington, DC: National Aeronautics and Space Administration
report SP-3080, 1973.)
© 2011 by Taylor and Francis Group, LLC
Heat Flux 129
purchased from vendors normally come with a calibra-
tion curve. It is helpful to have a blackbody calibrator
(see Figure 6.20) to ensure the calibration curve is still
valid after the gauge has been in use for some time or
something has happened to the gauge (e.g., it has been
Radiometer
Flame
Burner
lradm
Figure 6.15
Elevation view of the radiometer and the fame. (From Baukal, C. E.,
and Gebhart, B., International Journal of Heat and Mass Transfer, 40, no.
11, 2539–47, 1997.)
Radiometer
θ
radm
Flame
View restrictor
d
radm
d
f
Figure 6.16
Plan view of a radiometer and a fame. (From Baukal, C. E., and
Gebhart, B., International Journal of Heat and Mass Transfer, 40, no. 11,
2539–47, 1997.)
0
70
60
50
40 q

r
a
d

(
k
W
/
m
2
)
30
20
q
f
= 5.00 kW
q
f
= 15.0 kW
q
f
= 25.0 kW
L
radm
1 2 3 4 5 6 7
Figure 6.17
Thermal radiation (q″
rad
) versus nondimensional axial distance from
the burner exit (L
r
) and fring rate (q
f
) (5.0, 15.0, 25.0 kW) for stoichio-
metric oxygen-natural gas fames. (From Baukal, C. E., and Gebhart,
B., International Journal of Heat and Mass Transfer, 40, no. 11, 2539–47,
1997.)
Figure 6.13
(See color insert following page 424.) Example of a luminous fame.
(Courtesy of John Zink Co. LLC.)
Figure 6.14
(See color insert following page 424.) Surface radiation in a
furnace.
© 2011 by Taylor and Francis Group, LLC
130 Industrial Combustion Testing
dropped) that may affect the calibration. The calibrator
can be set to different temperatures, which means the
radiation is known because the furnace acts like a black-
body. The probe is inserted into the calibrator and the
known blackbody radiation from the furnace can be
compared against the calibration curve for the gauge.
The curve can be adjusted if necessary to match the
blackbody heat fux from the furnace. If no calibrator is
available, the gauge can be sent back to the vendor for
recalibration.
6.3.2 ellipsoidal radiometer
This probe (see Figure 6.21) consists of an ellipsoidal cav-
ity, internally coated with a highly refective metal, such
as polished gold. Radiation enters the cavity through
a small hole, located at one of the foci of the ellipsoid
cavity. The radiation is refected inside the ellipsoid,
onto a thermopile, located at the other focal point. The
thermopile converts the radiant fux to an electrical
signal. The radiometer is commonly calibrated against
a blackbody furnace (see Figure 6.20) having a known
internal surface temperature.
Several flame impingement studies ([21,25,34]) mea-
sured total radiant emission with hollow, gold-plated,
ellipsoidal radiometers. These were developed by
the International Flame Research Foundation (IFRF)
in IJmuiden, the Netherlands [89]. Measurements at
the IFRF indicated that radiation may account for as
much as half of the total heat flux to a target located
in a hot furnace. The radiation was primarily from
the hot refractory walls. Vizioz attempted to isolate
this radiation from the hot combustion gases. The
radiation reflected and emitted from a flat plate tar-
get was measured with a narrow angle radiometer.
Figure 6.20
Blackbody calibration furnace.
Ω = 0.35
0
10
20
30
q

r
a
d

(
k
W
/
m
2
)
40
50
60
70
1 2 3 4 5 6 7
L
radm
Ω = 1.00
Figure 6.18
Thermal radiation (q″
rad
) versus axial distance from the burner outlet
(L
r
) and oxidizer O
2
content of 35% and 100% (Ω = 0.35 and 1.00) for a
fring rate natural gas at 15.0 kW. (From Baukal, C. E., and Gebhart,
B., International Journal of Heat and Mass Transfer, 40, no. 11, 2539–47,
1997.)
Ge 77 K
Ge–30°C
Ge 22°C
D
*
(
λ
)
InAS–30°C
InAS 22°C
1
10
14
10
13
10
12
10
11
10
10 HgCdTe 77 K
10
9
10
8
2 3 4 5 6 7 8 10 12 20 30
HgCdTe–75°C
Wavelength (μm)
HgCdTe 77 K
Doped Ge 4 K
HgCdTe 77 K
HgCdTe–55°C
InSb 77 K
InAs 77 K
Ideal photovoltaic
(BLIP limit)
Ideal photoconductor
(BLIP limit)
Figure 6.19
Spectral characteristics of infrared detectors. (From Lide, D. R., CRC
Handbook of Chemistry and Physics, 79th ed., Boca Raton, FL: CRC
Press, 10–216, 1998.)
© 2011 by Taylor and Francis Group, LLC
Heat Flux 131
This consisted of a thermopile with a window in front
of it, to eliminate forced convection. The window and
thermopile were arranged so that only radiation from
certain angles is intercepted. The radiated flux from
the plate was assumed to be a constant fraction of the
incident radiation. The absorption and emission of
the gas layer, between the plate and the radiometer,
were estimated. Smith and Lowes [24] were unable
to obtain consistent results using this ellipsoidal
radiometer.
Butler and Webb [90] reported wall radiant heat fux
measurements, ranging from 100 to 500 kW/m
2
(30,000
to 160,000 Btu/ft
2
-hr), in an 80 MW
e
(2.7 × 10
8
Btu/hr)
industrial coal-fred boiler using an ellipsoidal radiom-
eter. Average wall radiant heat fuxes were measured at
six elevations in the boiler. In experiments located out-
doors at ambient conditions, Hustad, Røkke, and Sønju
[72] measured radiation perpendicular to the fame axis
using a wide angle radiometer. The measurements were
made at the middle of the fame height, where soot con-
centration and, therefore thermal radiation, were at their
peaks. It was found that the radiation was higher than
the model predictions. It was concluded that the mea-
surement location was not representative of the entire
fame. The calculated values were used, instead of the
actual measurements, in subsequent analyses. Hustad
[74] found that radiation perpendicular to the fame
axis was 26%, 40%, and 44% of the total heat fux for
CH
4
, C
4
H
10
, and C
3
H
8
, respectively at Ma = 1. For C
3
H
8
,
at Ma = 0.6, radiation was 26% of the total fux. These
results show that luminous radiation may be important,
even for gaseous fuels. Beltagui, Kenbar, and Maccallum
[91] used both a total heat fux probe and calorimetry to
measure the total heat fux in a furnace and a radiome-
ter to measure the radiant heat fux in the furnace. They
studied the characteristics of swirling fames fred verti-
cally into a water-cooled chamber 3 m (10 ft) long and 1
m (3 ft) i.d., divided into six sections for calorimetry. The
fring rate was 400 kW (1.4 × 10
6
Btu/hr) on natural gas
with equivalence ratios ranging from 0.70 to 1.35 and
5% excess air. The total heat fux probe measurements
were in good agreement with the calorimetry calcula-
tions for each of the six sections. The radiation measure-
ments showed that the radiant heat fux was at least half
of the total fux and in some cases nearly accounted for
all of the heat fux. The measurements also showed that
the heat fux distribution was much more uniform for
the case with no swirl and that the peak fux 70 kW/m
2

(22,000 Btu/ft
2
-hr) that occurred in the second section
downstream from the burner for the case with high swirl
(S = 2.25) was almost double the peak fux 40 kW/m
2

(13,000 Btu/ft
2
-hr) for the case without swirl.
Radiometers are generally low cost and simple to use.
They have good spatial resolution. They operate suc-
cessfully at high temperatures. However, they must be
calibrated. Ellipsoidal radiometers are diffcult to use in
very sooty or in particle-laden environments that may
foul the polished refectors.
6.3.3 Spectral radiometer
Ji and Baukal [92] studied the spectral radiation from
oxygen-enhanced fames. Spectral radiation was mea-
sured (see Figure 6.22) with a spectral radiometer from
Monolight Instruments (now known as Irradian Ltd.
in East Lothian, U.K.). Three scanning monochromator
Water
outlet
Water
inlet
Water jacket
Operating length
Cooling water
Cooling water
Constantan
wires
Nitrogen purge
Sensing
pellet
Mirrored
ellipsoidal
cavity
Figure 6.21
Drawing of an ellipsoidal radiometer. (From Baukal, C. E., The John Zink Combustion Handbook, Boca Raton, FL: CRC Press, 2001.)
© 2011 by Taylor and Francis Group, LLC
132 Industrial Combustion Testing
modules with appropriate sets of order-sorting flters
were used to cover the spectral range from 0.25 µm to
5.0 µm (see Figure 6.23). The spectral irradiance calibra-
tion was made with a quartz tungsten halogen lamp
(Oriel Model 6333, Stratford, Connecticut) in the 0.25–1.7
µm range and with an IR emitter (Oriel Model 6580) in
the 1.7–5.0 µm range. Figure 6.24 shows how the mea-
sured spectral radiation for an oxy-natural gas fame
varied with equivalence ratio. Stoichiometric fames
(ϕ = 1.00) had the most pronounced peaks. Fuel rich
fames (ϕ = 2.00) had lower peaks but more broadband
radiation. Fuel lean fames (ϕ = 0.67) were generally
in between. Figure 6.25 shows that fame radiation
increased as the O
2
content in the oxidizer increased,
where oxy-fuel produced signifcantly more radiation
than air–fuel combustion.
6.3.4 Other Techniques
In the Ji and Baukal [92] study referred to above, spectral
radiometers and the properties of molten glass were
used to determine the optimal wavelength range for
3.5 m
Spectrometers
Side view
Top view
Spectrometers
Burner
0.5 m
0.5 m
UV-VIS
NIR
MIR
Figure 6.23
Side view and top view of multispectrometer setup.
0.25
0
1
2
3
4
5
6
7
0.50 0.75 1.00 1.25 1.50 1.75 2.00 2.25 2.50 2.75 3.00 3.25 3.50 3.75 4.00 4.25 4.50 4.75 5.00
φ = 2.00
φ = 1.00
φ = 0.67
Wavelength λ (μm)
I
r
r
a
d
i
a
n
c
e

(
W

m

2

μ
m

1
)
Figure 6.24
Spectral fame radiation measurements for natural gas (5 kW) combusted with pure O
2
contents for different equivalence ratios. (From Ji, B.,
and Baukal, C. E., “Spectral Radiation Properties of Oxygen-Enhanced/Natural Gas Flames,” Proceedings of 1998 International Gas Research
Conference, Vol. 5, 422–33, November 8–11, 1998, San Diego, CA, 1998.)
Figure 6.22
Experimental apparatus for measuring spectral fame radiation.
© 2011 by Taylor and Francis Group, LLC
Heat Flux 133
thermal radiation to penetrate into the glass. However,
that method is time-consuming and expensive. A
more convenient and reliable way to determine the
penetrating radiation is desirable. While the scanning
spectral radiometer employed in that work provides
quantitative measurements of the penetrating radia-
tion, it requires spectral irradiance calibration, is time
consuming, and may not be as conveniently applied to
large-scale fames in industrial furnaces. A more conve-
nient way is to use total radiometers with appropriate
spectral flters. Total (heat fux) radiometers have a fat
spectral response in the 0.6–15.0 µm region. Without
any window or flter in front of it, a heat fux meter
gives the integrated total radiation intensity over the
entire 0.6–15.0 µm range.
Since one of the most important heat transfer mecha-
nisms in a glass furnace is absorption of the fame radi-
ation by the glass melt, one must consider how fame
radiation propagates through the glass melt over the
entire spectral range. Several studies on high tempera-
ture glass melt absorption spectra [93–95] show that the
spectral absorption coeffcient increases dramatically
at about 2.7 µm. The glass melt essentially becomes
opaque at longer wavelengths due to strong absorp-
tion from the O–H and Si–O bonds in silicate. This is a
universal feature of molten glass. The spectral absorp-
tion at wavelengths shorter than 2.7 µm varies with the
impurities and/or additives in the glass melt, the tem-
perature, and the furnace atmosphere [93–95]. For low
iron (FeO) content molten glass without other impuri-
ties or additives, the spectral absorption at wavelengths
shorter than 2.7 µm is relatively small and constant [96].
The spectral absorption data of such low iron content
glass melts [96] has been used for the case study in this
work. Figure 6.26 shows the spectral absorption coef-
fcient K of low iron content molten glass as reported
by Berg [96]. Also plotted in Figure 6.26 are the spec-
tral radiation intensity of a fame with Ω = 1.0, ϕ = 2.0,
q
f
 = 5 kW, and the intensity after the radiation propa-
gates through 1 cm (0.4 in.) of molten glass, respectively.
It is evident from Figure 6.26 that only radiation at
wavelengths λ < 2.7 µm has signifcant penetration into
molten glass. Since glass is a poor heat conductor, the
radiation energy at λ > 2.7 µm only heats the top sur-
face of the glass, which can cause overheating that may
reduce the glass quality. Therefore, only the penetrat-
ing (λ < 2.7 µm) radiation is the effective radiation for
molten glass.
In order to improve heat transfer effciency and
reduce fuel consumption in glass furnaces, it is
important that the penetrating is maximized, rather
than the total radiation intensity. To compare the pen-
etrating radiation intensity among the various fames,
the spectral radiation was numerically integrated for
λ < 2.7 µm. The results are summarized in Table 6.1.
For the penetrating radiation, the pure oxygen/natu-
ral gas fame produced fve times more radiation than
the air-fred fame, for the same fuel consumption and
stoichiometry. Taking this into account, the fuel sav-
ings using oxygen enrichment is more pronounced for
glass furnaces than the savings based only in terms
of the total radiation. For oxygen/natural gas fames
8
7
6
5
4
3
2
1
0
0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00 2.25 2.50
Wavelength λ (µm)
I
r
r
a
d
i
a
n
c
e

(
W

m

2

µ
m

1
)
2.75 3.00 3.25 3.50 3.75 4.00 4.25 4.50
100% O
2
60% O
2
21% O
2
4.75 5.00 5.25
Figure 6.25
Spectral fame radiation measurements for natural gas (5 kW) stoichiometrically combusted with oxidizers with variable O
2
contents (21%,
60%, 100%). (From Ji, B., and Baukal, C. E., “Spectral Radiation Properties of Oxygen-Enhanced/Natural Gas Flames,” Proceedings of 1998
International Gas Research Conference, edited by D. Dolenc, Vol. 5, 422–33, November 8–11, 1998, San Diego, CA, 1998.)
© 2011 by Taylor and Francis Group, LLC
134 Industrial Combustion Testing
with different fuel equivalence ratios, a fuel rich fame
(ϕ = 2.00) apparently produced less total radiation
than a stoichiometric fame (ϕ = 1.00), due to incom-
plete combustion (see Table 6.1). However, the fuel
rich fame emits more penetrating radiation than a
stoichiometric fame even though not all the fuel was
consumed.
By using an appropriate spectral flter to block out
the undesirable longer wavelength (λ > 2.7 µm) radia-
tion, a heat fux meter can then be used to conveniently
give the integrated radiation intensity that would be
most penetrating into molten glass. Ji and Baukal tested
several optical flters and compared the results against
a spectral radiometer. Among the flters tested, a Schott
BK7 optical glass flter exhibited a sharp cut-off trans-
mission at λ = 2.7 µm. It became opaque at longer wave-
lengths for flter thickness greater than 6 mm (0.25 in.).
Figure 6.27 shows a comparison of the measured fame
radiation spectrum after it passed through a room
temperature BK7 glass window, and the calculated
fame radiation spectrum after it passed through 1 cm
thick, high temperature, low iron content molten glass
(with the same spectral absorption coeffcient used
in Figure 6.26). The BK7 transmitted spectra closely
matched that of the molten glass.
This technique is a simple, reliable way of determin-
ing the penetrating portion of the radiation from a
fame. This method was used to measure the penetrat-
ing radiation from q
f
 = 5.0 kW fames with different fuel
equivalence and oxygen enrichment ratios. Similar to
the analysis of spectral radiometer results, the radia-
tion intensities were normalized to that of a fame with
q
f
 = 5.0 kW, Ω = 1.0, and ϕ = 1.0. The agreement between
the heat fux meter with a BK7 flter measurement and
the numerically integrated spectral radiometer mea-
surement was satisfactory as shown by the qualitative
trends for columns I
nrm(pen.)
and I
nrm(BK7)
in Table 6.1.
This validated the use of a Vatell heat fux meter with a
Schott BK7 optical glass flter to determine the penetrat-
ing radiation.
To estimate the radiation intensity beyond 5.0 µm, a
sapphire window was placed in front of the total heat
fux meter. The transmission effciency of a sapphire
window is nearly constant at about 84% for λ = 0.25–5.0
µm, but quickly drops beyond 5.0 µm and cuts off at
4.75 4.5 4.25 4 3.75 3.5 3.25
I at melt surface
I at 1 cm below surface
R
a
d
i
a
t
i
o
n

i
n
t
e
n
s
i
t
y

(
W

m

2

µ
m

1
)
K
K
3 2.75 2.5
Wavelength λ (mm)
M
o
l
t
e
n

g
l
a
s
s

s
p
e
c
t
r
a
l

a
b
s
o
r
p
t
i
o
n

c
o
e
f
f
i
c
i
e
n
t
K

(
c
m

1
)
2.25 2 1.75 1.5 1.25 1 0.75 0.5 0.25
7
6
5
4
3
2
1
0
5 5.25
20
18
16
14
12
10
8
6
4
2
0
Figure 6.26
Interaction of fame radiation with high temperature glass melt. (From Ji, B., and Baukal, C. E., “Spectral Radiation Properties of Oxygen-
Enhanced/Natural Gas Flames,” Proceedings of 1998 International Gas Research Conference, edited by D. Dolenc, Vol. 5, 422–33, November 8–11,
1998, San Diego, CA, 1998.)
Table 6.1
Flame Radiation Intensity Measurements
Ω φ I
nrm(total)
I
nrm(pen.)
I
nrm(BK7)
γ
1.00 0.67 0.90 ± 0.06 0.89 ± 0.06 0.82 ± 0.06 0.15 ± 0.01
1.00 0.80 0.99 ± 0.07 0.98 ± 0.07 0.95 ± 0.07 0.13 ± 0.01
1.00 1.00 1.00* 1.00* 1.00* 0.14 ± 0.01
1.00 1.33 0.93 ± 0.07 0.98 ± 0.07 0.99 ± 0.07 0.17 ± 0.01
1.00 1.50 0.91 ± 0.06 0.93 ± 0.07 0.87 ± 0.06 0.20 ± 0.01
1.00 1.60 0.84 ± 0.06 0.83 ± 0.06 0.88 ± 0.06 0.21 ± 0.01
1.00 1.79 0.80 ± 0.06 0.83 ± 0.06 0.83 ± 0.06 0.37 ± 0.03
1.00 1.89 0.75 ± 0.05 0.87 ± 0.06 0.96 ± 0.07 0.55 ± 0.04
1.00 2.00 0.80 ± 0.06 1.12 ± 0.08 1.10 ± 0.08 1.02 ± 0.07
0.60 1.00 0.73 ± 0.05 0.70 ± 0.05 0.75 ± 0.05 0.12 ± 0.01
0.21 1.00 0.26 ± 0.02 0.19 ± 0.01 0.19 ± 0.01 0.005 ± 0.001
Source: From Ji, B., and Baukal, C. E., “Spectral Radiation Properties
of Oxygen-Enhanced/Natural Gas Flames,” Proceedings of
1998 International Gas Research Conference, Vol. 5, 422–33,
November 8–11, 1998, San Diego, CA, 1998.
* The Ω = 1.0, and ϕ = 1.0 intensities were used as the reference in the
normalization.
© 2011 by Taylor and Francis Group, LLC
Heat Flux 135
6.0 µm. The difference between the total radiometer
readings with and without the sapphire window can
be used to estimate radiation intensity beyond 5.0 µm.
The results are listed in Table 6.2. For Ω = 1.00 and
Ω = 0.60 fames, radiation beyond 5.0 µm accounted
for about 17% of the total fame radiation. For Ω = 0.21
fames, the I
(λ > 5.0µm)
was about 25% of the total radiation.
Combustion gas radiation at wavelengths longer than
5.0 µm is dominated by an H
2
O band at 6.3 µm [97,98].
The decrease in the relative contribution from shorter
wavelength vibration band intensity in Ω = 0.21 fames
can be attributed to less population in the highly excited
vibrational levels in H
2
O molecules. This can be caused
by two factors in Ω = 0.21 fames: less violent combus-
tion reactions lead to weaker (if any) nonequilibrium
chemiluminescence and a cooler fame temperature
leads to fewer equilibrium thermal populations in the
upper vibrational levels. Although the estimation given
was only approximate, it gave a qualitative indication of
the relative contribution from long wavelength (λ > 5.0
µm) radiation.
Kilham [17] and Jackson and Kilham [18] used a
thermopile normal to the surface of heated refractory
cylinders. The combustion products from the fame,
fowing around the cylinder, were also in the feld
of view. Therefore, the sum of the radiation from the
gases, and from the cylinder surface, was measured.
Two screens with narrow slit openings were placed
between the fame and the detector, to limit the feld
of view to a small region around the heated cylinder.
Giedt, Cobb, and Russ [55] measured radiation for a
H/H
2
/CO mixture fame. From available data and
radiation measurements, the gas emissivity was esti-
mated to be 0.001. Nonluminous gas radiation was cal-
culated to be, at most, only 2% of the total heat fux.
Woodruff and Giedt [56] measured gas and surface
radiation, with two Leeds and Northrup mirror-type
rayotubes. Radiation was found to be a negligible
effect. Veldman, Kubota, and Zukoski [58] concluded
that the measured radiation, from an air/C
3
H
8
fame,
was negligible. No details of the measurement tech-
niques were given.
Baukal and Gebhart [99] used surfaces of different
emissivities to study fame radiation. These removable
5.5
5
4.5
4
3.5
3
2.5
2
1.5
1
0.5
0
0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00 2.25
Wavelength λ (mm)
I (w/BK7, measured)
I (w/o BK7, measured) I (through 1 cm glass, calc.)
F
l
a
m
e

r
a
d
i
a
t
i
o
n

i
n
t
e
n
s
i
t
y

(
W

m

2

µ
m

1
)
2.50 2.75 3.00 3.25 3.50 3.75 4.00 4.25 4.50 4.75 5.00
Figure 6.27
Transmission of fame radiation through a room temperature BK7 optical glass flter and through high temperature molten glass. (From Ji, B.,
and Baukal, C. E., “Spectral Radiation Properties of Oxygen-Enhanced/Natural Gas Flames,” Proceedings of 1998 International Gas Research
Conference, edited by D. Dolenc, Vol. 5, 422–33, November 8–11, 1998, San Diego, CA, 1998.)
Table 6.2
Estimation of the Radiation Intensity Beyond 5.0 µm
Ω φ I
(w/sap.)
I
(λ < 5.0µm)
I
(w/o)
I
(λ > 5.0µm)
(λ > 5.0 µm)%
1.00 0.67 0.0476 0.0567 0.0672 0.0105 16
1.00 0.80 0.0504 0.0600 0.0723 0.0123 17
1.00 1.00 0.0532 0.0633 0.0751 0.0118 16
1.00 1.33 0.0504 0.0600 0.0727 0.0127 17
1.00 1.50 0.0482 0.0574 0.0684 0.0110 16
1.00 1.60 0.0465 0.0554 0.0674 0.0120 18
1.00 1.79 0.0439 0.0523 0.0636 0.0113 18
1.00 1.89 0.0429 0.0511 0.0609 0.0098 16
1.00 2.00 0.0443 0.0527 0.0620 0.0093 15
0.60 1.00 0.0434 0.0517 0.0631 0.0114 18
0.21 1.00 0.0170 0.0202 0.0269 0.0067 25
Source: From Ji, B., and Baukal, C. E., “Spectral Radiation Properties
of Oxygen-Enhanced/Natural Gas Flames,” Proceedings of
1998 International Gas Research Conference, Vol. 5, 422–33,
November 8–11, 1998, San Diego, CA, 1998.
© 2011 by Taylor and Francis Group, LLC
136 Industrial Combustion Testing
surfaces were attached to copper calorimeters. In
order to minimize contact resistance [100], a high
thermal conductivity copper-based antiseize lubri-
cant was used at the interface between the various
surfaces tested and the copper calorimeters. Polished
metal surfaces were used as the low emissivity sur-
faces. Surfaces coated with a high temperature fat
black paint were used as the high emissivity surfaces.
One of the problems with the polished surfaces was
very rapid degradation of both the brass and the cop-
per surfaces. Within minutes, those surfaces lost much
of their refectivity, due to the contact with the fame
gases. The polished stainless steel surfaces remained
refective for over an hour. There was no noticeable
degradation of the blackened surfaces, for all three
target materials, after hours of operation. Figure 6.28
shows that the blackened surface had the highest heat
fux and the polished surface had the lowest fux. The
difference between the two fuxes was relatively small
and represents the radiation portion of the total heat
fux to the surface. The majority of the heat fux was
by convection.
Sivathanu, Gore, and Dolinar [101] measured the soot
volume fractions in highly turbulent and strongly radi-
ating fames that are common industrial-type fames.
They used laser extinction measurements made with a
three-line optical probe that had laser extinction at 632
nm and two wavelength emission pyrometry at 900 and
1000 nm.
6.4 Convective Heat Flux
In low temperature processes, radiation may be insignif-
cant and can be ignored, which means that the convec-
tive heat fux can be measured with a standard heat fux
gauge. However, in high temperature combustion pro-
cesses where radiation is very signifcant, it is diffcult to
directly measure the convective heat fux. The technique
that is commonly used is to measure the total heat fux and
the radiant fux and then to calculate the convective heat
fux by subtracting the two. This obviously only works
if there are no other modes of heat transfer that contrib-
ute to the total heat fux. A related method to determine
the convective heat fux is by using other measurements
along with a calculation. To do this, the temperatures of
the surface where the convection is occurring and of the
gas fowing over the surface, along with the gas velocity
are needed. Using the appropriate forced convection heat
transfer equation for the given geometry, the convective
heat fux can be calculated. This method is useful as a
check against other methods that may be used.
Another method of determining the convective heat
fux by direct measurement involves adding a radiant
shield between the heat source and the heat fux gauge
to exclude radiation from being received by the gauge.
The problem is that the shield often disturbs the fow
and the measured fux is not representative of the actual
convective fow. Another problem is that the surface tem-
perature under the shield will be lower than it would be
without the shield, so the measurement would need to
be done quickly before the wall temperature under the
shield drops too much. In addition, although radiation
from the primary source (burners) may be shielded, the
rest of the furnace walls may still contribute radiation
to the gauge.
A third technique that could be used is to make a
perfectly refective heat fux gauge so that any radiation
received by the gauge is refected away and not mea-
sured by the gauge. Then (in theory) the gauge would
only measure convection, assuming no other heat trans-
fer mechanisms were important. Although in theory
this sounds plausible, the reality is that it is very dif-
fcult to have a perfect refector in an industrial combus-
tion environment.
A fourth technique that could be used is to use a heat
fux gauge embedded in the wall where convection is to
be measured and then to turn off the radiant source and
then quickly take measurements with the gauge. The
radiation from the hot furnace walls can be calculated
Polished (ε ~ 0.028)
Untreated (ε ~ 0.05 – 0.60)
Blackened (ε ~ 0.95)
850
800
750
700
650
600
q


(
k
W
/
m
2
)
550
500
450
0 1 2 3
L
4 5 6
Figure 6.28
Total heat fux (q”) versus axial distance between the burner out-
let and the impingement surface for O
2
/natural gas fames (fring
rate = 15.0 kW) impinging on the stagnation point (R
eff
 = 0.16) of
brass surfaces with various surface radiant characteristics. (From
Baukal, C. E., and Gebhart, B., Thermal and Fluid Science, 15, 323–25,
1997.)
© 2011 by Taylor and Francis Group, LLC
Heat Flux 137
and subtracted from the total fux. Again, this method
suffers from similar problems as the other methods
above in that it relies on a transient condition and accu-
rate calculations.
References
1. Arai, N., Matsunami, A., and Churchill, S. W. “A Review
of Measurements of Heat Flux Density Applicable to the
Field of Combustion.” Experimental Thermal and Fluid
Science 12, no. 4 (1996): 452–60.
2. Baukal, C. E. Heat Transfer in Industrial Combustion. Boca
Raton, FL: CRC Press, 2004.
3. Baukal, C. E., ed. The John Zink Combustion Handbook.
Boca Raton, FL: CRC Press, 2001.
4. Colannino, J. “Mathematical Models for Characterizing
and Predicting Heat Flux Profles from Ethylene Cracking
Units.” Proceedings of the 2007 AIChE Spring National
Meeting, Houston, TX, April 23–26, 2007.
5. Colannino, J. Modeling of Combustion Systems: A Practical
Approach. Boca Raton, FL: CRC Press, 2006.
6. Hayes, R., Singh, P., Foote, D., and Baukal, C. E. “Heat
Flux from Process Burners.” Proceedings of the ASME
IMECE, New York, November 11–16, 2001, Heat Transfer
Division, Vol. 369, No. 4, pp. 161–64, 2001.
7. Rafdi, N., and Blasiak, W. “Heat Transfer Character-
istics of HiTAC Heating Furnace Using Regenerative
Burners.” Applied Thermal Engineering 26, no. 16 (2006):
2027–34.
8. Hayes, R. R., Brewster, S., Webb, B. W., McQuay, M. Q.,
and Huber, A. M. “Crown Incident Radiant Heat Flux
Measurements in an Industrial, Regenerative, Gas-Fired,
Flat-Glass Furnace.” Experimental Thermal and Fluid Science
24 (2001): 35–46.
9. Newbold, J., Webb, B. W., McQuay, M. Q., and Huber,
A. M. “Combustion Measurements in an Industrial Gas-
Fired Flat-Glass Furnace.” Journal of the Institute of Energy
70 (1997): 71–81.
10. McQuay, M. Q., and Webb, B. W. “Effect of Rebuild on
the Combustion Performance of an Industrial Gas-Fired
Flat Glass Furnace.” Combustion Science and Technology
150, no. 1 (2000): 77–97.
11. Hoogendoorn, C. J., Ballintijn, C. M., and Dorresteijn, W.
R. “Heat-Flux Studies in Vertical Tube Furnaces.” Journal
of the Institute of Fuel 43 (1970): 511–16.
12. Goldstein, R. J., and Chiang, H. D. “Measurement of
Temperature and Heat Transfer.” In Handbook of Heat
Transfer Applications, edited by W. M. Rohsenow, J. P.
Hartnett, and E. N. Ganic´, 2nd ed. New York: McGraw-
Hill, 12-1–12-94, 1985.
13. Okoh, C. I., and Brown, R. A. Combustion Experimentation
Handbook. Gas Research Institute report GRI-88/0143,
Chicago, IL, 1988.
14. Bachmann, R. C., Chambers, J. T., and Giedt, J. T.
“Investigation of Surface Heat-Flux Measurements with
Calorimeters.” ISA Transactions 4 (1965): 143–51.
15. Beasley, D. E., and Figliola, R. S. “A Generalised (sic)
Analysis of a Local Heat Flux Probe.” Journal of Physics
E-Scientifc Instruments 21 (1988): 316–22.
16. Ducharme, M. B., Frim, J., and Tikuisis, P. “Errors in Heat
Flux Measurements Due to the Thermal Resistance of
Heat Flux Disks.” Journal of Applied Physiology 69, no. 2
(1990): 776–84.
17. Kilham, J. K. “Energy Transfer from Flame Gases to
Solids.” Third Symposium on Combustion and Flame and
Explosion Phenomena, 733–40. Baltimore, MD: The Williams
and Wilkins Co., 1949.
18. Jackson, E. G., and Kilham, J. K. “Heat Transfer from
Combustion Products by Forced Convection.” Industrial
& Engineering Chemistry 48, no. 11 (1956): 2077–79.
19. Davies, R. M. “Heat Transfer Measurements on
Electrically-Boosted Flames.” 10th Symposium (Inter-
national) on Combustion, The Combustion Institute,
Pittsburgh, PA, 755–66, 1965.
20. Fells, I., and Harker, J. H. “An Investigation Into Heat
Transfer from Unseeded Propane-Air Flames Augmented
with D.C. Electrical Power.” Combustion and Flame 12
(1968): 587–96.
21. Vizioz, J.-P., and Lowes, T. M. “Convective Heat Transfer
from Impinging Flame Jets.” International Flame
Research Foundation report F 35/a/6, IJmuiden, the
Netherlands, 1971.
22. Davies, D. R. “Heat Transfer from Working Flame
Burners.” BS thesis, University of Salford, UK, 1979.
23. Posillico, C. J. “Raman Spectroscopic and LDV Measure-
ments of a Methane Jet Impinging Normally on a Flat
Water-Cooled Boundary.” PhD thesis, Polytechnic
Institute of New York, 1986.
24. Smith, R. B., and Lowes, T. M. “Convective Heat Transfer
from Impinging Tunnel Burner Flames—A Short Report
on the NG-4 Trials.” International Flame Research Founda-
tion report F 35/a/9, IJmuiden, the Netherlands, 1974.
25. Rajani, J. B., Payne, R., and Michelfelder, S. “Convective
Heat Transfer from Impinging Oxygen-Natural Gas
Flames—Experimental Results from the NG5 Trials.”
International Flame Research Foundation report F
35/a/12, IJmuiden, the Netherlands, 1978.
26. Baukal, C. E., and Gebhart, B. “Heat Transfer from
Oxygen-Enhanced/Natural Gas Flames Impinging
Normal to a Plane Surface.” Experimental Thermal and
Fluid Science 16, no. 3 (1998): 247–59.
27. American Society for Testing and Materials Standard E
422 – 83, Standard Method for Measuring Heat Flux Using
a Water-Cooled Calorimeter, Philadelphia, PA, 1983.
28. Anderson, J. E., and Stresino, E. F. “Heat Transfer from
Flames Impinging on Flat and Cylindrical Surfaces.”
Journal of Heat Transfer 85, no. 1 (1963): 49–54.
29. Reed, T. B. “Heat-Transfer Intensity from Induction
Plasma Flames and Oxy-Hydrogen Flames.” Journal of
Applied Physics 34, no. 8 (1963): 2266–69.
30. Fay, R. H. “Heat Transfer from Fuel Gas Flames.” Welding
Journal, Research Supplement, (1967): 380s–83s.
31. Cookson, R. A., and Kilham, J. K. “Energy Transfer from
Hydrogen-Air Flames.” Ninth Symposium (International)
on Combustion, 257–63. New York: Academic Press,
1963.
© 2011 by Taylor and Francis Group, LLC
138 Industrial Combustion Testing
32. Kilham, J. K., and Dunham, P. G. “Energy Transfer from
Carbon Monoxide Flames.” 11th Symposium (International)
on Combustion, The Combustion Institute, Pittsburgh, PA,
899–905, 1967.
33. Cookson, R. A., Dunham, P. G., and Kilham, J. K.
“Stagnation Point Heat Flow Meter.” Journal of Scientifc
Instruments 42 (1965): 260–62.
34. Beér, J. M., and Chigier, N. A. “Impinging Jet Flames.”
Combustion and Flame 12 (1968): 575–86.
35. Buhr, E., Haupt, G., and Kremer, H. “Heat Transfer from
Impinging Turbulent Jet Flames to Plane Surfaces.” In
Combustion Institute European Symposium 1973, edited
by F. J. Weinberg, 607–12. New York: Academic Press,
1973.
36. Kremer, H., Buhr, E., and Haupt, R. “Heat Transfer from
Turbulent Free-Jet Flames to Plane Surfaces.” In Heat
Transfer in Flames, edited by N. H. Afgan and J. M. Beér,
463–72. Washington, DC: Scripta Book Company, 1974.
37. Hoogendoorn, C. J., Popiel, C. O., and van der Meer, T.
H. “Turbulent Heat Transfer on a Plane Surface in
Impingement Round Premixed Flame Jets.” Proceedings of
6th International Heat Transfer Conference, 107–12. Toronto,
Vol. 4, August 7–11, 1978.
38. Popiel, C. O., van der Meer, T. H., and Hoogendoorn, C. J.
“Convective Heat Transfer on a Plate in an Impinging
Round Hot Gas Jet of Low Reynolds Number.”
International Journal of Heat and Mass Transfer 23 (1980):
1055–68.
39. Hargrave, G. K., and Kilham, J. K. “The Effect of
Turbulence Intensity on Convective Heat Transfer From
Premixed Methane-Air Flames.” Institution of Chemical
Engineers Symposium Series 2, no. 86 (1984): 1025–34,
1984.
40. Hargrave, G. K., Fairweather, M., and Kilham, J. K. “Forced
Convective Heat Transfer from Premixed Flames—Part
2: Impingement Heat Transfer.” International Journal of
Heat and Fluid Flow 8, no. 2 (1987): 132–38.
41. Kataoka, K., Shundoh, H., and Matsuo, H. “Convective
Heat Transfer Between a Flat Plate and a Jet of Hot
Gas Impinging On It.” In Drying ’84, edited by A. S.
Mujumdar, 218–27. New York: Hemisphere/Springer-
Verlag, 1984.
42. Cassiano, J., Heitor, M. V., and Silva, T. F. “Combustion
Tests on an Industrial Glass-Melting Furnace.” Fuel 73,
no. 10 (1994): 1638–42.
43. Shorin, S. N., and Pechurkin, V. A. “Effectivnost’ teplo-
perenosa na poverkhnost’ plity ot vysokotemperaturnoi
strui produktov sjoraniya razlichnykh gazov” (The
Effectiveness of Heat Transfer to the Surface of a Plate
from a High-Temperature Jet of Combustion Products of
Various Gases). Teoriya i Praktika Szhiganiya Gaza 4 (1968):
134–43.
44. Schulte, E. M., and Kohl, R. F. “A Transducer for
Measuring High Heat Transfer Rates.” Review of Scientifc
Instruments 41, no. 12 (1970): 1732–40.
45. Schulte, E. M. “Impingement Heat Transfer Rates from
Torch Flames.” Journal of Heat Transfer 94 (1972): 231–33.
46. You, H.-Z. “An Investigation of Fire-Plume Impingement
on a Horizontal Ceiling: 2—Impingement and Ceiling-Jet
Regions.” Fire & Materials 9, no. l (1985): 46–56.
47. Hager, J. M., Langley, L. W., Onishi, S., and Diller, T. E.
“Microsensors for High Heat Flux Measurements.”
Journal of Thermophysics 7, no. 3 (1993): 531–34.
48. Gardon, R. “A Transducer for the Measurement of Heat
Flow Rate.” Journal of Heat Transfer 82 (1960): 396–98.
49. Grey, J. “Probe Measurements in High Temperature Gases
and Dense Plasmas.” In Measurements in Heat Transfer,
edited by E. R. G. Eckert and R. J. Goldstein, 2nd ed.
Washington, DC: Hemisphere Publishing, 337–374, 1976.
50. Ash, R. L. “Response Characteristics of Thin Foil Heat
Flux Sensors.” AIAA Journal 7, no. 12 (1969): 2332–35.
51. Borell, G. J., and Diller, T. E. “A Convection Calibration
Method for Local Heat Flux Gages.” Journal of Heat
Transfer 109 (1987): 83–89.
52. van der Meer, T. H. “Stagnation Point Heat Transfer from
Turbulent Low Reynolds Number Jets and Flame Jets.”
Experimental Thermal and Fluid Science 4 (1991): 115–26.
53. Barnes, A. “Heat Flux Sensors, Part 2: Applications.”
Sensors 16, no. 2 (1999): 54–57.
54. Gebhart, B. Heat Transfer. New York: McGraw-Hill, 1971.
55. Giedt, W. H., Cobb, L. L., and Russ, E. J. “Effect of
Hydrogen Recombination on Turbulent Flow Heat
Transfer.” ASME Paper 60-WA-256, New York, 1960.
56. Woodruff, L. W., and Giedt, W. H. “Heat Transfer
Measurements from a Partially Dissociated Gas with
High Lewis Number.” Journal of Heat Transfer 88 (1966):
415–20.
57. Milson, A., and Chigier, N. A. “Studies of Methane-Air
Flames Impinging on a Cold Plate.” Combustion and
Flame 21 (1973): 295–305.
58. Veldman, C. C., Kubota, T., and Zukoski, E. E. “An
Experimental Investigation of the Heat Transfer from
a Buoyant Gas Plume to a Horizontal Ceiling—Part 1.
Unobstructed Ceiling.” National Bureau of Standards
report NBS-GCR-77-97, Washington, DC, 1975.
59. Matsuo, M., Hattori, M., Ohta, T., and Kishimoto, S. “The
Experimental Results of the Heat Transfer by Flame
Impingement.” International Flame Research Foundation
report F 29/1a/1, IJmuiden, the Netherlands, 1978.
60. Rauenzahn, R. M. “Analysis of Rock Mechanics and Gas
Dynamics of Flame-Jet Thermal Spallation Drilling.”
PhD thesis, MIT, Cambridge, MA, 1986.
61. Gatowski, J. A., Smith, M. K., and Alkidas, A. C. “An
Experimental Investigation of Surface Thermometry and
Heat Flux.” Experimental Thermal and Fluid Science 2, no. 3
(1989): 280–92.
62. Hornbaker, D. R., and Rall, D. L. “Thermal Perturbations
Caused by Heat-Flux Transducers and Their Effect on
the Accuracy of Heating-Rate Measurements.” ISA
Transactions 3 (1964): 123–30.
63. Kilham, J. K., and Purvis, M. R. I. “Heat Transfer from
Hydrocarbon-Oxygen Flames.” Combustion and Flame 16
(1971): 47–54.
64. Nawaz, S. “Heat Transfer from Oxygen Enriched
Methane Flames.” PhD thesis, The University of Leeds,
U.K., 1973.
65. Fairweather, M., Kilham, J. K., and Nawaz, S. “Stagnation
Point Heat Transfer from Laminar, High Temperature
Methane Flames.” International Journal of Heat and Fluid
Flow 5, no. 1 (1984): 21–27.
© 2011 by Taylor and Francis Group, LLC
Heat Flux 139
66. Fairweather, M., Kilham, J. K., and Mohebi-Ashtiani,
A. “Stagnation Point Heat Transfer from Turbulent
Methane-Air Flames.” Combustion Science and Technology
35 (1984): 225–38.
67. Conolly, R., and Davies, R. M. “A Study of Convective
Heat Transfer from Flames.” International Journal of Heat
and Fluid Flow 15 (1972): 2155–172.
68. Kilham, J. K., and Purvis, M. R. I. “Heat Transfer from
Normally Impinging Flames.” Combustion Science and
Technology 18 (1978): 81–90.
69. Ivernel, A., and Vernotte, P. “Etude expérimentale de
l’amélioration des transferts convectis dans les fours
par suroxygénation du comburant.” Revue Generale de
Thermique 18, nos. 210–211 (1979): 375–91.
70. Horsley, M. E., Purvis, M. R. I., and Tariq, A. S.
“Convective Heat Transfer From Laminar and Turbulent
Premixed Flames.” Heat Transfer 1982, edited by U.
Grigull, E. Hahne, K. Stephan, and J. Straub, 409–15.
Washington, DC: Hemisphere, Vol. 3, 1982.
71. Hemeson, A. O., Horsley, M. E., Purvis, M. R. I., and Tariq,
A. S. “Heat Transfer from Flames to Convex Surfaces.”
Institution of Chemical Engineers Symposium Series 2, no.
86 (1984), Rugby, UK: Institution of Chemical Engineers,
969–78.
72. Hustad, J. E., Røkke, N. A., and Sønju, O. K. “Heat
Transfer to Pipes Submerged in Lifted Buoyant Diffusion
Flames.” In Experimental Heat Transfer, Fluid Mechanics,
and Thermodynamics, 1991, edited by J. F. Keffer, 567–74.
New York: Elsevier, 1991.
73. Hustad, J. E., and Sønju, O. K. “Heat Transfer to Pipes
Submerged in Turbulent Jet Diffusion Flames.” In Heat
Transfer in Radiating and Combusting Systems. Berlin:
Springer-Verlag, 474–90.
74. Hustad, J. E., Jacobsen, M., and Sønju, O. K. “Radiation
and Heat Transfer in Oil/Propane Jet Diffusion Flames.”
Institution of Chemical Engineers Symposium Series 10, no.
129 (1992): 657–63.
75. Gray, W. A., and Müller, R. Engineering Calculations in
Radiative Heat Transfer. Oxford, UK: Pergamon, 1974.
76. Stradomskiy, M. V., Maksimov, Y. A., and Malyarov, V. S.
“Calibration of High-Temperature Heat Flux Sensors.”
Heat Transfer–Soviet Research 20, no. 4 (1988): 562–68.
77. Ludwig, C. B., Malkmus, W., Reardon, J. E., and Thomson,
J. A. L. Handbook of Infrared Radiation from Combustion
Gases. Washington, DC: National Aeronautics and Space
Administration report SP-3080, 1973.
78. Siegel, R., and Howell, J. R. Thermal Radiation Heat
Transfer. 2nd ed. Washington, DC: Hemisphere, 1981.
79. Baukal, C. E., and Gebhart, B. “Oxygen-Enhanced
Natural Gas Flame Radiation.” International Journal of
Heat and Mass Transfer 40, no. 11 (1997): 2539–47.
80. Gore, J. P., Skinner, S. M., Stroup, D. W., Madrzykowski,
D., and Evans, D. D. “Structure and Radiation
Properties of Large Two-Phase Flames.” In Heat Transfer
in Combustion Systems, edited by N. Ashgriz, J. G.
Quintiere, H. G. Semerjian, and S. E. Slezak, 77–86. New
York: American Society of Mechanical Engineers, HTD-
Vol. 122, 1989.
81. Lide, D. R., ed. CRC Handbook of Chemistry and Physics,
79th ed. Boca Raton, FL: CRC Press, 1998.
82. Matthews, L., Harris, A., and Garcia, G. “Radiative Flux
Measurements in a Sooty Pool Fire Using a Multihead
Transpiration Radiometer.” In Heat Transfer Phenomena
in Radiation, Combustion, and Fires, edited by R. K. Shah,
375–80. New York: ASME HTD-Vol. 106, 1989.
83. Moffat, R. J., Hunn, B. D., and Ayers, J. F. “Development
of a Transpiration Radiometer.” Instrument Society of
America, Conference Proceedings, Paper No. 613, pp.
613.1-613.7, 1971.
84. Brajuskovic, B., Matovic, M., and Afgan, N. “A Heat
Flux-Meter for Ash Deposit Monitoring Systems – I. Ash
Deposit Prevention.” International Journal of Heat and
Mass Transfer 34, no. 9 (1991): 2291–-2301.
85. Brajuskovic, B., and Afgan, N. “A Heat Flux-Meter for
Ash Deposit Monitoring Systems—II. ‘Clean’ Heat Flux-
Meter Characteristics.” International Journal of Heat and
Mass Transfer 34, no. 9 (1991): 2303–15.
86. Martins, N., Carvalho, M. G., Afgan, N. H., and Leontiev,
A. L. “Experimental Verifcation and Calibration of the
Blow-Off Heat Flux Sensor.” Applied Thermal Engineering
18, no. 6 (1998): 481–89.
87. Martins, N., Carvalho, M. G., Afgan, N. H., and Leontiev,
A. I. “Design and Sensitivity Analysis of a New Gauge
for Radiation Heat Flux Assessment.” Heat & Technology
16, no. 2 (1998): 77–84.
88. Ploteau, J. P., Glouannec, P., and Noel, H. “Conception
of Thermoelectric Flux Meters for Infrared Radiation
Measurements in Industrial Furnaces.” Applied Thermal
Engineering 27, nos. 2–3 (2007): 674–81.
89. Chedaille, J., and Braud, Y. Vol. 1: Measurements in Flames.
New York: Crane, Russak & Co., 1972.
90. Butler, B. W., and Webb, B. W. “Measurements of Local
Temperature and Wall Radiant Heat Flux in an Industrial
Coal-Fired Boiler.” In Heat Transfer in Combustion
Systems—1990, edited by B. Farouk, W. L. Grosshandler,
D. G. Lilley, and C. Presser, 49–56. New York: ASME
HTD-Vol. 142, 1990.
91. Beltagui, S. A., Kenbar, A. M. A., and Maccallum,
N. R. L. “Heat Transfer and Emission Studies in a Gas
Fired Furnace.” In Boilers & Furnaces. Proceedings of
International Symposium on Combustion & Emissions
Control, Cardiff, UK, September 1993, Institute of Energy,
pp. 275–96, 1993.
92. Ji, B., and Baukal, C. E. “Spectral Radiation Properties
of Oxygen-Enhanced/Natural Gas Flames.” Proceedings
of 1998 International Gas Research Conference, edited by
D. Dolenc, Vol. 5, 422–33. November 8–11, 1998, San
Diego, CA, 1998.
93. Franz, H. “Infrared Absorption of Molten Soda-Lime-
Silica Glasses Containing Transition Metal Oxides.”
International Congress on Glass, Science and Technology 9,
no. 1 (1971): 243–60.
94. Endrys, J., Geotti-Bianchini, F., and De Riu, L. “Study of
the High-Temperature Spectral Behavior of Container
Glass.” Glastechnische Berichte Glass Science and Technology
70, no. 5 (1997): 126–36.
95. Banner, D., and Klarsfeld, S. “High Temperature Infrared
Spectra of Silicate Melts.” In Physics of Non-crystalline
Solids, edited by L. D. Pye, W. C. La Course, and H. J.
Stevens, 371–75, London, 1992.
© 2011 by Taylor and Francis Group, LLC
140 Industrial Combustion Testing
96. Berg, J. I. “Near Infrared Absorption Coeffcient of Molten
Glass by Emission Spectroscopy.” International Journal of
Thermophysics 2, no. 4 (1981): 381–94.
97. Faeth, G. M., Gore, J. P., Chuech, S. G., and Jeng, S.-M.
“Radiation from Turbulent Diffusion Flames.” Annual
Review of Fluid Mechanics & Heat Transfer 2 (1989): 1–38.
98. Ludwig, C. B., Malkmus, W., Reardon, J. E., and Thomson,
J. A. L. Handbook of Infrared Radiation from Combustion
Gases. Washington, DC: National Aeronautics and Space
Administration report SP-3080, 1973.
99. Baukal, C. E., and Gebhart, B. “Surface Condition Effects
on Flame Impingement Heat Transfer.” Thermal and Fluid
Science 15 (1997): 323–35.
100. Holman, J. P. Heat Transfer. 10th ed. New York: McGraw-
Hill, 2010.
101. Sivathanu, Y. R., Gore, J. P., and Dolinar, J. “Transient
Scalar Properties of Strongly Radiating Jet Flames.” In
Heat and Mass Transfer in Fires and Combustion Systems,
edited by W. L. Grosshandler and H. G. Semerjian, 45–56.
New York: ASME HTD-Vol. 148, ASME, 1990.
© 2011 by Taylor and Francis Group, LLC
141
7.1 Introduction
There are a number of reasons for sampling the exhaust
gases from an industrial combustion process [1]. One is
to determine whether the emissions are within regula-
tory limits. Another is to determine the effciency and
effectiveness of the pollution control technology being
used in a given process. Stricter limits or ineffective pol-
lution control technologies may lead to design changes
in the equipment. Sampling is also useful for predict-
ing any potential problems in the development of a new
process or modifcation of the existing process. The U.S.
Environmental Protection Agency has established sam-
pling methods for different pollutants in 40 CFR 60 [2].
These are available electronically over the Internet [3].
7
Pollution Emissions
Charles E. Baukal, Jr.
CONTENTS
7.1 Introduction ....................................................................................................................................................................141
7.2 Some Statistics ............................................................................................................................................................... 142
7.3 Sampling ........................................................................................................................................................................ 144
7.3.1 Extractive Sampling .......................................................................................................................................... 148
7.3.2 In-Situ Sampling ............................................................................................................................................... 152
7.3.3 Ambient Condition Effects .............................................................................................................................. 152
7.4 Testing Equipment ........................................................................................................................................................ 160
7.4.1 Exhaust Gas Flow ............................................................................................................................................. 160
7.4.1.1 Pitot Tube ............................................................................................................................................. 160
7.4.1.2 Direct Measurement ...........................................................................................................................161
7.4.1.3 Calculation ...........................................................................................................................................162
7.4.1.4 Non-Contact .........................................................................................................................................162
7.4.2 Gas Composition ................................................................................................................................................162
7.4.3 Particulates ........................................................................................................................................................ 166
7.5 Source Testing ................................................................................................................................................................167
7.6 Continuous Monitoring ............................................................................................................................................... 169
7.7 Spot Checking ............................................................................................................................................................... 170
7.8 Example Measurements ............................................................................................................................................... 170
7.8.1 Oxygen-Enhanced Combustion...................................................................................................................... 170
7.8.1.1 Experimental Setup ........................................................................................................................... 170
7.8.1.2 Results .................................................................................................................................................. 172
7.8.1.3 Discussion ........................................................................................................................................... 173
7.8.1.4 Conclusions and Recommendations ............................................................................................... 173
7.8.2 Fuel Composition Effects ................................................................................................................................. 173
7.8.2.1 Test Setup .............................................................................................................................................174
7.8.2.2 Sample Results .................................................................................................................................... 175
7.8.2.3 Conclusions ..........................................................................................................................................176
7.8.3 Ghost NO
x
..........................................................................................................................................................176
7.8.3.1 Introduction ........................................................................................................................................ 177
7.8.3.2 Test Description .................................................................................................................................. 177
7.8.3.3 Results .................................................................................................................................................. 178
7.8.3.4 Conclusions ......................................................................................................................................... 181
References ................................................................................................................................................................................ 181
© 2011 by Taylor and Francis Group, LLC
142 Industrial Combustion Testing
There are many useful general references on experimen-
tal measurements [4–8] as well as specifc references on
pollution measurement techniques (e.g., [9–15]).
Crawford (1976) lists the following steps in the process
of measuring pollutants in an air stream [16]:
1. Collection of the air sample containing the
pollutant
2. Conditioning of the sample
3. Separation of the pollutants from the air in the
sample
4. Measurement of the desired properties of the
pollutants
5. Measuring the sample fow rate
6. Disposal of the sample gas after completing the
measurements
It is noted that not all of these steps are required in par-
ticular circumstances. The last step is normally not a
factor for most industrial combustion processes where
the gas sample is vented to the atmosphere. This is
only a factor where the pollutants in the sample may
be poisonous, radioactive, or toxic in the concentrations
present.
The objective of this chapter is to discuss the general
principles in measuring pollutants in industrial com-
bustion exhaust gas streams [17].
7.2 Some Statistics
There are some limited statistics that may be needed
for measuring and reporting pollution emissions. These
are briefy discussed here without derivation or proof.
The reader is referred to environmental statistics books
for those details [18–29]. The frst concept is the mean or
average value of a set of numbers:
x
n
x
i
i
n
=
=

1
1
, (7.1)
where x is the average value, x
i
is the ith value, i is the
index going from 1 to n, and n is the total number of
observations.
The range of readings is:
Range = x
min
→ x
max
. (7.2)
Some operating permits may be written on a not-
to- exceed basis, which means that a given pollutant
concentration cannot exceed a given value. In that
case, only the maximum value over a given time
period may be of interest. Some permits are written so
that the average concentration cannot exceed a specifc
value, so it is important to know how the permit is
written.
Another statistical concept that may be used in report-
ing pollutant emissions is referred to as a moving aver-
age. Some regulations are written based on the average
of a set of readings taken over a specifc time period that
is moving. For example, a pollutant may be limited to
a certain maximum concentration over an eight-hour
moving average.
The moving average takes the most recent data for
the specifed period. Figure 7.1 shows a set of daily
averages for a 10-day time span with a line through
the data, which is a linear regression ft. Note that
38
40
42
44
1 2 3 4 5 6 7 8 9 10
Day
A
v
e
r
a
g
e

N
O
x

(
p
p
m
v
d
)

a
t

3
%

O
2
Figure 7.1
Average emissions over a 10-day time span with a linear regression ft. (From Baukal, C. E., Industrial Combustion Pollution and Control, Marcel
Dekker, New York, 2004.)
© 2011 by Taylor and Francis Group, LLC
Pollution Emissions 143
the vertical axis has been exaggerated to show the
relatively small shift in the averages during that time
span. The graph shows that average daily NO
x
appears
to be increasing. This may indicate some type of prob-
lem that needs to be corrected before regulatory limits
are exceeded. An example might be that the removal
effciency of a downstream NO
x
posttreatment sys-
tem may be declining and the equipment is in need
of service.
Another statistical measure that is useful for report-
ing data is referred to as the standard deviation. This is
a measure of the dispersion of the data from the average
value. Figure 7.2 shows two sets of data taken on two
different days over the same hours of the day. The data
have been artifcially chosen so that the average over
the given period is the same for both days, but there is
considerably more “scatter” on Day 2, which means that
the data are farther from the average value compared
to Day 1. The Day 2 data then have a higher standard
deviation than that of the Day 1 data.
The unbiased or “n-1” standard deviation is
defned as:
σ
u
i
i
n
x x
n
=


=

( )
.
2
1
1
(7.3)
The biased or “n” standard deviation is defned as:
σ
b
i
i
n
x x
n
=

=

( )
,
2
1
(7.4)
where the difference between the two is the divisor. For
a large sample size where n is large, these two will be
essentially the same.
The unbiased standard deviation is typically used
for emissions reporting and will be used here for illus-
tration purposes. The Day 1 data are assumed to be
normally distributed about the mean value as shown
in Figure 7.3. This means that if enough data were col-
lected and the data were truly normally distributed
about the mean, then they would ft the curve shown
in the fgure. The graph also shows the limits of one
standard deviation from the mean both to the left
and to the right. As can be seen, the majority of the
data would fall in the region bounded by one stan-
dard deviation to the left and to the right of the mean.
Therefore, it is common to report data in the following
form:
x x = ± σ. (7.5)
The standard deviation shows how tightly distributed
the data are about the mean. The smaller the standard
deviation, the more closely the data are to the mean
value. Figure 7.2 showed that the Day 2 data were more
scattered compared to the Day 1 data. Figure 7.4 shows
a comparison of the normal distribution curves for the
Day 1 and Day 2 data, both of which have the same
mean value. The curve for Day 1 is taller and thin-
ner, which shows that most of the data are closer to
the mean. The curve for Day 2 is shorter and fatter,
which shows that most of the data are farther from the
mean.
A high relative standard deviation could indicate that
there may be a problem with the measurement system
30
35
40
45
50
12:00
PM
1:00
PM
2:00
PM
3:00
PM
4:00
PM
5:00
PM
6:00
PM
7:00
PM
8:00
PM
9:00
PM
10:00
PM
11:00
PM
Time of the day
N
O
x

(
p
p
m
v
d
)

a
t

3
%

O
2
Day 1
Day 2
Average
Figure 7.2
Distribution about a mean value of data taken on two consecutive days over the same time span. (From Baukal, C. E., Industrial Combustion
Pollution and Control, Marcel Dekker, New York, 2004.)
© 2011 by Taylor and Francis Group, LLC
144 Industrial Combustion Testing
where there is not high repeatability if the data are not
truly changing. A high standard deviation could also
indicate that there are wider swings in the data that
could be caused by changing conditions. For example,
if the fuel composition is changing with time, which is
sometimes the case in petrochemical applications, this
could cause the NOx to change signifcantly. Another
example is a changing ambient environment where
there may be wide swings in the temperature and rela-
tive humidity of the incoming combustion air that can
affect NOx emissions. A high relative standard devia-
tion is not generally desirable because more cushion
will be needed between the permitted or regulated
emission limits and the measured emissions because of
the higher possibility of signifcant excursions above the
mean value.
7.3 Sampling
An intersociety committee put together a guidebook for
air sampling, primarily in the ambient atmosphere [30].
The following general physical precautions relevant
to extractive gas sampling were noted: ensure that the
sample is homogeneous, be aware of any absorption
and diffusion effects in the sampling system, check for
mechanical defects in the sampling system, and cali-
brate at the design sampling fow rate. It is also noted
that any interferences between chemicals in a sampling
system must also be known and corrected for in order
to ensure accurate and reliable measurements. The
U.S. EPA has developed an extensive manual to aid
in estimating the costs of pollution control equipment
0%
2%
4%
6%
8%
10%
12%
14%
16%
20 25 30 35 40 45 50 55 60
NO
x
(ppmvd) at 3% O
2
F
r
e
q
u
e
n
c
y

(
%
)
Day 1
Day 2
Average
Figure 7.4
Comparison of normal distributions for Day 1 with a smaller standard deviation and Day 2 with a larger standard deviation, both with the
same mean value. (From Baukal, C. E., Industrial Combustion Pollution and Control, Marcel Dekker, New York, 2004.)
0%
2%
4%
6%
8%
10%
12%
14%
16%
30 32 34 36 38 40 42 44 46 48
NOx (ppmvd) at 3% O
2
F
r
e
q
u
e
n
c
y

(
%
)
Day 1
Average
-1 Std. Dev.
+1 Std. Dev.
Figure 7.3
Normal distribution of Day 1 data about the mean of 38.7 ppmvd NO
x
at 3% O
2
.
© 2011 by Taylor and Francis Group, LLC
Pollution Emissions 145
[31]. A chapter in the manual is devoted to emissions
monitoring [32].
It is important that the position of the sample probe is
known accurately if there is a considerable variation in the
gas composition profle across the duct [33]. It is preferred
that the measurements be made suffciently downstream
of fow disturbances like elbows or dampers to minimize
the variation across the duct. If this is not possible as is
the case when physical restrictions limit how the exhaust
ducting can be run and where the measurements can be
made, then more traverses are recommended.
The EPA Method 1 (see Appendix B for a complete
list of EPA Methods) recommends that both sample and
velocity measurements be made at least 8 internal diam-
eters downstream of any fow disturbance (e.g., elbow,
duct contraction, or expansion) and at least 2 internal
diameters upstream from any fow disturbance. If this is
not possible, then measurements may be made as close
as 2 internal diameters downstream and 0.5 internal
diameters upstream of disturbances, but more measure-
ments across the duct cross section are needed. Figure 7.5
shows the recommended minimum number of traverse
(sampling) points for measuring either the gas sample
composition or the gas velocity where no particulates are
in the gas stream. The fgure shows that at least 8 or 9
measurement points are recommended for internal stack
diameters between 12 and 24 in. where the measurement
location is at least 8 internal diameters downstream and
2 internal diameters upstream from any upstream dis-
turbances. For internal stack diameters greater than 24
in. in internal diameter, at least 12 sampling points are
recommended if the measurement location is at least 7
internal diameters downstream and 1.75 internal diam-
eters upstream of any fow disturbance. A minimum of
16 sample points are recommended for internal stack
diameters 12 in. in internal diameter or greater where
the sampling location is between 2 and 7 (8 for internal
stack diameters between 12 and 24 in.) internal diame-
ters downstream and between 0.5 and 1.75 (2 for internal
stack diameters between 12 and 24 in.) internal diameters
upstream from a fow disturbance. Figure 7.6 shows the
recommended minimum number of traverse points if
there are particulates in the fow. As can be seen, more
traverses are generally needed when particulates are
present.
The minimum distance to a disturbance determines
the minimum number of recommended traverse points.
For example, if the stack diameter is 18 in. in internal
diameter, the gas stream contains no particulates, and
the sample location is 10 internal diameters downstream
from the closest downstream disturbance but only
1 internal diameter upstream from the closest upstream
disturbance, then the distance to the upstream distur-
bance determines the minimum number of traverses.
According to Figure 7.5, only 8 or 9 traverses would be
needed based on the downstream distance (10 diame-
ters) to a disturbance but at least 16 traverses are recom-
mended because the sample location is only 1 internal
diameter upstream from the location.
For rectangular ducts, an equivalent diameter is cal-
culated using the following
D
LW
L W
e
=
+
2
, (7.6)
Measurement
site
Disturbance
Disturbance
Stack diameter > 0.61 m (24 in.)
Duct diameters upstream from flow disturbance* (distance A)
Higher number is for
rectangular stacks or ducts
Stack diameter = 0.30 to 0.61 m (12-24 in.)
12
2
50
0.5 1.0 1.5 2.0 2.5
30
M
i
n
i
m
u
m

n
u
m
b
e
r

o
f

t
r
a
v
e
r
s
e

p
o
i
n
t
s
20
10
0
40
3 4 5
Duct diameters downstream from flow disturbance* (distance B)
6 7 8 9 10
B
A
16
8 or 9
* From point of any type of
disturbance (bend, expansion, contraction, etc.)
a
a
Figure 7.5
Minimum number of traverse points for sample or velocity measurements in when particulates are not present. (From U.S. EPA, 40 CFR 60:
Standards of Performance for New Stationary Sources, Washington, DC: U.S. Environmental Protection Agency, 2001.)
© 2011 by Taylor and Francis Group, LLC
146 Industrial Combustion Testing
where L is the length of the rectangular duct cross section,
W is the width of the rectangular duct cross section, and
D
e
is the equivalent diameter. Note that L = W for square
ducts and that all of these dimensions are internal.
Table 7.1 shows the recommended layout for rectan-
gular ducts based on the number of traverse points from
the above fgures. For example, if 12 traverse points
are recommended, then the table suggests a 4 × 3 lay-
out. Figure 7.7 shows an example of a rectangular duct
divided into 12 equal subareas in a 4 × 3 matrix. Then,
the samples are measured in the centroid of each sub-
area as shown in the fgure.
Table 7.2 shows the recommended layout for circu-
lar ducts based on the number of traverse points from
the above fgures. For example, if 12 traverse points are
recommended, then the table suggests the location of the
sample points. Figure 7.8 shows an example of a circu-
lar duct divided into 12 equal subareas in a symmetric
matrix. Then the samples are measured in the centroid of
each subarea as shown in the fgure. The distances to each
centroid are given as a percentage of the total diameter.
The EPA Method 1A applies to ducts with equivalent
internal diameters of less than 12 in. but greater than or
equal to 4 in. The method does not apply to swirling or
cyclonic fows. If particulates are present, one or two tra-
verse points should be located at least 8 internal diam-
eters downstream and 10 internal diameters upstream
from any fow disturbances. Note that velocity measure-
ments should be made 8 internal diameters downstream
from the gas sampling location for this case. Figure 7.9
shows the recommended sampling arrangement for
small ducts. Again, Figure 7.5 should be used to deter-
mine the number of sample points for fows without
particulates and Figure 7.6 should be used for fows with
Measurement
site
Disturbance
Disturbance
Stack diameter > 0.61 m (24 in.)
Duct diameters upstream from flow disturbance* (distance A)
Higher number is for
rectangular stacks or ducts
12
0
2 3 4 5 6 7 8 9 10
M
i
n
i
m
u
m

n
u
m
b
e
r

o
f

t
r
a
v
e
r
s
e

p
o
i
n
t
s
* From point of any type of
disturbance (bend, expansion, contraction, etc)
Stack diameter = 0.30 to 0.61 m (12-24 in.)
10
20
30
40
50
0.5 1.0 1.5 2.0 2.5
Duct diameters downstream from flow disturbance* (distance B)
B
A
16
20
24 or 25
a
8 or 9
a
a
Figure 7.6
Minimum number of traverse points for sample or velocity measurements when particulates are present. (From U.S. EPA, 40 CFR 60: Standards
of Performance for New Stationary Sources, Washington, DC: U.S. Environmental Protection Agency, 2001.)
Typical centroid measurement location
Figure 7.7
Rectangular cross-sectional duct divided into 12 equal subareas.
(From Baukal, C. E., Industrial Combustion Pollution and Control, Marcel
Dekker, New York, 2004.)
Table 7.1
Recommended Measurement Layout for
Rectangular Cross-Sectional Area Ducts.
Cross Section Layout for Rectangular Stacks
(40 CFR 60: Appendices, p. 11)
# Traverse Points in Layout Recommended Matrix
9 3 × 3
12 4 × 3
16 4 × 4
20 5 × 4
25 5 × 5
30 6 × 5
36 6 × 6
42 7 × 6
49 7 × 7
Source: U.S. EPA, 40 CFR 60: Standards of Performance
for New Stationary Sources, Washington, DC:
U.S. Environmental Protection Agency, 2001.
© 2011 by Taylor and Francis Group, LLC
Pollution Emissions 147
particulates. In either case, the minimum number of
sample points is 8 for circular ducts and 9 for rectangu-
lar ducts. For circular ducts, the number of sample points
should be divisible by 4. The main difference is that the
distances from the gas sampling location and from the
velocity measurement location to the nearest upstream
and downstream disturbances should be calculated
(four total equivalent internal diameters). The distance
requiring the most number of traverse points should be
used as the minimum number of traverse points.
There are two techniques for sampling exhaust
products that are referred to as extractive and in-situ.
Mullinger and Jenkins [34] compare both techniques in
Section 8.3 of their book. These two techniques are dis-
cussed here in the next two sections.
Typical centroid measurement location
6
5
4
3
2
1
Figure 7.8
Circular cross-sectional duct divided into 12 equal subareas. (From
Baukal, C. E., Industrial Combustion Pollution and Control, Marcel
Dekker, New York, 2004.)
Table 7.2
Recommended Measurement Layout for Circular Cross-Sectional Area Ducts. Location of Traverse Points in Circular
Stacks in Percentage of Stack Diameter from Inside Wall to Traverse Point (40 CFR 60: Appendices, pp. 12–13)
Traverse Point
on a Diameter
Number of Traverse Points on a Diameter
2 4 6 8 10 12 14 16 18 20 22 24
1 14.6 6.7 4.4 3.2 2.6 2.1 1.8 1.6 1.4 1.3 1.1 1.1
2 85.4 25.0 14.6 10.5 8.2 6.7 5.7 4.9 4.4 3.9 3.5 3.2
3 75.0 29.6 19.4 14.6 11.8 9.9 8.5 7.5 6.7 6.0 5.5
4 93.3 70.4 32.3 22.6 17.7 14.6 12.5 10.9 9.7 8.7 7.9
5 85.4 67.7 34.2 25.0 20.1 16.9 14.6 12.9 11.6 10.5
6 95.6 80.6 65.8 35.6 26.9 22.0 18.8 16.5 14.6 13.2
7 89.5 77.4 64.4 36.6 28.3 23.6 20.4 18.0 16.1
8 96.8 85.4 75.0 63.4 37.5 29.6 25.0 21.8 19.4
9 91.8 82.3 73.1 62.5 38.2 30.6 26.2 23.0
10 97.4 88.2 79.9 71.7 61.8 38.8 31.5 27.2
11 93.3 85.4 78.0 70.4 61.2 39.3 32.3
12 97.9 90.1 83.1 76.4 69.4 60.7 39.8
13 94.3 87.5 81.2 75.0 68.5 60.2
14 98.2 91.5 85.4 79.6 73.8 67.7
15 95.1 89.1 83.5 78.2 72.8
16 98.4 92.5 87.1 82.0 77.0
17 95.6 90.3 85.4 80.6
18 98.6 93.3 88.4 83.9
19 96.1 91.3 86.8
20 98.7 94.0 89.5
21 96.5 92.1
22 98.9 94.5
23 96.8
24 99.9
Source: U.S. EPA, 40 CFR 60: Standards of Performance for New Stationary Sources, Washington, DC: U.S. Environmental Protection
Agency, 2001.
Gas
sample
location
Nearest
upstream
disturbance
Nearest
downstream
disturbance
Velocity
measurement
location
>8D >2D >8D
D
Gas flow direction
Figure 7.9
Recommended sampling arrangement for small ducts. (From Baukal,
C. E., Industrial Combustion Pollution and Control, Marcel Dekker, New
York, 2004.)
© 2011 by Taylor and Francis Group, LLC
148 Industrial Combustion Testing
7.3.1 extractive Sampling
Isokinetic sampling pertains to extractive gas sam-
pling where the sample is withdrawn out of the gas
stream at the same velocity as the exhaust gas stream
velocity. The EPA defnes isokinetic sampling as “sam-
pling in which the linear velocity of the gas entering
the sampling nozzle is equal to that of the undisturbed
gas stream sample point” [35]. This is particularly
important when measuring larger particulates, espe-
cially those greater than about 5 µm. Figure 7.10 shows
the three possibilities for withdrawing a sample from
the exhaust gas stream. When the sample velocity is
less than the gas stream velocity, some particles in the
gas stream fow around the probe and the measured
particulate concentration is lower than the actual
particulate concentration. When the sample veloc-
ity is greater than the gas stream velocity, then some
particles in the gas stream are preferentially drawn
into the sample probe and the measured particulate
concentration is greater than the actual particulate
concentration. The sample fow rate is normally easily
varied by either controlling the sample pump suction
Stream lines (a) (b)
(c) (d)
Stream lines
Stream lines Stream lines
Sample
Sample
Stream lines
bent at
sample point
Sample velocity too low
nonisokinetic
Too many large particles collected
Sample velocity too high
nonisokinetic
Too few large particles collected
Stream lines
bent at
sample point
Sample-tube not in line
nonisokinetic
Too few large particles are collected
Stream lines
bent on
entering
probe
S
a
m
p
l
e
Sample
Illustration of isokinetic
sampling
Concentration
of all sizes
the same here
as here
No bend
of
stream lines
Figure 7.10
Schematics of isokinetic and nonisokinetic sampling. (From Lodge, J. P., ed., Methods of Air Sampling and Analysis, 3rd ed., Chelsea, MI: Lewis
Publishers, 1977. With permission.)
© 2011 by Taylor and Francis Group, LLC
Pollution Emissions 149
rate or more typically by adjusting a control valve in
the sampling system.
The challenge in isokinetic sampling is measuring
both the sample gas velocity and the exhaust gas stream
velocity. In most cases, the exhaust gas stream velocity
varies across the exhaust duct and also varies as a func-
tion of time as changes in the combustion system cause
the exhaust gas volume to vary due to changing process
requirements. It is not suffcient to measure the exhaust
gas velocity at the location where the sample probe will
be located with, for example, a Pitot tube and assume that
the exhaust gas velocity will be constant at all times and
under all conditions. Isokinetic conditions are approxi-
mated when the static pressure is equal inside and out-
side a properly designed sample probe (see Figure 7.11).
The combined Pitot-sample probe is used to measure
the gas velocity in the vicinity of the probe and then to
adjust the sample fow rate so that the average velocity
at the sample probe inlet is equal to the gas velocity as
measured by the Pitot probe. The gas velocity is mea-
sured using the Pitot probe by determining the pres-
sure difference across the probe as shown in Figure 7.12.
This can be done by simply using a differential pressure
gauge (Figure 7.13), manometer (Figure 7.14), or differen-
tial pressure transmitter (Figure 7.15), and controlling
the sample fow rate until the measured differential
pressure is zero.
Another type of isokinetic sampling probe is shown
in Figure 7.16. While this probe is more diffcult to build
because of the small passage sizes, it is easier to oper-
ate than the probe recommended by the U.S. EPA. Two
static pressure measurements are made: one for the outer
stream and one for the inner stream. These are connected
to a differential pressure measuring device (see discus-
sion above) and the gas sample fow is adjusted until
the static pressure differential is zero, which means that
the static pressure is the same inside and outside the
probe. The static pressure holes on both the inside and
outside of the probe go around the entire circumference
at equally spaced distances apart. These holes are con-
nected to the appropriate manifold inside the probe so
that the static pressure measurements are actually aver-
aged values around the circumference of the probe in
case there are any signifcant fow deviations. The probe
tip is aerodynamically designed to minimize distur-
bances to the out exhaust gas stream fow.
Another important aspect of extractive sampling is
proper conditioning of the sample stream. If the exhaust
gas stream contains particulates, these can clog up the
sample lines. Some type of high temperature flter (see
Figure 7.17) or shield (see Figure 7.18) is often attached to
the sample probe to prevent particulates from entering
the sampling system. Figure 7.19 shows a schematic of
a typical sampling system. In order to prevent further
Sampling nozzle
Sampling nozzle
A. Bottom view; showing minimum pitot tube-nozzle separation.
Type S pitot tube
Nozzle entry
plane
Impact pressure opening
plane
Sampling probe
B. Side view, to prevent pitot tube from interfering with gas
flow streamlines approaching the nozzle. the impact pressure
opening plane of the pitot tube shall be even with or above the
nozzle entry plane.
Static pressure opening
plane
X ≥ 1.90 cm (
2
/
4
in.) for D = 1.3 cm (
1
/
2 in.)
Type S pitot tube
D
t
D
t
D
n
Figure 7.11
Isokinetic sample probe. (From U.S. EPA, 40 CFR 60: Standards of Performance for New Stationary Sources, Washington, DC: U.S. Environmental
Protection Agency, 2001.)
© 2011 by Taylor and Francis Group, LLC
150 Industrial Combustion Testing
reactions in the sampling system, the gas sample is
typically quenched as quickly as possible to “freeze”
the gas chemistry or the sample is maintained above
the dew point to prevent water from condensing out
until it reaches the water knockout unit. This is some-
times referred to as hot-wet sampling where the sample
is maintained above the dew point until the sample
reaches the condenser. Figure 7.20 shows an example
of a heated sample probe to prevent catalytic reactions
inside the probe itself. The sample is not diluted after it
has been withdrawn from the exhaust gas stream.
One common type of water knockout unit is an elec-
tronic water condenser (see Figure 7.21). Another type
uses a permeation dryer where nitrogen on one side of
a membrane removes water from the gas sample on the
other side of the membrane. The gas sample condition-
ing system can be purchased prepackaged as shown in
Figure 7.22. It is also necessary to measure the sample
fow rate to ensure that the proper fows are going to
the analyzers. Figure 7.23 shows a sampling system that
includes fow meters for measuring and controlling the
sample fow rates to the analyzers.
Another alternative to removing water from the gas
sample prior to sending it to the analyzers is to dilute
the sample so that the effective moisture content is not
a problem for the analyzers. This is sometimes referred
Temperature sensor
Type S pitot tube
Manometer
Leak-free connections
*Suggested (interference free)
pitot tube/thermocouple spacing
7.62 cm (3 in.)*
1.90 - 2.54 cm*
(0.75 - 1.0 in.)
Figure 7.12
Type S Pitot tube manometer assembly for measuring gas velocity. (From U.S. EPA, 40 CFR 60: Standards of Performance for New Stationary
Sources, Washington, DC: U.S. Environmental Protection Agency, 2001.)
Figure 7.13
Differential pressure gauge. (From Baukal, C. E., Industrial Combustion
Pollution and Control, Marcel Dekker, New York, 2004.)
Figure 7.14
Manometers for measuring differential pressure. (From Baukal, C.
E., Industrial Combustion Pollution and Control, Marcel Dekker, New
York, 2004.)
© 2011 by Taylor and Francis Group, LLC
Pollution Emissions 151
to as cool-dry sampling where the sample is greatly
diluted immediately after extraction from the exhaust
gas stack. Specially designed probes are used to sam-
ple particle-laden fows according to EPA Method 5, as
shown in Figure 7.24.
Once the gas sample has been properly conditioned, it
can be analyzed in one or more analyzers, depending on
what gases are being measured. Figure 7.25 shows a sin-
gle analyzer used to measure the oxygen concentration
in the sample. A multigas analyzer is used to measure
gases such as nitrogen oxides, sulfur oxides, oxygen,
carbon dioxide, and carbon monoxide.
There are a number of potential sampling prob-
lems [32]:
probes and lines clogging with contamination •
heated lines failing in cold climates causing •
water to freeze and block lines
Figure 7.15
Differential pressure transmitter. (From Baukal, C. E., Industrial
Combustion Pollution and Control, Marcel Dekker, New York, 2004.)
Gas sample
Outer stream
static pressure
Inner stream
static pressure
Outer stream
pressure tap
Inner stream
pressure tap
F
l
u
e

g
a
s

s
t
r
e
a
m
To gas
analyzers
Figure 7.16
Nozzle for isokinetic sampling probe (not to scale). (From Baukal, C.
E., Industrial Combustion Pollution and Control, Marcel Dekker, New
York, 2004.)
Figure 7.17
Sample frit flter. (Courtesy of Millennium Instruments, Spring
Grove, IL.)
Figure 7.18
Sample shield. (Courtesy of Millennium Instruments, Spring
Grove, IL.)
Condenser
Heated
sample
Sample
pump
Water
drain
pump
To drain
Filter
To analyzers
Flowmeter
Figure 7.19
Gas sample fow schematic.
© 2011 by Taylor and Francis Group, LLC
152 Industrial Combustion Testing
probe flter causing loss of pollutant as it passes •
through the probe media (scrubbing)
dilution probe causing temperature, pressure, •
gas density effects, and water droplet evapora-
tion when dilution air is added to the sample
gas
water entrainment •
leaks in the tubing or elsewhere in the system •
adsorption of pollutant to the wall, flter, tub- •
ing, or other components
adsorption of pollutant to the water, which is •
removed by a conditioning system
7.3.2 in-Situ Sampling
In-situ sampling involves analyzing the exhaust gases in
the exhaust stack without extracting them. In-situ ana-
lyzers are available for SO
2
, CO, O
2
, NO
x
, and CO
2
. They
are also available for measuring opacity, particulates,
and gas fow rate. In-situ sampling eliminates the need
for sample transport and conditioning that are required
in extractive sampling. There are two types of in-situ
sampling: point and path. Point sampling measures at
a specifc point or location in the exhaust stack while
path measures across the stack, giving an averaged
value. Another important advantage of in-situ sampling
is rapid response since the transport time in extractive
sampling can be signifcant.
However, there are some additional challenges of
using in-situ sensors due to the harsh conditions that
usually include high temperatures and may include
particulates and high gas fow rates. Another potential
problem is that the analytical equipment may be located
in inconvenient locations, possibly at elevated locations
on tall exhaust stacks, which can make maintenance
and troubleshooting more diffcult compared to extrac-
tive systems that are typically located in conditioned
enclosures or rooms at ground level. In-situ sampling
systems can also become plugged or clogged in exhaust
gas streams with high particulate levels.
One common type of in-situ sensor is made of zirco-
nia (ZrO
2
), which is used for wet O
2
measurements (e.g.,
see Figure 2.28). Light absorption techniques involving
UV or IR spectrometers are also used to measure gases
like CO, CO
2
, SO
2
, and NO.
7.3.3 ambient Condition effects
Bussman and Baukal [36] have shown that the ambient
environment can have a signifcant impact on the fue
gas composition, specifcally the pollutants NO
x
and
CO. The environment variables include ambient air tem-
perature, relative humidity, barometric pressure, and
wind velocity. Climatic changes in ambient air tempera-
ture and humidity can signifcantly impact the excess
O
2
in a heater or furnace. If the excess air is not properly
controlled, it can ultimately lead to dramatic changes in
NO
x
and CO emissions.
When the temperature of the ambient air changes, its
density also changes. For example, dry air at 60°F (16°C)
and 14.7 psia has a density of 0.0765 lb
m
/ft
3
(1.23 kg/m
3
).
At 100°F (38°C) the density drops to 0.0710 lb
m
/ft
3
(1.14
SP2000-H heated gas sample probe
Figure 7.20
Heated gas sample probe. (Courtesy of M&C, Ratingen, Germany.)
Figure 7.21
Electronic water condenser. (Courtesy of PERMA PURE LLC, Toms
River, NJ.)
© 2011 by Taylor and Francis Group, LLC
Pollution Emissions 153
SS-5 panel-mount sample conditioning system
Figure 7.22
Prepackaged sample conditioner. (Courtesy of M&C, Ratingen, Germany.)
CSS-3 rack-mount compact sampling system
Figure 7.23
Sampling system. (Courtesy of M&C, Ratingen, Germany.)
Figure 7.24
Method 5 sampling probe. (Courtesy of Environmental Supply, Durham, NC.)
© 2011 by Taylor and Francis Group, LLC
154 Industrial Combustion Testing
kg/m
3
). If a natural draft burner, operating at a given
draft, experiences an increase in air density due to a
decrease in temperature, it will pull more combustion
air into the heater on a mass basis, resulting in higher
excess O
2
. Conversely, if the air temperature increases,
excess O
2
would decrease.
The presence of water vapor in the combustion air can
also affect heater excess O
2
levels. As the moisture in the
air increases, the excess O
2
drops because part of the com-
bustion air is displaced by the water vapor. The amount
of water vapor that air can hold depends on the tempera-
ture and pressure. When the air can’t hold all the water,
then the moisture condenses as dew. Relative humidity
is the term commonly used to describe the amount of
water vapor in the air and is defned as the amount of
moisture in the air compared to the maximum amount
the air could hold at that temperature. Figure 7.26 shows
the effects of temperature and relative humidity on the
amount of water vapor in air. For example, at 100°F (38°C)
and 100% relative humidity, air holds about 6.5% water
vapor by volume, or about 275 grains of H
2
O per pound
of dry air (7000 grains = 1 lb
m
 = 0.454 kg).
Figure 7.27 shows theoretical results for how the O
2

concentration decreases with an increase in air tempera-
ture and relative humidity. These calculations are based
on a constant furnace draft level with 3% excess O
2
cor-
responding to air at 60°F (16°C) and 0% relative humid-
ity (baseline point). The results show that the excess O
2

drops off as the air temperature increases. Also notice
that the variation of O
2
is more dramatic with changes
in air temperature as the relative humidity increases.
Next, a couple of scenarios are given to demonstrate
how these variations can result in dramatic changes in
NOx and CO emissions.
The American Petroleum Institute (API) Publication
535 [37] states that “NO
x
concentrations will increase as
the excess oxygen increases in raw gas burners … this is
true for typical refnery heater excess oxygen (1–5% O
2
,
wet basis) rates.” Figure 7.28 shows the API 535 curve
(shaded area) demonstrating the effect of excess O
2
on
NOx emissions from raw gas burners. The NO
x
covers
a fairly broad range at a given excess O
2
. This variation
in data may be due to burner type, fuel composition,
burner design, and so on.
The API 531 curve does not characterize the effect of
excess O
2
on NO
x
emissions for all operating conditions.
For example, Figure 7.28 also shows test results of a dif-
fusion-type wall burner fring at 1.65 × 10
6
Btu/hr (0.483
MW) with two fuel compositions: (1) 100% natural gas
Figure 7.25
Oxygen analyzer. (Courtesy of AMETEK, Inc., Paoli, PA.)
0
1
2
3
4
5
6
7
8
9
30 40 50 60 70 80 90 100 110 120
Air temperature (degrees F)
V
o
l
u
m
e

%

w
a
t
e
r

v
a
p
o
r

i
n

a
i
r
0
50
100
150
200
250
300
350
400
H
u
m
i
d
i
t
y

(
g
r
a
i
n
s

p
e
r

l
b

d
r
y

a
i
r
)
20%
4
0
%
6
0
%
8
0
%
1
0
0
%
Relative
humidity
Figure 7.26
Effects of temperature and relative humidity on the amount of water
vapor in air at an atmospheric pressure of 14.7 psia (101 kPa). (From
Bussman, W., and Baukal, C., “Ambient Condition Effects on Process
Heater Emissions,” Proceedings of the Int’l Mechanical Engineering
Congress & Exhibition, Paper IMECE2008-68284, Boston, MA,
November 2008. With permission.)
2.5
2.0
1.5
1.0
0.5
0.0
3.0
3.5
30 40 50 60 70 80 90 100 110 120
Air temperature (degrees F)
P
e
r
c
e
n
t

O
2
,

d
r
y

v
o
l
u
m
e
2
0
%
4
0
%
6
0
%
8
0
%
1
0
0
%
0
%
Relative
humidity
Figure 7.27
Percent O
2
as a function of ambient air temperature and relative
humidity fring methane (or propane). (From Bussman, W., and
Baukal, C., “Ambient Condition Effects on Process Heater Emissions,”
Proceedings of the Int’l Mechanical Engineering Congress &
Exhibition, Paper IMECE2008-68284, Boston, MA, November 2008.)
© 2011 by Taylor and Francis Group, LLC
Pollution Emissions 155
(NG), and (2) 60% NG, 40% hydrogen (by volume). The
data for each fuel are corrected for dilution to 3% O
2
and
a furnace temperature of 1600°F (870°C). When fring
100% NG, the effects of O
2
on NO
x
follow the API curve
fairly closely. However, when fring with the hydrogen
mixture, the data deviate signifcantly from the API
curve. These results show that increasing the hydrogen
content increases the sensitivity of excess O
2
on NO
x
pro-
duction for this particular burner. For example, suppose
this burner is operating at 3% O
2
with NO
x
emissions
of 20 ppmvd fring both the NG and high-hydrogen
fuel cases. If the excess O
2
were to increase to 6% at a
constant furnace temperature, the NO
x
would increase
to 27 and 32 ppmvd for the NG and high-hydrogen
case, respectively. Experimental data clearly show that
increasing the excess O
2
results in an increase in NO
x
for
diffusion-style burners. The increase is attributed to the
combination of high fame temperatures and excess O
2

available to combine with N
2
to form NO
x
.
Consider a hypothetical example demonstrating
changes in NO
x
. Suppose during a plant start-up, opera-
tors tune a heater full of diffusion-style burners to 2%
excess O
2
. The time of year is midsummer with an air
temperature of 100°F (38°C) and 80% relative humidity.
At night in the late summer months, the temperature
drops to 50°F (10°C) and 0% relative humidity. If the
operators fail to trim the O
2
during this climate change,
Figure 7.29 shows the excess O
2
will increase from 2%
to approximately 4%. Using the data from Figure 7.28,
NO
x
will increase by 40% for the NG fuel case and 58%
for the high-hydrogen fuel case. This demonstrates that
dramatic changes in NO
x
can occur if excess O
2
is not
trimmed.
Carbon monoxide is produced by the incomplete
combustion of fuels containing carbon. The CO burns
by reacting with O
2
to form carbon dioxide (CO
2
). The
CO is generally easy to convert into CO
2
given suffcient
temperature, oxygen, and mixing; however, it can be
formed if there is insuffcient oxygen available for com-
plete combustion.
Figure 7.30 shows the effects of excess O
2
on CO emis-
sions for a diffusion-style wall burner, where excess
O
2
is measured at the top of the radiant section. Data
show that as the excess O
2
decreases to about 1%, CO
emissions start to increase. As the excess O
2
approaches
0.5%, the CO production rapidly accelerates, which is
often referred to as “CO breakthrough.”
Consider a hypothetical example demonstrating
changes in CO. Suppose plant operators working the
nightshift tune a heater with diffusion-style burners to
2% excess O
2
. During adjustments, the ambient tempera-
ture is 60°F (16°C) and 0% relative humidity. As midday
approaches, the weather conditions reach 100°F (38°C)
and 80% relative humidity. If the dayshift operators fail
to adjust the heater, the excess O
2
will drop to nearly
0% (see Figure 7.31). Using the data from the diffusion-
style burner in Figure 7.30, CO breakthrough will occur
resulting in high CO emissions out the stack. Operating
a heater under these conditions poses a safety risk and
can cause long-term heater problems.
Another weather-related variable that can impact
heater excess O
2
is the atmospheric (barometric) pres-
sure. Atmospheric pressure is closely approximated
by the hydrostatic pressure caused by the weight of air
above the measurement point. Low pressure areas have
less air mass above their location, whereas high pres-
sure areas have more air mass above their location. As
the atmospheric pressure changes, the air density also
changes due to the compression effect of the air mass.
Natural variations in the atmospheric pressure occur
as a consequence of changing weather conditions. For
example, an incoming storm can cause a large drop
in atmospheric pressure. As the atmospheric pressure
drops, the density of the air (at a constant temperature)
also drops. For a heater operating at a given draft, this
would result in a reduction in excess O
2
because the
reduction in air density results in less combustion air
going through the burner.
Figure 7.32 shows the effects of atmospheric pressure
on excess O
2
where the baseline point is 3% O
2
with ambi-
ent conditions of 60°F (16°C), 14.7 psia (101 kPa). Results
show that as the atmospheric pressure decreases, the
O
2
also decreases. The rate of change of excess O
2
, with
atmospheric pressure, is not signifcantly dependent on
the air temperature.
Weather data show that atmospheric pressure, at a
given location, can vary as much as 0.6 psia (4 kPa) in a
short period of time [38]. This corresponds to a change
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
0 1 2 3 4 5 6 7 8 9
Furnace O
2
(percent dry on volume basis)
R
a
t
i
o

o
f

N
O
x

a
t

n
e
w

c
o
n
d
i
t
i
o
n

t
o

b
a
s
e
l
i
n
e

c
o
n
d
i
t
i
o
n
Diffusion wall burner
60% NG (vol.), 40% H
2
Diffusion wall burner
100% NG
Pre-mix wall burner
- No secondary air
- H2 (10 to 70% vol.), bal. NG
API 535
Figure 7.28
Effects of excess O
2
on NO
x
. (From Bussman, W., and Baukal,
C., “Ambient Condition Effects on Process Heater Emissions,”
Proceedings of the Int’l Mechanical Engineering Congress &
Exhibition, Paper IMECE2008-68284, Boston, MA, November 2008.)
© 2011 by Taylor and Francis Group, LLC
156 Industrial Combustion Testing
in heater excess O
2
of about 0.5% based on the results
in Figure 7.32. A drop of 0.5% in excess O
2
can be sig-
nifcant, especially, if a burner is operating near the CO
breakthrough point.
Varying winds, especially storm force winds, can
cause dramatic changes in pressure inside the stack and
at the burner inlet. This can signifcantly impact heater
draft resulting in swings in excess O
2
. When wind blows
over the top of a stack, as illustrated in Figure 7.33, it can
contribute to either a back pressure or a suction pressure
inside the stack. If the momentum of the fue gas is low
relative to the wind, then a negative pressure inside of
the stack can be formed. Since the heater box is in direct
dynamic communication with the stack, this contributes
to an increase in heater draft. On the other hand, if the
fue gas momentum is dominant, the wind can create a
back pressure resulting in a reduction in heater draft.
Figure 7.34 shows experimental results for the effects
of wind on heater draft and excess O
2
. Data were col-
lected over a 10-minute period for a diffusion-style
burner fring NG at a heat release of 10 × 10
6
Btu/hr
(2.9 MW). Over that period, the burner damper was
full-open and the stack damper was set at 58% open.
The stack damper was exposed to the full force of the
wind while the burner inlet was somewhat isolated
from the wind due to equipment and structures located
in the area. For the high-wind case, the speed ranged
from about 8 to 35 mph (3.6 to 16 m/s) causing variations
in draft of about 0.25 in. WC (0.062 kPa) and swings in
excess O
2
of about 1%. However, for the low-wind case,
the speed varied from about 0 to 10 mph (0 to 4.5 m/s)
causing variations in draft of 0.05 inches WC (0.012 kPa)
and swings in excess O
2
of only about 0.2%. Clearly, the
wind blowing over a heater stack can have a signifcant
impact on the draft and excess O
2
. Next, consider the
effects of wind blowing past the air intake of a burner.
Wind blowing past a burner intake can have a sig-
nifcant impact on the air fow rate into the heater. The
impact depends on the burner intake design, and wind
velocity and direction. Figure 7.35 shows several burners
with a variety of intake designs. Due to the diverse air
intake designs, burners can be affected by the wind dif-
ferently. For example, the burner depicted in Figure 7.35b
will be less susceptible to a headwind as compared to
the burner in Figure 7.35c because the front plate acts as
a barrier to protect it from the force of the wind.
6
4.0
3.0
3.5
58% increase in NOx
40% increase in NOx
Furnace O
2
(percent dry on volume basis)
Diffusion wall burner
60% NG (vol.), 40% H2
Diffusion wall burner
100%NG
2.0
2.5
1.0
1.5
0.5
5
4
3
P
e
r
c
e
n
t

O
2
,

d
r
y

v
o
l
u
m
e
R
a
t
i
o

o
f

N
O
x

a
t

n
e
w

c
o
n
d
i
t
i
o
n

t
o

b
a
s
e
l
i
n
e
2
1
0
30 40 50 60 70 80
Air temperature (degrees F)
Heater tuned at these conditions:
- Temperature = 100° F
- Relative humidity = 80%
- Atmospheric pressure = 14.7 psia
Weather changes to these conditions:
- Temperature = 50° F
- Relative humidity = 0%
- Atmospheric pressure = 14.7 psia
90 100 110 120 0 1 2 3 4 5 6 7 8 9
Figure 7.29
Example demonstrating how variations in ambient air temperature and humidity can result in dramatic changes in NO
x
emissions. (From
Bussman, W., and Baukal, C., “Ambient Condition Effects on Process Heater Emissions,” Proceedings of the Int’l Mechanical Engineering
Congress & Exhibition, Paper IMECE2008-68284, Boston, MA, November 2008. With permission.)
3500
3000
2500
2000
1500
1000
500
0
0 0.5 1 1.5
Heater excess O
2
(dry volume)
C
O

c
o
r
r
e
c
t
e
d

t
o

3
%

O
2
2 2.5 3
Region of CO breakthrough
Figure 7.30
Experimental data for a diffusion-style burner showing the CO break-
through. (From Bussman, W., and Baukal, C., “Ambient Condition
Effects on Process Heater Emissions,” Proceedings of the Int’l
Mechanical Engineering Congress & Exhibition, Paper IMECE2008-
68284, Boston, MA, November 2008. With permission.)
© 2011 by Taylor and Francis Group, LLC
Pollution Emissions 157
A diffusion-style burner, fring NG at its design maxi-
mum heat release of 10 × 10
6
Btu/hr (2.9 MW), was tested
in a crosswind direction with the damper set at 100%
open. Wind speeds up to 35 mph (16 m/s) were simu-
lated using a variable speed blower positioned perpen-
dicular to the burner intake as illustrated in Figure 7.36.
During all tests, the ambient wind speed remained less
than 2 mph (0.9 m/s) and therefore, had an insignifcant
affect on the test results. Experimental data show that
for this particular burner design, a crosswind of 20 mph
(8.9 m/s) did not have a signifcant impact on the excess
O
2
; however, at 35 mph (16 m/s) it did. At a wind speed
of 35 mph (16 m/s), the excess O
2
varied from 0.5 to 2.3%
over a period of fve minutes with an average of about
1.5% excess O
2
. During this time period, the CO emis-
sions varied from 0 to about 200 ppmvd (uncorrected
for O
2
). One would expect that as the wind speed fur-
ther increased, the heater would operate in a fuel rich
condition.
Figure 7.37 demonstrates the effects of air humidity
on NO
x
emissions for two burner styles. The circular
H
e
a
t
e
r

e
x
c
e
s
s

O
2

(
p
e
r
c
e
n
t

d
r
y

v
o
l
u
m
e
)
0.0
30 40 50 60 70
Weather changes to these conditions
- Temperature air = 100° F
- Relative humidity = 80%
- Atmospheric pressure = 14.1 psia
Heater tuned at these conditions
- Temperature air = 60° F
- Relative humidity = 0%
- Atmospheric pressure = 14.7 psia
80
Air temperature (degrees F)
90 100 110 120
0.5
1.0
1.5
2.0
2.5
3.0
Region of CO breakthrough
Heater excess O
2
(percent dry volume)
0
0
500
1000
1500
C
O

c
o
r
r
e
c
t
e
d

t
o

3
%

O
2
2000
2500
3000
3500
4000
0.5 1 1.5 2 2.5 3
Figure 7.31
Example demonstrating how variations in ambient temperature and humidity can result in dramatic changes in CO emissions. (From
Bussman, W., and Baukal, C., “Ambient Condition Effects on Process Heater Emissions,” Proceedings of the Int’l Mechanical Engineering
Congress & Exhibition, Paper IMECE2008-68284, Boston, MA, November 2008. With permission.)
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
P
e
r
c
e
n
t

O
2
,

d
r
y

v
o
l
u
m
e
A
ir tem
perature =
32°F
A
ir tem
peratu
re =
60°F
A
ir tem
p
eratu
re =
100°F
Baseline point at 3% O
2
60°F, 14.7 psia
13.0 13.5 14.0 14.5 15.0
Atmospheric pressure (psia)
Figure 7.32
Calculated percent O
2
as a function of atmospheric pressure fring
methane (or propane). (From Bussman, W., and Baukal, C., “Ambient
Condition Effects on Process Heater Emissions,” Proceedings of
the Int’l Mechanical Engineering Congress & Exhibition, Paper
IMECE2008-68284, Boston, MA, November 2008. With permission.)
Wind flowing over the top of a stack
con contribute to either a heater back
pressure or a suction pressure
depending on flow rates
Wind flowing past a burner inlet can
create either a positive or negative
pressure depending on burner orientation
Wind
Flue gas
Wind
Heater
Figure 7.33
Illustration showing ways the wind can impact heater excess O
2
.
(From Bussman, W., and Baukal, C., “Ambient Condition Effects
on Process Heater Emissions,” Proceedings of the Int’l Mechanical
Engineering Congress & Exhibition, Paper IMECE2008-68284, Boston,
MA, November 2008. With permission.)
© 2011 by Taylor and Francis Group, LLC
158 Industrial Combustion Testing
symbols represent data from a large industrial heater
full of premix, radiant wall burners (data for both high
and low NO
x
burners) [39]. The triangular symbols
represent experimental data fring a single, diffusion-
style, wall burner. All of the data are normalized to a
NO
x
value corresponding to zero humidity. The NO
x

value, at zero humidity, is obtained by extrapolating
a best-ft exponential curve through the experimen-
tal data. Experimental data clearly show that for both
the premix and diffusion-style burners, NO
x
decreases
with an increase in air humidity. For example, a varia-
tion in air humidity from 0 to 120 grains corresponds
to a 45% change in NO
x
for the high NO
x
burner. As
previously discussed, this represents a change in the
W
i
n
d

s
p
e
e
d
(
m
p
h
)
40
30
20
10
0
F
l
o
o
r

d
r
a
f
t
(
i
n
c
h
e
s

W
.
C
.
)
0.75
0.50
0.25
High wind condition
E
x
c
e
s
s

O
2
(
d
r
y

v
o
l
u
m
e
)
3
2
1
0
0
W
i
n
d

s
p
e
e
d
(
m
p
h
)
40
30
20
10
0
F
l
o
o
r

d
r
a
f
t
(
i
n
c
h
e
s

W
.
C
.
)
0.75
0.50
0.25
E
x
c
e
s
s

O
2
(
d
r
y

v
o
l
u
m
e
)
3
2
1
0
0
Low wind condition
Figure 7.34
Data trends showing wind effects on heater draft and excess O
2
at high and low wind speeds over a 10-minute period. (From Bussman, W.,
and Baukal, C., “Ambient Condition Effects on Process Heater Emissions,” Proceedings of the Int’l Mechanical Engineering Congress &
Exhibition, Paper IMECE2008-68284, Boston, MA, November 2008. With permission.)
(a) (b)
Air intake
Air intake Air intake
Air intake
Air intake
Air intake
Air intake
(c)
(d) (e) (f )
Figure 7.35
Illustrations showing burner designs with various air intake confgurations. (From Bussman, W., and Baukal, C., “Ambient Condition Effects
on Process Heater Emissions,” Proceedings of the Int’l Mechanical Engineering Congress & Exhibition, Paper IMECE2008-68284, Boston, MA,
November 2008. With permission.)
© 2011 by Taylor and Francis Group, LLC
Pollution Emissions 159
theoretical fame temperature of only about 2.3%. The
data also show the percentage NO
x
reduction is more
pronounced for the high NO
x
burner as compared to
the low NO
x
burner. Based on theoretical results previ-
ously discussed, the fame temperature from a low NO
x

burner is less sensitive to variations in air humidity and
hence, changes in NO
x
.
This section has shown that ambient conditions can
signifcantly impact NO
x
and CO emissions. Tight con-
trol of burner operating conditions is essential to mini-
mize these emissions. This can be done in two ways:
manual or automatic control. Manual control means
that operators must frequently adjust dampers (air reg-
isters) on every burner as ambient conditions change.
It is not enough to adjust only some of the burners to
achieve the target O
2
level. While this could be done, the
fames would not be uniform, which could lead to mul-
tiple operational problems such as uneven heating and
fame impingement on process tubes. Some heaters may
have dozens of burners and each operator is normally
responsible for multiple heaters. Because they already
have many other responsibilities, this is not a practical
option for most plants.
The other method is automatic control where the
excess O
2
is continuously adjusted to a target level. The
two ways this is commonly done are using a control
motor on each burner (see Figure 7.38) or with some type
of mechanical linkage connecting multiple burners to a
single control motor (see Figure 7.39). The motors may
4.0
3.5
3.0
2.5
A
v
e
r
a
g
e

e
x
c
e
s
s

O
2

(
%

d
r
y
)
Excess O
2
= 0.5 to 2.3%
CO = 0 to 200 ppmvd
Average wind speed (mph)
0 10
Wind direction
Air intake
20 30 40
2.0
1.5
1.0
0.5
0.0
Figure 7.36
Experimental data showing the effects of a crosswind past a burner
intake on excess O
2
. (From Bussman, W., and Baukal, C., “Ambient
Condition Effects on Process Heater Emissions,” Proceedings of
the Int’l Mechanical Engineering Congress & Exhibition, Paper
IMECE2008-68284, Boston, MA, November 2008. With permission.)
0.4
0.5
0.6
0.7
0.8
0.9
1.0
1.1
Humidity, (grains per lb dry air)
N
O
x
,

n
o
r
m
a
l
i
z
e
d

t
o

z
e
r
o

h
u
m
i
d
i
t
y
Field data - Low NO
x
, premix, radiant wall burners
Field data - High NO
x
, premix, radiant wall burners
Test data - Wall-fired, diffusion burner
Exponential fit of Low NO
x
, premix data
Exponential fit of High NO
x
, premix data
Exponential fit of Wall-fired, diffusion data
0 20 40 60 80 100 120 140
Figure 7.37
Experimental data and model results showing the effects of air humid-
ity on NO
x
emissions. (Adapted from Bussman, W., and Baukal, C.,
“Ambient Condition Effects on Process Heater Emissions,” Proceedings
of the Int’l Mechanical Engineering Congress & Exhibition, Paper
IMECE2008-68284, Boston, MA, November 2008. With permission.)
Figure 7.38
Control motor on each burner. (From Bussman, W., and Baukal, C.,
“Ambient Condition Effects on Process Heater Emissions,” Proceedings
of the Int’l Mechanical Engineering Congress & Exhibition, Paper
IMECE2008-68284, Boston, MA, November 2008. With permission.)
Shaft connecting burners
Burners connected by shaft
Figure 7.39
Multiple burners mechanically connected by a shaft to a single con-
trol motor. (From Bussman, W., and Baukal, C., “Ambient Condition
Effects on Process Heater Emissions,” Proceedings of the Int’l
Mechanical Engineering Congress & Exhibition, Paper IMECE2008-
68284, Boston, MA, November 2008. With permission.)
© 2011 by Taylor and Francis Group, LLC
160 Industrial Combustion Testing
be electrically or pneumatically actuated. Quick discon-
nects are normally included so individual burners can
be quickly and easily manually controlled or taken out
of service if necessary.
7.4 Testing Equipment
This section briefy considers the equipment commonly
used to measure variables of importance when quanti-
fying pollution emissions. These variables include the
exhaust gas fow rate, the gas composition, and particu-
late emissions (solids in the gas stream). Note that there
are many other variations that may be used in a given
installation and new types of equipment are constantly
being developed to make these measurements easier,
more repeatable, and more accurate.
7.4.1 exhaust gas Flow
The exhaust gas fow is usually not directly measured
but is often determined through a combination of mea-
surements, calculations, and assumptions. For exam-
ple, the gas velocity is measured at one or more points
across the exhaust duct. Since it is not known a priori
if the gas sample is homogeneous, it must be assumed
that there is some variation across the exhaust gas duct
until proven otherwise. This is done by making multiple
traverses across the duct. The cross-sectional area of the
duct is subdivided into smaller areas for the purposes
of making measurements. The gas velocity in each sub-
area is assumed to be constant. Then the gas fow rate
through that subarea is simply:
Q v A
i i i
= , (7.7)
where Q
i
is the gas fow rate through sub area i, v
i
is the
gas velocity through the sub area, and A
i
is the area. The
sum of all the subareas equals the total cross-sectional
area of the duct:
A A
i
i
n
=
=

1
, (7.8)
where A is the total duct cross-sectional area and n is the
total number of subareas. Then, the total exhaust gas fow
rate equals the sum of the individual gas fow rates:
Q Q
i
i
n
=
=

1
. (7.9)
7.4.1.1 Pitot Tube
The EPA Method 2 discusses how to determine the vol-
umetric stack gas fow rate using an S-Type Pitot tube
that includes a temperature sensor (see Figure 7.40).
The method only applies to fows that are not swirl-
ing or cyclonic. This method combines measurements
and calculations to determine the gas fow rate. Specifc
guidelines are given for what materials should be used
and what dimensions are critical. Each Pitot probe must
be calibrated with a known coeffcient that should be
engraved on the probe. The static and dynamic measur-
ing holes on the probe must not be plugged, which can
be a problem in particulate-laden fows. The pressure
“head” or difference between the static and dynamic
measurements is typically very small and can be mea-
sured with an inclined manometer (see Figure 7.14) or
a similar device like a magnehelic gauge. Both of these
need to be calibrated to ensure accuracy.
The equation governing the response of a Pitot probe
is commonly given as (Becker and Brown 1974) [40]:
v
p p
st
=
− 2( )
t
p
C ρ
(7.10)
where p
t
is the total or impact pressure, p
st
is the static
pressure, and C
p
is the probe calibration constant.
According to EPA Method 2, the average stack gas veloc-
ity can be calculated for an S-Type Pitot tube using:
v K C
p p T
p M
s p p
t st s
s s
=
− ( )
, (7.11)
where
v
s
 = average stack gas velocity (ft/s)
K
p
 = velocity equation constant
C
p
 = Pitot tube probe constant
D
4 static
holes
3/8 D
Impact opening
1/2 D 4

D
1
0

D
Figure 7.40
Standard hemispherically nosed Pitot tube from 40 CFR 60 Method
2C. (From U.S. EPA, 40 CFR 60: Standards of Performance for New
Stationary Sources, Washington, DC: U.S. Environmental Protection
Agency, 2001.)
© 2011 by Taylor and Francis Group, LLC
Pollution Emissions 161
p
t
– p
st
 = difference between total and static pressures
measured with Pitot tube (in. H
2
O)
T
s
 = absolute stack temperature (°R)
P
s
 = absolute stack pressure (in. Hg)
M M B B
s d
= − + ( ) , 1 18
ws ws
(7.12)
where
M
s
 = molecular weight of wet stack gas (lb/lb-mole)
M
d
 = molecular weight of dry stack gas (lb/lb-mole)
B
ws
 = fraction by volume of water in the gas stream
(determined by Methods 4 or 5).
The EPA Method 2C concerns determination of the
exhaust gas fow rate using a standard hemispherically
nosed Pitot probe as shown in Figure 7.40. The previous
equations for the S-Type Pitot probe apply to the stan-
dard Pitot tube, although the probe constant will be dif-
ferent. Becker and Brown [40] studied the response of
Pitot probes in turbulent nonreacting gas streams. The
following terminology is used to specify the recommen-
dations of their study:
d
1
 = total pressure hole i.d.
d
2
 = probe o.d.
d
3
 = static pressure hole i.d.
l
1
  = internal length before any change in diameter
or direction of total impact tube
l
2
  = external length before any change in diameter
or direction of probe
l
3
  = length from probe tip to centerline of static
pressure hole
In that study, the following conditions were assumed:
1. The probe is long enough (l
2
/d
2
 > 6) so that the
downstream geometry has a negligible effect
on the response.
2. The internal geometry (l
1
/d
1
 > 3) has a negligible
effect on the response.
3. Compressibility effects are small (Mach
number < 0.3).
Cho and Becker [41] recommend a single static pressure
inlet.
A thermocouple or similar device can be used to
measure the gas temperature that is needed to com-
pute the gas density. A barometer is needed to mea-
sure the atmospheric pressure, also needed to compute
the gas density. In lieu of a barometer, the barometric
pressure can be obtained from the National Weather
Service.
According to EPA Method 2, the average dry stack gas
volumetric fow rate is calculated as follows:
Q B v A
T P
T P
s
s
s
= −






3600 1 ( )
ws
std
std
(7.13)
where
Q = is the volumetric fow rate in dry standard cubic
feet per hour (dscf/hr)
B
ws
 = fraction by volume of water in the gas stream
(determined by Methods 4 or 5)
v
s
 = average stack gas velocity (ft/s)
A = cross-sectional area of the stack (ft
2
)
T
std
 = standard absolute temperature = 528°R
P
s
 = absolute stack pressure (in. Hg)
T
s
 = absolute stack temperature (°R)
P
std
 = standard absolute pressure (29.92 in. Hg).
The EPA Methods 2G and 2F discuss the use of two-
and three-dimensional Pitot probes, respectively, which
are useful for swirling or cyclonic fows, but are used
infrequently in industrial combustion applications and
are therefore not discussed further here. The calibration
procedures are considerably more diffcult than those
for one-dimensional Pitot probes.
The EPA Method 2H discusses how to determine a
correction for the velocity decay near the wall in stacks
greater than 1.0 m (3.3 ft) in diameter. Two procedures
are discussed: one uses velocity measurements and the
other uses a generic adjustment factor.
7.4.1.2 Direct Measurement
The EPA Method 2A discusses the direct measure-
ment of the gas volume fow rate through pipes and
small ducts. These devices include, for example, posi-
tive displacement meters and turbine meters. This is
a much simpler and faster technique than measuring
multiple traverse points in the exhaust gas stack and
then computing the fow rate based on several different
types of measurements (pressure difference, gas tem-
perature, barometric conditions, and moisture in the
stack gases). Direct measuring devices also need to be
properly calibrated to ensure accuracy. These usually
cost much more than using Pitot tubes and manome-
ters but require far less labor to operate. They can also
mitigate the problem of varying conditions across the
exhaust duct. However, they are typically only used on
smaller diameter ducts and they may introduce a sig-
nifcant pressure drop in the exhaust system that may
require either additional fan power or a reduction in
the exhaust gas volume fow rate, which is usually not
desirable.
The EPA Method 2D discusses the measurement of
the gas volume fow rates in small pipes and ducts using
rotameters, orifce plates, or other similar devices to
measure fow rate or pressure drop. A rotameter consists
© 2011 by Taylor and Francis Group, LLC
162 Industrial Combustion Testing
of a glass vertical cylinder with a slight taper inside and
some type of foat. There are graduated markings along
the length of the cylinder. The gas fow comes in from
the bottom and causes the foat to rise proportionally
to the fow rate. Appropriate curves are then used for
the particular gas fowing through the rotameters to
determine the fow rate. Orifce plates (see Figure 7.41)
measure the pressure drop across a known and precise
hole or orifce. Using the upstream pressure and the gas
temperature and properties, the gas fow rate can then
be calculated.
7.4.1.3 Calculation
This is a simple and straightforward method for cal-
culating the average exhaust gas fow rate, which is
often of suffcient accuracy for many situations. As
will be shown, this method uses a combination of
measurements, assumptions, and calculations to deter-
mine the gas fow rate. There are also some variations
of the technique as will be shown. However, in gen-
eral, the technique involves calculating the gas fow
rate based on either the measured input fow rates, on
the measured stack gas composition, or a combina-
tion of the two as a check to make sure that the results
make sense.
In nearly all industrial combustion systems, the
fuel fow rate is measured. In many cases it must be
measured in order to calculate emissions according to
regulatory requirements. For example, some air per-
mits are based on a given maximum fuel fow rate.
For some applications that do not have continuous
emissions monitoring systems, an emission factor
may be assumed using EPA AP-42 [42]. These factors
are usually calculated based on an emission rate per
unit fuel input (e.g., lb NO
2
/10
6
Btu-input). If there
is a total emission limit on the plant, then the maxi-
mum total fuel fow rate can be calculated. This type
of permit often requires regular reports on the fuel
consumption that is used to calculate the estimated
emissions.
The combustion air fow rate is also sometimes mea-
sured. Together with the fuel fow rate, the estimated
exhaust gas fow rate can be calculated by using an O
2

measurement in the exhaust stack to estimate air infl-
tration into the furnace (assuming that the measured O
2

is higher than the calculated O
2
).
While measuring the combustion air fow may not
be necessary to calculate the exhaust gas fow, measur-
ing only the O
2
in the stack (either wet or dry) only tells
how much air has gotten into the furnace. It does not tell
where it came from so it is possible that the burners are
running fuel rich with a large amount of air infltration.
Therefore, it is usually desirable to measure the combus-
tion air fow to ensure the operating conditions of the
burners.
7.4.1.4 Non-Contact
Optical sensing systems are available for measuring the
exhaust gas fow rate. These are similar to direct mea-
surement techniques except that those are normally
thermal or mechanical, as opposed to optical. A sche-
matic of an optical fow measurement system is shown
Figure 7.42. A photograph of an optical sensor is shown
in Figure 7.43. Figure 7.44 shows an ultrasonic time of
fight instrument for non-intrusively measuring fows
in small and large stacks and ducts. The velocity mea-
surement is independent of the fue gas temperature,
pressure, and density.
7.4.2 gas Composition
The EPA Method 3 is used to determine the dry molecu-
lar weight of an exhaust gas stream. This is needed, for
example, to calculate the exhaust gas fow rate used in
calculating the pollutant emission rates. The method
specifcally refers to determining the CO
2
and O
2
concen-
trations in a gas stream. The technique uses single-point
grab sampling (batch extractive sample), single-point
integrated sampling, or multipoint integrated sampling.
The sample is typically analyzed using an Orsat ana-
lyzer. This method is not commonly used in industrial
combustion applications where instrument analyzers
are typically used.
The Method 3A concerns the determination of O
2
and
CO
2
using instrument analyzers. A typical O
2
analyzer
is shown in Figure 7.45. This method discusses test
and calibration procedures. Method 3B is specifc to
Figure 7.41
Orifce plate fow measurement system. (From Baukal, C. E., Industrial
Combustion Pollution and Control, Marcel Dekker, New York, 2004.)
© 2011 by Taylor and Francis Group, LLC
Pollution Emissions 163
gas streams that contain signifcant quantities of gases
other than O
2
, CO
2
, CO, and N
2
(excluding water, which
can be easily removed from the stream) that could affect
the gas analysis results. For example, signifcant quanti-
ties of SO
2
or HCl can interfere with CO
2
measurements
using an Orsat analysis. Method 4 is used to determine
the moisture content in an exhaust gas stream. This is
important in determining the total exhaust gas volume
fow rate. A gas sample is removed from the exhaust gas
stream at a constant fow rate. Moisture is removed from
the sample and determined either volumetrically or gra-
vimetrically. The sample probe and transport lines are
Customer data
monitoring
Output
40 CFR part 75 requires measurement
of gas flow to obtain mass emissions.
Te optical flow sensor makes a drift-
free measurement across the entire
stack or duct and calculates an average
reading without contacting the gas
stream. Tus giving a true cross-stack
flow measurement of the process.
Control panel
Stack or duct
Power
Airflow
Figure 7.42
Schematic of an optical exhaust gas fow sensing system. (Courtesy of OSI, Gaithersburg, MD.)
Figure 7.43
Optical fow sensor for measuring stack gas fow rates. (Courtesy of
OSI, Gaithersburg, MD.)
Figure 7.44
In-situ exhaust gas fow meter. (Courtesy of Teledyne Monitor Labs,
Denver, CO.)
Figure 7.45
O
2
instrument analyzer. (From Baukal, C. E., Industrial Combustion
Pollution and Control, Marcel Dekker, New York, 2004.)
© 2011 by Taylor and Francis Group, LLC
164 Industrial Combustion Testing
heat traced to prevent water from condensing out of the
sample until it reaches the condenser, which might be
an electronic device (see Figure 7.21) or an ice bath for
example. Figure 7.46 shows a schematic of the recom-
mended sampling train used to measure the moisture
content in an exhaust gas stream.
The EPA Method 6 provides procedures for measur-
ing sulfur dioxide emissions from stationary sources
where the gas sample is extracted from the exhaust
stack. Ammonia, water-soluble cations, and fuorides
cause interferences with SOx measurements. Method
6A concerns sulfur dioxide, moisture, and carbon diox-
ide measurements from fossil fuel combustion sources
by chemically separating the SO
2
and CO
2
components,
where different reagent chemicals are used. Method 6C
discusses the use of instrument analyzers to measure
sulfur dioxide emissions from stationary sources. This is
the most commonly used method in industrial combus-
tion processes. These analyzers typically use ultraviolet
(UV), nondispersive infrared (NDIR), or fuorescence
techniques. A schematic of an accepted sampling sys-
tem is shown in Figure 7.47. Method 8 discusses deter-
mining sulfuric acid and sulfur dioxide emissions from
stationary sources.
The EPA Method 7 concerns the determination of
nitrogen oxide emissions from stationary sources.
This method is based on chemical separation using
a reagent from a grab sample taken from the exhaust
stack. Methods 7A and 7B give procedures for the use of
ion chromatography and ultraviolet spectrophotometry,
respectively, for measuring nitrogen oxide emissions,
also from a grab sample. Methods 7C and 7D discuss
Filter
(either in stack
or out of stack)
Stack wall
Probe
Condenser-ice bath system including silica gel tube
Temperature
sensors
Orifice
Dry gas
meter
By-pass
valve
Main
valve
Vacuum
gauge
Vacuum
line
Check
valve
Air-tight pump
Figure 7.46
EPA Method 4 moisture sampling train: reference method. (From U.S. EPA, 40 CFR 60: Standards of Performance for New Stationary Sources,
Washington, DC: U.S. Environmental Protection Agency, 2001.)
Excess
sample vent
Needle valve
Sample
by-pass vent
3% H
2
O
2
(15 ml each)
Midget
impingers
Drying
tube
Rotameter
Dry gas
meter
Ice bath
Figure 7.47
Method 6C sampling system schematic for SO
2
measurements. (From U.S. EPA, 40 CFR 60: Standards of Performance for New Stationary Sources,
Washington, DC: U.S. Environmental Protection Agency, 2001.)
© 2011 by Taylor and Francis Group, LLC
Pollution Emissions 165
the use of the alkaline permanganate/calorimetric and
alkaline permanganate/ion chromatographic methods,
respectively, with extractive samples from the exhaust
stack. Method 7E discusses the use of instrument ana-
lyzers (see Figure 7.48) to measure NO
x
emissions from
gas samples continuously extracted from the exhaust
stack. This is the most commonly used technique in
industrial combustion applications.
The EPA Method 10 discusses measuring carbon
monoxide emissions from stationary sources from con-
tinuous samples extracted from an exhaust stack where
the sample is measured with nondispersive infrared
(NDIR) analyzer. Possible interferences include water,
carbon dioxide, and carbon monoxide. Method 10A tells
how to make certifed carbon monoxide measurements
from continuous emission monitoring systems (CEMS)
at petroleum refneries.
The EPA Method 18 discusses measurement tech-
niques for determining gaseous organic compound
emissions from exhaust stacks. The technique involves
separating the major organic compounds from the
sample and then individually quantifying the constitu-
ents using fame ionization, photoionization, electron
capture, or other appropriate techniques. Diluting the
sample is also used in certain cases to mitigate the effects
of moisture. Method 25 is a similar process for deter-
mining the total gaseous nonmethane organic (volatile
organic compounds [VOCs]) emissions from a stack.
These measurements are made on a continuous gas
sample with a fame ionization detector (FID). Carbon
dioxide and water vapor can potentially interfere with
the measurements. Figure 7.49 shows a schematic of
the recommended sampling train to determining VOC
emissions. Method 25A concerns measuring VOCs with
a fame ionization analyzer (FIA) for determining the
concentrations of samples containing primarily alkanes,
alkenes, and/or arenas. Method 25B concerns the use of
an (NDIR) analyzer to measure VOCs consisting pri-
marily of alkanes.
The EPA Method 23 discusses how to measure the
emissions of dioxins and furans from stationary sources.
A sample is withdrawn isokinetically and collected on
a glass fber flter. The sample is then separated using
high resolution gas chromatography and measured by
high resolution mass spectrometry. A schematic of an
approved gas sampling train is shown in Figure 7.50.
The condenser and adsorbent trap must be specially
designed for this application. Specifc flters, reagents,
adsorbents, and sample recovery are required. The
sample analysis is also extensive. These types of mea-
surements should be done by qualifed professionals
to ensure high accuracy, especially because of the toxic
nature of the constituents and the strict regulatory
requirements.
Figure 7.48
Typical NO
x
analyzer. (From Baukal, C. E., Industrial Combustion
Pollution and Control, Marcel Dekker, New York, 2004.)
Stack wall
Stainless steel
probe
Stainless steel
filter holder
Heated box
Dry ice
Sample
tank
Sample
tank
valve
Flow
control
valve
Manometer
Vaccum
pump
Regulating
valve
Dual range
rotameter
Thermocouples
Thermocouple
Temperature
controller
Purge valve
Sample
valve
Condensate trap
Rotameter
Figure 7.49
EPA Method 25 sampling train schematic for measuring volatile organic compounds. (From U.S. EPA, 40 CFR 60: Standards of Performance for
New Stationary Sources, Washington, DC: U.S. Environmental Protection Agency, 2001.)
© 2011 by Taylor and Francis Group, LLC
166 Industrial Combustion Testing
7.4.3 Particulates
The EPA Method 5 is used to determine particulate
emissions from stationary sources. Method 5B refers to
the determination of nonsulfuric acid particulate matter
emissions from stationary sources. Method 5D concerns
determining particulate matter emissions from posi-
tive pressure fabric flters, also known as baghouses,
where this method has particular emphasis on the
sample location. Method 5E describes the procedure for
determining particulate emissions from fberglass wool
manufacturing. Method 5F concerns nonsulfate partic-
ulate emissions from stationary sources, which includes
a modifed method for separating the sample from the
flter. Method 5I concerns the determination of low con-
centrations of particulate emissions from stationary
sources. This method differs from Method 5 with the
following modifcations: improved sample handling
procedures, use of a lightweight sample flter assembly,
and the use of low residue grade acetone to recover the
particulate matter collected on the flter.
For particulate measurements, the gas sample must
be withdrawn isokinetically (see Section 7.3.1) to ensure
that a representative sample of particles is withdrawn
from the gas stream. If the extraction rate is too high,
then the measured particulate concentration will be too
high. If the extraction rate is too low, then the measured
particulate concentration will be too low. The construc-
tion of the sampling system is also important to ensure
water and particulates do not come out of the extracted
sample until it reaches the measurement location in the
system. For example, the probe should be lined with
borosilicate or quartz glass tubing and heated to pre-
vent condensation and to prevent the particulates from
coating the walls.
The extracted sample containing particulates passes
through a glass fber flter maintained at a temperature of
120 ± 14°C (248 ± 25°F). A schematic of a typical particu-
late sampling train is shown in Figure 7.51. The sample is
extracted for a known period of time, long enough to col-
lect a sample of suffcient mass. Moisture is removed from
the particulate sample captured on the flter. The dried
sample and flter are then weighed. This is compared to
the weight of the clean flter prior to sampling. The differ-
ence in the weights then gives the weight of particulates
captured. Using the measured sample extraction time,
sample fow rate, and total exhaust gas fow rate, the par-
ticulate matter emission rate can then be calculated.
The EPA Method 9 concerns the visual determination
of opacity from emissions from stationary sources. The
opacity may be caused by particulates and/or by steam.
This method requires a certifed observer trained in
Gas
flow
Stack wall
Heated glass liner
“S” type
pilot tube
Inclined
manometer
Filter holder
Recirculation
pump
Empty
Ice
100 ml HPLC water
Empty
coarse
Silica gel
Vacuum
gauge
Vacuum
pump
P
Dry gas
meter
Fine
Calibrated
orifice Gas
exit
Inclined
manometer
Heated area
Temperature
sensor
Temperature
sensor
Temperature
sensor
Condenser
XAD-2 trap
Figure 7.50
EPA Method 23 sampling train schematic for measuring dioxins and furans. (From U.S. EPA, 40 CFR 60: Standards of Performance for New
Stationary Sources, Washington, DC: U.S. Environmental Protection Agency, 2001.)
© 2011 by Taylor and Francis Group, LLC
Pollution Emissions 167
this measurement technique. An alternative technique
known as LIDAR based on light backscattering can be
used to determine opacity remotely as described in
Alternate Method 1. Method 18 discusses a particulate
matter measurement technique that can be used in lieu
of Method 5 when it is known that there are no tem-
perature effects so that the measurements can be made
directly in the exhaust stack. Method 22 discusses the
visual determination of smoke emissions from fares.
This method does not require a certifed observer as in
Method 9 as the opacity is not quantifed. It is merely
a visual determination if smoke is present from faring
(see Chapter 28).
7.5 Source Testing
Source testing involves making “offcial” pollution emis-
sion measurements used either to determine permit
levels or to determine compliance with an air permit.
Often an outside, third-party company specializing in
making these measurements is hired. These companies
usually must be certifed by a recognized agency to
make these measurements. The company will usually
bring a truck to the site equipped with the appropriate
measurement and calibration equipment. This source
testing must be done in applications where there is no
certifed continuous emission monitoring system. It
may also be done even where there is a CEMS as an
independent verifcation. It is also used by equipment
vendors trying to certify new technology. For example,
a burner manufacturer may hire a frm to make off-
cial measurements to demonstrate reduced pollution
emissions compared to other technologies. Figure 7.52
shows an example of measurements being made from
an exhaust stack.
The frequency of source testing depends on a number
of factors. Measurements are usually made when an air
permit is being established for the frst time and when-
ever signifcant modifcations are made to an existing
permitted system. Annual verifcation tests may also
be required, especially where no emissions monitoring
system exists. These verifcation tests may not require
“offcial” third-party measurements, depending on
the rules of the given governing agency. An inspector
may make unannounced spot checks with a portable
Temperature sensor
Temperature
sensor
Temperature
sensor
Temperature
sensor
Temperature
sensor
Probe
Type S pitot
tube
Type S pitot
tube
Gooseneck
nozzle
Heat traced
glase-lined
probe
Glass filter
holder
Heated area
Stack
wall
Manometer
Orifice
Water Empty Silica
gel
Dry gas
meter
By-pass
valve
Main
valve
Air-tight
pump
Vacuum
gauge
Vacuum
line
Check
valve
Impingers
Ice
water
bath
Impinger train optional, may be replaced
by an equivalent condenser
Figure 7.51
Method 5 particulate sampling train schematic. (From U.S. EPA, 40 CFR 60: Standards of Performance for New Stationary Sources, Washington,
DC: U.S. Environmental Protection Agency, 2001.)
© 2011 by Taylor and Francis Group, LLC
168 Industrial Combustion Testing
analyzer to verify compliance, which is discussed below
in Section 7.7.
An alternative to source testing is to measure the
emissions of the burner in a test facility (see Figure 7.53).
Figure 7.54 shows a pilot-scale test furnace that could
be used for regulatory source monitoring. While this is
not always permissible, depending on the rules of the
local regulatory agency, it may be used to establish per-
mit limits. It may also be used in lieu of expensive feld
source tests that may not be economically practical for
smaller emission sources.
Figure 7.52
Measuring emissions from an exhaust stack.
Figure 7.53
John Zink R&D Test Center. (Courtesy of John Zink Co., LLC.)
Figure 7.54
Test furnace to simulate a section of an ethylene cracking furnace.
(Courtesy of John Zink Co., LLC.)
© 2011 by Taylor and Francis Group, LLC
Pollution Emissions 169
7.6 Continuous Monitoring
Depending on regulations and the size of the emission
source, some industrial combustion processes may be
required to have a continuous emissions monitoring
system. For example, some air permits may be writ-
ten in such a way as to limit emissions over some time
interval, such as a 24-hour rolling average. Another
common limit is that emissions of a particular pollutant
may not exceed a given concentration at any time. There
are many other variations as well. In order to determine
whether a plant is in compliance, a certifed continuous
emission monitoring system may be required. These
are not only fairly expensive to purchase, but also have
rigorous and ongoing maintenance and calibration
requirements to ensure high accuracy and reliability.
This is compared to other pollutant emission sources
that may only require an annual test. In some cases,
only an initial test may be required after the equipment
is frst started or after any signifcant modifcations
have been made.
The U.S. EPA has developed a series of performance
specifcations (PS) for continuous emissions monitoring
systems (see Figure 7.55) for measuring a wide range of
pollutant emissions [43]. These specifcations are listed
in Table 7.3. Note that two of the specifcations (4 and
4A) have identical titles. PS-4 is for the general mea-
surement of CO emissions while PS-4A is for measur-
ing lower concentrations (< 200 ppmv) of CO emissions.
These are to be used in conjunction with the EPA meth-
ods discussed above. The performance specifcations
include discussions of calibration procedures and rela-
tive accuracy requirements. The EPA has also developed
quality assurance procedures to be used for compliance
determination [44].
Parker (1995) lists the following factors as important
in a continuous monitoring program [45]:
1. knowledge of the process to be monitored
2. knowledge of the relevant regulatory requir-
ements
3. using the correct monitors
4. using appropriately trained personnel with the
proper calibration, quality control, and quality
assurance procedures
5. relevant data collection, review and repor-
ting
6. ongoing maintenance program
7. comprehensive cost tracking, control, and
budgeting
The three major components of a continuous emis-
sions monitoring system are: (1) the sampling and
conditioning system, (2) the gas analyzers and moni-
tors, and (3) the data acquisition system (DAS). The
frst two have been discussed above. The DAS may
be as simple as manually recording emissions data
at scheduled and regular time intervals. It may be as
sophisticated as a completely automated computer-
recorded data management system. It may also be
some combination of the two where some of the data
is manually recorded while other data is automati-
cally recorded by a computer or some similar type of
electronic device. For example, some of the emissions
such as particulate emissions may not be easily col-
lected on a continuous basis while other emissions
such as NO
x
are easily collected through continuous
monitoring.
Figure 7.55
Continuous emission monitoring system. (Courtesy of Horiba
Instruments, Inc., Irvine, CA.)
© 2011 by Taylor and Francis Group, LLC
170 Industrial Combustion Testing
7.7 Spot Checking
Spot checks may be made by the frm owning the indus-
trial combustion equipment or by inspectors checking
compliance with air permits. Portable analyzers are usu-
ally used to make spot check measurements of exhaust
gas emissions. Figure 7.56 shows an example of a portable
kit used for making feld measurements. The kit includes
the analyzer, the sample probe, a sample conditioner,
and the appropriate tubing and connectors for the sys-
tem. This can be easily transported to a site and quickly
assembled for portable measurements. Figure 7.57 shows
an example of a similar portable measurement analysis
system. A more sophisticated “portable” analyzer will
generally have higher accuracy and repeatability com-
pared to other smaller analyzer systems. However, it is
not nearly as portable and is less convenient, for exam-
ple, for carrying on airplanes to remote sites.
7.8 Example Measurements
7.8.1 Oxygen-enhanced Combustion
7.8.1.1 Experimental Setup
Experiments using different probes were conducted in
Air Products’ combustion laboratory furnace shown in
Figure 7.58 [46]. The furnace design has been described
elsewhere [47]. Briefy, the furnace can accommodate
one burner fring up to 10 × 10
6
Btu/hr. The furnace
maximum operating temperature is 1800 K (2800°F) and
can be somewhat controlled by means of steam genera-
tion in sidewall water panels. There is a damper at the
furnace exit allowing furnace pressure control. For the
probe emission tests, the furnace was kept at a positive
Figure 7.56
Portable gas analyzer (Testo 330). (Courtesy of ETA, Garner, NC.)
Table 7.3
EPA Performance Specifcations for Continuous Emission Monitoring System (40 CFR 60: Appendix B)
Specifcation # Specifcation Title
1 Specifcations and test procedures for continuous opacity monitoring systems in stationary sources
2 Specifcations and test procedures for SO
2
and NO
x
continuous emission monitoring systems for stationary sources
3 Specifcations and test procedures for O
2
and CO
2
continuous emission monitoring systems for stationary sources
4 Specifcations and test procedures for carbon monoxide continuous emission monitoring systems for stationary sources
4A Specifcations and test procedures for carbon monoxide continuous emission monitoring systems for stationary sources
4B Specifcations and test procedures for carbon monoxide and oxygen continuous emission monitoring systems for stationary
sources
5 Specifcations and test procedures for TRS (total reduced sulfur) continuous emission monitoring systems for stationary
sources
6 Specifcations and test procedures for continuous emission rate monitoring systems for stationary sources
7 Specifcations and test procedures for hydrogen sulfde continuous emission monitoring systems for stationary sources
8 Specifcations and test procedures for volatile organic compound continuous emission monitoring systems for stationary
sources
8A Specifcations and test procedures for total hydrocarbon continuous emission monitoring systems for stationary sources
9 Specifcations and test procedures for gas chromatographic continuous emission monitoring systems for stationary sources
15 Specifcations and test procedures for extractive FTIR (Fourier transform infrared) continuous emission monitoring
systems for stationary sources
Source: U.S. EPA, 40 CFR 60: Standards of Performance for New Stationary Sources, Washington, DC: U.S. Environmental Protection Agency,
2001.
© 2011 by Taylor and Francis Group, LLC
Pollution Emissions 171
pressure to minimize air infltration. The average tem-
perature was about 1380 K (2030ºF).
A K-Tech burner, shown in Figure 7.59, was fred at
about 0.64 MW (2.2 × 10
6
) Btu/hr. During the experi-
ments the oxygen/natural gas volumetric ratio was var-
ied between 1.8 (fuel rich) and 2.2 (fuel lean).
A permanent CEMS and a portable handheld unit
Enerac 2000 were used in these experiments. The CEMS
is a permanently installed sampling system for monitor-
ing the exhaust gas composition at the exit of the furnace.
It consists of a water-cooled stainless steel probe, heat-
traced transport lines, water removal system, and gas
analyzers. NO is measured with a chemiluminescent and
nondispersive infrared (NDIR) analyzers. O
2
is measured
with a paramagnetic analyzer. CO is measured with an
NDIR analyzer and a gas chromatograph. The Enerac
2000 analyzer is a portable unit for feld measurements of
combustion gases. It is capable of measuring concentra-
tions of NO, CO, O
2
, SO
2
, and combustibles in gas stream
temperatures up to 1400 K (2000ºF). Electrochemical cells
are used to determine the gas concentrations. This type
of instrument may not be accepted for offcial regulatory
monitoring. However, it may be used in observing trends
and establishing relationships between different combus-
tion variables. The original probe for the Enerac 2000 was
a 9.5 mm (3/8 in.) o.d., 1 m (3 ft) long tube made of Inconel.
This tube had a thermocouple mounted at the end that
enables monitoring of the sampling gas temperature as
well as temperature of that end of the Inconel tube. In
addition to this tube, a quartz tube of similar dimensions
was used for the experiments.
The Enerac 2000 analyzer probes were inserted in the
furnace through a port on the 46 cm (18 in.) thick wall
(below and to the right of the K-Tech burner). The exper-
iment was designed to allow quick changes between the
tubes. The insertion length of the tubes was varied dur-
ing the experiments allowing rough probe temperature
control, as shown in Figure 7.60.
Figure 7.57
Portable gas analyzer. (Courtesy of ECOM, Gainsville, GA.)
Figure 7.58
Air Products’ combustion laboratory furnace. (Slavejkov, A. G., and
Baukal, C. E., Transport Phenomena in Combustion, Taylor & Francis,
Boca Raton, FL, 1996.)
Oxygen
Fuel
Fuel
Side view
Front view
Figure 7.59
K-Tech burner. (Slavejkov, A. G., and Baukal, C. E., Transport Phenomena
in Combustion, Taylor & Francis, Boca Raton, Florida, 1996.)
Sampling port position varied
between –14 and 12 inches with
respect to hot face of the wall
Furnace
Wall = 18 in
Inconel tube
Quarts tube
Refractory plug
Enerac 2000
Figure 7.60
Sampling setup with Enerac 2000 analyzer. (Slavejkov, A. G., and
Baukal, C. E., Transport Phenomena in Combustion, Taylor & Francis,
Boca Raton, FL, 1996.)
© 2011 by Taylor and Francis Group, LLC
172 Industrial Combustion Testing
In parallel with the Enerac 2000 measurements,
the fue gas was monitored using the standard CEMS
technique. This allowed comparison between results
obtained with different instruments.
7.8.1.2 Results
The results of the measurements are given in Table 7.4
through Table 7.6. Note that the probe position indicates
the distance of the probe tip from the hot face of the fur-
nace wall. Thus, the positions greater than 0 show that
the tip extends inside the furnace. The measurements
shown in the tables were made in the sequence from left
to right.
For the fuel lean experiments, it may be seen from
Table 7.4 that the Enerac NO and O
2
readings were by
about 30% higher than the CEMS readings. There was
practically no variation between the results obtained
using different tubes. The Enerac CO measurements
showed zero except for the frst reading, which was well
below the 575 ppm CO measured with the CEMS.
For the fuel rich experiments shown in Table 7.5, the
NO readings with the Inconel probe were much lower
than the other readings. Also, there is a trend between
Table 7.4
Measurements at Oxygen: Natural Gas Ratio of 2.2 (Fuel Lean)*
Analysis System Enerac 2000 CEMS
Probe material Inconel Inconel Inconel Inconel Quartz Quartz Stainless
Position (inches) 0 –8 0 –4 6 12 in the stack
O
2
(%) 5.4 4.4 4.5 5.0 6.4 7.5 3.8
NO (ppmvd) 425 430 445 445 421 410 300
CO (ppmvd) 200 0 0 0 0 0 575
Probe temp (ºF)  > 2000 1972  > 2000  > 2000  > 2000  > 2000 water-cooled
*T
furnace average
 = 2030ºF. Firing rate = 2.2 × 106 Btu/hr.
Source: From Slavejkov, A. G., and Baukal, C. E., Transport Phenomena in Combustion, Taylor & Francis,
Boca Raton, FL, 1996.
Table 7.5
Measurements at Oxygen: Natural Gas Ratio of 1.8 (Fuel Rich)*
Analysis System Enerac 2000 CEMS
Probe material Inconel Inconel Inconel Inconel Quartz Quartz Stainless
Position (inches) –4 12 –10 –14 12 12 in the stack
O
2
(%) 0 0 0 0 0 0 1.5
NO (ppmvd) 11 7 30 107 214 170 210
CO (ppmvd) over over over over over over 42,000
Probe temp (ºF)  > 2000  > 2000 1515 1310  > 2000  > 2000 water cooled
*T
furnace average
 = 2040ºF. Firing Rate = 2.2 × 106 Btu/hr.
Source: From Slavejkov, A. G., and Baukal, C. E., Transport Phenomena in Combustion, Taylor & Francis,
Boca Raton, FL, 1996.
Table 7.6
Transitional Measurements*
Analysis System Enerac Enerac CEMS Enerac Enerac CEMS
Probe Material Inconel Quartz Stainless Inconel Quartz Stainless
Position (inches) 12 12 in the stack 12 12 in the stack
O
2
(%) 0.5 1.2 0.4 1.1 3.5
NO (ppmvd) 240 327 233 334
CO (ppmvd) 0 0 2800 0 0 2000
Probe temp (ºF)  > 2000  > 2000  > 2000  > 2000  > 2000 water cooled
*T
furnace average
 = 2040ºF. Firing Rate = 2.2 × 106 Btu/h. Transition from oxygen: natural gas ratio of 1.8 (fuel
rich) to 2.2 (fuel lean).
Source: From Slavejkov, A. G., and Baukal, C. E., Transport Phenomena in Combustion, Taylor & Francis,
Boca Raton, FL, 1996.
© 2011 by Taylor and Francis Group, LLC
Pollution Emissions 173
the Inconel probe temperature and the NO readings.
That is, at T
probe
 < 1400 K (2000ºF) the NO readings were
signifcantly greater than at T
probe
 > 1400 K (2000ºF). For
the quartz probe, however, the NO readings were much
higher even though the probe was inside the furnace for
both measurements. The Enerac O
2
readings were all at
zero and the CO was off the scale ( > 5000 ppm). At the
same time, the CEMS showed 1.5% O
2
, about the same
NO as the quartz tube and 4.2% CO.
Following these two sets of measurements, the oxy-
gen/natural gas ratio was set to 2.2 (fuel lean). The data
shown in Table 7.6 were taken during the transitional
period when the furnace CO was being reduced from
the initial very high reading of 42,000 ppm. The read-
ings shown in the table were again taken in the sequence
from left to right within fve minutes from adjusting the
fow controls.
It may be seen from Table 7.6 that a reproducible dif-
ference in the NO and O
2
readings was detected for
the Inconel and quartz tubes. The Inconel tube always
showed lower NO and O
2
readings. The NO readings
with the CEMS were not made for these transitional
measurements because of the time limitation.
7.8.1.3 Discussion
The discrepancy between the Enerac 2000 readings with
the Inconel and quartz tubes, as well as its temperature
dependence, clearly indicates that the probe material
plays a key role. From the above data, it may be con-
cluded that Inconel is a good catalyst of the following
reactions at temperatures above 980 K (1300ºF).
CO + NO → CO
2
 + N (7.14)
CO + O
2
→ CO
2
 + O (7.15)
The quartz tube has also catalyzed these reactions but
with lower effciency.
For the fuel lean case in Table 7.4, there was much more
O
2
and NO than CO. In other words the excess quan-
tities of oxidizers in reactions 7.14 and 7.15 resulted in
complete consumption of CO. The difference in the NO
and O
2
readings between the Enerac 2000 and the CEMS
was probably due to the calibration, instrument accu-
racy or the difference in probe location in the furnace.
Here, the NO readings with the Enerac 2000 were con-
sistently higher than those obtained with the CEMS.
The fuel rich case shown in Table 7.5, indicates that the
4.2% CO furnace concentration was able to completely
eliminate NO when sampled through the Inconel tube.
However, the NO started increasing after the Inconel tube
temperature was lowered by pulling it out of the furnace.
This temperature dependence is shown in Figure 7.61.
The higher NO readings at lower temperatures, espe-
cially below 1100 K (1500ºF), indicate a decrease in the
Inconel catalytic effciency for the reactions 7.14 and 7.15.
However, even at about 980 K (1300ºF), the Enerac NO
readings were still much lower than those obtained with
the CEMS. The quartz tube showed about equal read-
ings of NO as the CEMS. However, in view of the frst
data set that showed NO consistently higher, it appears
that the reactions 7.14 and 7.15 were weakly catalyzed in
the quartz tube too, which lowered the NO readings.
The last transitional data set with fuel lean condi-
tions, in Table 7.6, showed a reproducible indication of
this catalytic activity. The 2000–2800 ppm of CO was not
enough to consume all of O
2
and NO. This time, the oxi-
dizers were in excess.
7.8.1.4 Conclusions and Recommendations
The above example has demonstrated some of the prob-
lems one can expect in using a hot probe for sampling
gases in high temperature furnaces away from the fame.
For reducing conditions inside a furnace, the measure-
ments may show much lower NO readings. Similarly, if
the furnace is operating under oxidizing conditions the
measurements will show much lower CO readings. The
experiments have shown that both metal and quartz
uncooled probes can affect the readings provided the
probe surface temperature exceeds 980 K (1300ºF).
To avoid the surface catalytic effects, a water-cooled
probe must be used when sampling high temperature
combustion products. This is of particular importance
to oxyfuel combustion where measured species con-
centrations are much higher due to the elimination of
nitrogen. Recommended probe materials and cooling
requirements to avoid reaction of the various gases
inside the probe are given elsewhere [48].
7.8.2 Fuel Composition effects
The composition of the fuel supplied to a combustion
system has a signifcant impact on the NO
x
emissions.
In the petrochemical and chemical process industries,
120
100
80
60
N
O

r
e
a
d
i
n
g
s
,

p
p
m
v
d
40
20
0
1200 1300 1400 1500 1600
Temperature, deg F
1700 1800 1900 2000 2100
Figure 7.61
Temperature dependence of NO readings for Enerac 2000 inconel
probe at reducing furnace conditions.
© 2011 by Taylor and Francis Group, LLC
174 Industrial Combustion Testing
there is a very wide range of fuel blends used for process
heating. These fuels are often by-products from a refn-
ing process. They typically contain hydrocarbons rang-
ing from C1 to C4, hydrogen, and inert gases like N
2
and
CO
2
. In a given plant or refnery, burners used in process
heaters may need to be capable of fring on multiple fuels
that are present at different times (e.g., start-up, normal
operation, upset conditions, etc.). In many cases, the NO
x

emissions from the heaters may be limited by regula-
tions regardless of what fuel composition is being fred.
Therefore, it is critical that the effects of the fuel compo-
sition on NO
x
emissions be understood and quantifed
to ensure that permitted values are not exceeded. This
study [49] investigated the effects of fuel composition
on NO
x
emissions from an industrial-scale burner. The
acquired data provided additional insight into the effects
on NO
x
over the range of fuel compositions consisting
of various fractions of three primary components: H
2
,
C
3
H
8
, and CH
4
. The results quantify 3-way fuel compo-
nent effects on NO
x
emissions and provide insight into
fundamental parameters involved and how they inter-
act, which provides information on which to base future
research, design, and development work.
7.8.2.1 Test Setup
To demonstrate the strong dependence of NO
x
on fuel
gas composition, experiments were conducted using an
industrial burner fred into a pilot-scale furnace. Tests
were conducted using a conventional-type burner (see
Figure 7.62) with a single fuel gas tip and fameholder.
The burner was fred vertically upward in a rectangu-
lar furnace (see Figure 7.63). The test furnace was a rect-
angular heater with internal dimensions of 2.4 m (8 ft)
wide, 3.7 m (12 ft) long, and 4.6 m (15 ft) tall. The furnace
was cooled by a water jacket on all four walls. The inte-
rior of the water-cooled walls was covered with varying
layers of refractory lining to achieve the desired furnace
temperature.
A velocity thermocouple (also known as a suction
thermocouple or suction pyrometer; see Chapter 5) was
used to measure the furnace and stack gas temperatures.
The furnace draft was measured with an automatic,
temperature-compensated, pressure transducer as well
as an inclined manometer connected to a pressure tap in
the furnace foor. Fuel fow rates were measured using
(a) (b)
Figure 7.62
Raw gas burner: (a) side view, (b) top view. (From Baukal, C. E., The John Zink Combustion Handbook. Boca Raton, FL, CRC Press, 2001.)
Figure 7.63
Test Furnace 1 at the John Zink Co., R&D test center. (From Baukal, C. E.,
The John Zink Combustion Handbook. Boca Raton, FL, CRC Press, 2001.)
© 2011 by Taylor and Francis Group, LLC
Pollution Emissions 175
calibrated orifce meters (see Figure 7.41), fully corrected
for temperature and pressure. Emission levels were
measured using state-of-the-art continuous emissions
monitors (CEMs) to measure emissions species concen-
trations of NO
x
, CO, and O
2
.
The experimental matrix consisted of fring the burner
at a constant heat release (2.2 MW or 7.5 × 10
6
Btu/hr)
and excess air level (15%) with 15 different fuel blends
comprised of varying amounts of H
2
, C
3
H
8
, and Tulsa
Natural Gas (TNG).* For testing and analysis purposes
TNG was treated as a single fuel component for conve-
nience. The TNG, which is comprised of approximately
93% CH
4
, is a more economical choice than pure CH
4

for experimental work and the analysis is simplifed
by treating it as a single component. All 15 fuel com-
positions were tested on each of six different fuel gas
tips, which differed in port diameter sizes, to enable the
acquisition of additional information regarding effects
resulting from differing fuel pressures.
7.8.2.2 Sample Results
Figure 7.64 shows the variation in relative measured
NO
x
emissions resulting from different concentrations
(volume basis) of H
2
in a fuel blend composed with a
balance of TNG for each of the six different fuel gas tips
*

The nominal composition by volume of TNG is: 93.4% CH
4
, 2.7%
C
2
H
6
, 0.60% C
3
H
8
, 0.20% C
4
H
10
, 0.70% CO
2
, and 2.4% N
2
.
tested. The plot, which illustrates NO
x
levels on a con-
centration basis, clearly shows the correlation between
increased H
2
content and higher NO
x
emission levels.
The slope of the profle is exponentially increasing,
qualitatively similar to that predicted by the plotted the-
oretical calculations shown in reference [49]. The effect
of H
2
is signifcant, with the sharpest increase in NO
x

levels taking place as concentration levels of H
2
in the
fuel mixture rise from 75% to 100%.
The variation in relative measured NO
x
emissions
resulting from different concentrations (volume basis)
of C
3
H
8
in a fuel blend composed with a balance of TNG
is shown in Figure 7.65. The slope of the increase in NO
x

levels corresponding to increased concentrations of
C
3
H
8
is shown to be relatively constant or slightly declin-
ing over the gradient in C
3
H
8
concentration, in contrast
with the exponentially increasing profle of the H
2
-TNG
plot-in. The profle showing the effect of C
3
H
8
content is
also seen to be similar to the corresponding calculated
trends shown in Hayes and colleagues research [49].
Figure 7.66 shows the fnal two-component fuel blend
results being examined, which describe the variation in
relative measured NO
x
emissions resulting from differ-
ent concentrations (volume basis) of H
2
in a fuel blend
composed with a balance of C
3
H
8
. The upper plot, which
shows measured relative NO
x
on a volume concentration
basis, illustrates that for a given tip, geometry, and port
size, the measured NO
x
concentrations actually decrease
Tip 5 Tip 4
Tip 3 Tip 2 Tip 1
Tip 6
40%
50%
60%
70%
0% 20% 40% 60% 80% 100%
H
2
in TNG (vol. %)
R
e
l
a
t
i
v
e

N
O
x

(
%

o
f

m
a
x
.
,

p
p
m
v

b
a
s
i
s
)
90%
80%
100%
Figure 7.64
Measured NO
x
(percentage of the maximum ppmvd value) as a func-
tion of the fuel blend composition for H2/TNG blends combusted
with 15% excess air where both the fuel and the air were at ambi-
ent temperature and pressure. (From Baukal, C. E., The John Zink
Combustion Handbook. Boca Raton, FL, CRC Press, 2001.)
40%
50%
60%
70%
80%
90%
100%
0% 20% 40% 60% 80% 100%
C
3
H
8
in TNG (vol. %)
R
e
l
a
t
i
v
e

N
O
x

(
f
r
a
c
t
i
o
n

o
f

m
a
x
.
,

p
p
m
v

b
a
s
i
s
)
Tip 1 Tip 2 Tip 3
Tip 4 Tip 5 Tip 6
Figure 7.65
Measured NO
x
(percentage of the maximum ppmvd value) as a func-
tion of the fuel blend composition for C
3
H
8
/TNG blends combusted
with 15% excess air where both the fuel and the air were at ambi-
ent temperature and pressure. (From Baukal, C. E., The John Zink
Combustion Handbook. Boca Raton, FL, CRC Press, 2001.)
© 2011 by Taylor and Francis Group, LLC
176 Industrial Combustion Testing
slightly with increasing H
2
content up to 75% H
2
content,
then sharply increase with H
2
concentration.
Due to the decrease in total dry products of combus-
tion from the burning of H
2
expressing NO
x
in terms
of concentration (ppmvd) does not fully represent the
actual mass rate of NO
x
emissions produced. The lower
plot, which shows the variation in measured NO
x
lev-
els on a mass per unit heat release basis, illustrates that
the overall emissions of NO
x
on a mass basis decrease
with increasing fuel hydrogen content and continue to
decrease or remain relatively fat even in the high-hy-
drogen content region that produced a sharp increase in
NO
x
levels on a volume concentration basis.
7.8.2.3 Conclusions
This study showed experimentally that NO
x
is highly
dependent on the fuel blend composition. Adiabatic
fame temperature and fuel pressure are both identi-
fed as signifcant fundamental parameters affecting
NO
x
emission levels when considering the effect of fuel
composition on NO
x
levels. For a conventional burner,
with NO
x
on a concentration basis, the adiabatic fame
temperature is dominant, with fuel pressure remaining
signifcant in affecting NO
x
emission levels. The high-
est NO
x
levels on a volume concentration basis occurred
at the highest hydrogen content fuel compositions at
lower fuel pressures. On a mass per heat release basis
however, the highest relative NO
x
levels were achieved
for fuel compositions containing large fractions of C
3
H
8
.
This appears to result from some combined characteris-
tics of a high-propane mixture including: very low fuel
pressure for a given heat release in comparison with
the other fuels; somewhat higher adiabatic fame tem-
perature than CH
4
; and a substantially larger amount of
total dry products of combustion produced for a given
heat release when compared with H
2
. In summary, the
results of this work provide both quantitative and quali-
tative information to improve emission performance
prediction and design of burners with application to a
wide variation of fuel compositions.
7.8.3 ghost NO
x
Field NO
x
measurements have been made in a number
of process heaters, at locations before and after the con-
vection sections. In nearly all cases, NO
x
readings after
the convection section have been higher than before the
convection section. In some cases, the increase in NO
x

has been as much as 50%. This phenomenon has been
lightheartedly referred to as “ghost NO
x
” because of its
unknown origin. It indicates that either additional NO
x

is being formed in the convection section, that there
is some systematic measurement problem, or some
60%
65%
70%
75%
80%
85%
90%
95%
100%
0% 20% 40% 60% 80% 100%
R
e
l
a
t
i
v
e

N
O
x
(
%

o
f

m
a
x
.
,

l
b
/
1
0
6

B
t
u

b
a
s
i
s
)
60%
65%
70%
75%
80%
85%
90%
95%
100%
0% 20% 40% 60% 80% 100%
H
2
in C
3
H
8
(vol %)
Tip 1 Tip 2 Tip 3 Tip 4 Tip 5 Tip 6
R
e
l
a
t
i
v
e

N
O
x
(
%

o
f

m
a
x
.
,

p
p
m
v

b
a
s
i
s
)
Figure 7.66
Measured NO
x
(percentage of the maximum value in both ppmv and lb/10
6
Btu) as a function of the fuel blend composition for H
2
/C
3
H
8
blends
combusted with 15% excess air where both the fuel and the air were at ambient temperature and pressure.
© 2011 by Taylor and Francis Group, LLC
Pollution Emissions 177
combination of the two. Extensive pilot scale measure-
ments have also been made to study this phenomenon.
These suggest that the nitrogen oxide (NO
x
 = NO + NO
2
)
readings can be signifcantly different depending on the
sample location within the furnace. This section dis-
cusses the feld and laboratory measurements, discusses
possible sources of the ghost NO
x
, and recommends fur-
ther work to study this unusual phenomenon.
7.8.3.1 Introduction
Environmental regulations continue to reduce the
allowable NO
x
emissions from process heaters [17]. This
has led to the development of ultra low NO
x
burners
[50,51]. One important issue has become the measure-
ment of emissions as the levels get lower as slight errors
are signifcant, because they are now a much higher
proportion of the overall emission level. This section
considers the potential error associated with the overall
NO
x
(NO + NO
2
) emissions from process heaters.
Field data suggests that the nitrogen oxide (NO
x
 = 
NO + NO
2
) readings can be signifcantly different depend-
ing on the sample location within the furnace. For exam-
ple, feld data shows that the NO
x
emissions in the furnace
stack can be as much as 50% higher than for a sample
extracted before the convection section. This anomaly is
referred to here as ghost NO
x
. The purpose of this section
is to discuss possible reasons why this apparent anomaly
exists and to present results of data collected in the feld at
an ethylene cracking refnery, and in a pilot-scale test fur-
nace that attempt to simulate this occurrence. Although
not the subject of this section, this phenomenon has also
been observed in industrial boilers during the develop-
ment of ultra low NO burner technology.
There are many possible hypotheses for why the ghost
NO
x
anomaly might occur.
1. NO
x
meter does not measure NO
2
, only NO
2. NO
2
readily absorbs into water during
sampling
3. NO
2
catalytically decomposes in the convection
section
4. Thermal NO
x
formation increases due to longer
residence time
5. NO
x
emissions are not uniformly distributed in
the radiant section
6. NO
2
to NO reduction converter in analyzer is
not effcient
7. NO
x
formation chemistry produces additional
NO
x
in the convection section
One or a combination of these could be a culprit in the
NO
x
anomaly. Each of these hypotheses is discussed
elsewhere [52].
7.8.3.2 Test Description
Lab tests were performed in a pilot-scale test furnace
with the following arrangement:
Two (2) foor burners and two (2) radiant wall •
burners
Total heat release • ~ 4.30 MW (14.7 × 10
6
Btu/hr)
Percentage O •
2
in the exhaust products, variable
Fuel composition, Tulsa natural gas (TNG) •
Inside furnace dimensions, H 14 m •  × L 2.7
m × W 1.8 m (45′ × 9′ × 6′)
Convection tubes, carbon steel •
Figure 7.67 is a schematic showing the test arrange-
ment. The setup allowed for a sample probe to be
inserted at three locations just before the convection sec-
tion or at three locations in the middle of the convection
section as shown in the fgure. The probe depth could
also be adjusted to various depths within the convection
section. The fue gas sample was pulled through a heat
traced Tefon tube to a remotely located chemilumines-
cent NO
x
analyzer. Flue gas samples were also collected
from a sample port located approximately 3 m (10 ft)
below the top of the furnace and stack exit plane. Those
samples were also pulled through a heat traced Tefon
tube to the same NO
x
analyzer.
Tests were performed using state-of-the-art continu-
ous emissions monitors (CEMs) to measure emissions
species concentrations of NO
x
, CO, and O
2
. Stack sam-
ples were continuously extracted at high fow rates to
minimize response time. A continuous sample was fed
to a condenser/conditioner, which cools the sample to
1°C dew point. Once the moisture is removed, the dry
sample fows to the CEM for species analysis.
The reduction converter effciency was determined
using NO
2
span gas with 16.5 ppm. With this span gas,
the analyzer read a value of 14.6 ppm, for a conversion
effciency of 88%. This compared very favorably with the
87% conversion effciency reported in Bussman, Baukal,
and Waibel [52] for MnO
2
.
Field tests were performed in a full-scale ethylene
cracking furnace comprised of multiple foor and wall
burners operating at a total heat release of approxi-
mately 68.9 MW (235 × 10
6
Btu/hr). The setup allowed
for a sample probe to be inserted at various locations
through site ports within the furnace radiant section
and crossover section located just before the convection
section. The probe length was approximately 3 m (10 ft)
long and could be adjusted to various depths within the
furnace. The fue gas sample was pulled through a heat
traced Tefon tube to a remotely located CEM.
Tests were performed using state-of-the-art CEMS.
Stack samples were continuously extracted at high fow
© 2011 by Taylor and Francis Group, LLC
178 Industrial Combustion Testing
rates to measure emissions species concentrations of NO
x
,
CO, and O
2
. A continuous sample was fed to a condenser/
conditioner, which cools the sample to 1°C dew point.
Once the moisture was removed, the dry sample fows
to the CEM for species analysis. The reduction converter
effciency of 96% was determined using NO
2
span gas.
7.8.3.3 Results
Lab Tests. Figures 7.68 and 7.69 are plots of the NO at
locations before and in the middle of the convection
section at various depths. This data was collected at a
total furnace heat release of 4.3 MW (14.7 × 10
6
Btu/hr)
burning 100% Tulsa natural gas ( > 90% CH
4
) at 3% O
2
in
the dry exhaust products. The data clearly show that the
NO
x
concentration varies by a maximum of 15% over a
distance of 46 cm (18 in.) through the convection section,
demonstrating that the NO
x
was highly stratifed within
the convection section.
The stratifcation of NO
x
might be due to recirculating
fow patterns inside the furnace. It is speculated that the
fow pattern of furnace fue gases might look similar to
that as illustrated in Figure 7.70. As the hot combustion
products turn 90º into the convection section, they cre-
ate a recirculation zone in the upper corner of the fur-
nace. Also in Figure 7.70 is a streak photograph showing
small particles fowing through two 90º elbows [53].
Notice in this photograph that there is a small recircula-
tion zone in the upper corner of the frst 90º elbow. This
recirculation zone, containing relatively cool gas, might
Top view of convection section
showing sample probe locations
measurement depths at
d = 6, 12, 18 and 24 inches
17_5 in
29_25 in
40 in
40 in
29_25 in
29 in
Stack
Sample
probe
5 ft
27 in
d
3
2
1
6
5
4
6 ft L × 9 ft W
45 ft H
Radiant wall burners
Floor burners
Convection section with
carbon steel tubes
15 ft
Figure 7.67
Schematic showing test furnace and sample probe locations. (From Bussman, W., Baukal, C., Waibel, R. “Ghost NO
x
.” Presented at the Air &
Waste Management Association’s 98th Annual Conference & Exhibition, Control #28, Minneapolis, MN, 2005.)
© 2011 by Taylor and Francis Group, LLC
Pollution Emissions 179
extend into the convection section of the furnace and
create the NO
x
stratifcation observed.
Also notice in Figure 7.70 that the NO appears to be
slightly lower at a location before the convection section
than at a location in the middle of the convection sec-
tion for all depths tested. This suggests that the carbon
steel convection tubes might be acting as a catalyst and
converting the NO
2
to NO, one of the possible culprits
for the ghost NO
x
anomaly.
Figures 7.71 and 7.72 are plots of NO
x
(NO + NO
2
) at
locations before and in the middle of the convection sec-
tion at various depths. The data clearly show that NO
x

varied by a maximum of 11% over a distance of 46 cm
(18 in.). Notice that the NO
x
appeared to be lower at a
location before the convection section than at a location
in the middle of the convection section, for all depths
tested. It is unclear why this occurs because the total
NO
x
(NO + NO
2
) should be conserved between these
two locations. Perhaps the recirculation pattern in the
upper corner of the frebox creates this anomaly. At a
probe depth of 61 cm (24 in.), the average NO
x
in the stack
appears to be in the center of the data scatter for NO
x
col-
lected at various locations in the convection section.
46
48
50
52
54
56
58
30
40
50
60
70
80
90
100
6
8
10
12
14
16
18
20
22
W
e
s
t

p
o
s
i
t
i
o
n

(
i
n
.
)
D
epth
(in
.)
A
v
e
.

N
O

(
p
p
m
d
v
)
46
48
50
52
54
56
58
In convection
section
Before convection
section
Figure 7.68
(See color insert following page 424.) NO emissions at locations
before and in the middle of the convection section at various depths.
(From Bussman, W., Baukal, C., Waibel, R. “Ghost NO
x
.” Presented at
the Air & Waste Management Association’s 98th Annual Conference &
Exhibition, Control #28, Minneapolis, MN, 2005.)
50
51
52
53
54
55
56
57
58
0 5 10 15 20 25 30
Probe depth (inches)
N
O

+

N
O
2

e
m
i
s
s
i
o
n
s

(
p
p
m
v
d
)
Middle of convection section
After convection section
Average in stack
Figure 7.69
Top surface is in middle of convection section, bottom surface before
convection section. (From Bussman, W., Baukal, C., Waibel, R. “Ghost
NO
x
.” Presented at the Air & Waste Management Association’s 98th
Annual Conference & Exhibition, Control #28, Minneapolis, MN,
2005.)
Hot
combustion
products
Re-circulation
zone
Hot combustion products results in high
NO
x
concentration levels in this region
Re-circulation zone
Figure 7.70
Illustration showing theorized fow pattern within test furnace.
(From Bussman, W., Baukal, C., Waibel, R. “Ghost NO
x
.” Presented at
the Air & Waste Management Association’s 98th Annual Conference
& Exhibition, Control #28, Minneapolis, MN, 2005.) Also shown is a
streak photograph of particles fowing through two 90º turns.
© 2011 by Taylor and Francis Group, LLC
180 Industrial Combustion Testing
The convection tube skin temperature in the test fur-
nace might not have been high enough to provide sub-
stantial conversion of NO
2
to NO. In the frst few rows
of the convection section of a typical ethylene cracking
furnace, the tube skin temperature can reach values
above 1300 K (1800°F).
Field Tests. Figure 7.73 is a plot of the NO
x
(NO + NO
2
),
corrected to 3% O
2
, at various locations in the upper 3 m
(10 ft) of the radiant section. This data was collected at a
total furnace heat release of 68.9 MW (235 × 10
6
Btu/hr)
burning a typical refnery fuel at approximately 5.7% O
2

in the stack. The data show that the overall average NO
x

value is 26% lower in the upper zone of the radiant sec-
tion than at the stack exit. The data also show that varia-
tions in probe readings within the radiant section vary
signifcantly (approximately 40%), but are fairly constant
at the stack exit.
Figure 7.74 is a plot of the NO and NO
2
at various
locations in the upper 3 m (10 ft) of the radiant section;
the data corresponds to the plotted values in Figure 7.73.
The data show that the overall average NO
2
value is 152%
lower in the stack than in the upper zone of the radiant
section. However, the NO is 147% higher in the stack
than in the upper zone of the radiant section. This trend
appears to suggest NO
2
is converted to NO as the fue gas
passes through the convection section.
46
48
50
52
54
56
58
0 5 10 15 20 25 30
Probe depth (inches)
N
O

e
m
i
s
s
i
o
n
s

(
p
p
m
v
d
)
Before convection section
Middle of convection section
Average in stack
Figure 7.71
NO
x
(NO + NO
2
) at locations before and in the middle of the con-
vection section at various depths. (From Bussman, W., Baukal, C.,
Waibel, R. “Ghost NO
x
.” Presented at the Air & Waste Management
Association’s 98th Annual Conference & Exhibition, Control #28,
Minneapolis, MN, 2005.)
51
52
53
54
55
56
57
58
30
40
50
60
70
80
90
100
6
8
10
12
14
16
18
20
22
A
v
e
.

N
O
x

(
N
O

+

N
O
2
)

(
p
p
m
d
v
)
51
52
53
54
55
56
57
58
W
e
s
t

p
o
s
i
t
i
o
n

(
i
n
.
)
D
epth
(in
.)
In convection
section
Before convection
section
Figure 7.72
(See color insert following page 424.) Top surface is in middle of con-
vection section, bottom surface before convection section. (Bussman,
W., Baukal, C., Waibel, R. “Ghost NO
x
.” Presented at the Air & Waste
Management Association’s 98th Annual Conference & Exhibition,
Control #28, Minneapolis, MN, June, 2005.)
35
30
N
O
2

+

N
O

c
o
r
r
e
c
t
e
d

t
o

3
%

O
2
25
20
15
10
5
0
1 2 3
Sample number
Stack (NO
2
+ NO) – Average = 29.4 ppm
Probe (NO
2
+ NO) – Average = 23.3 ppm
4 5
Figure 7.73
NO
x
(NO + NO
2
) at various locations in the upper 3 m (10 ft) of the
radiant section of the feld test furnace. (From Bussman, W., Baukal,
C., Waibel, R. “Ghost NO
x
.” Presented at the Air & Waste Management
Association’s 98th Annual Conference & Exhibition, Control #28,
Minneapolis, MN, 2005.)
30
E
m
i
s
s
i
o
n
s

c
o
r
r
e
c
t
e
d

t
o

3
%

O
2
25
20
15
10
5
0
1 2 3
Sample number
Stack NO – Average = 27.4 ppm
Probe NO – Average = 18.6 ppm
Stack NO
2
– Average = 1.9 ppm
Probe NO
2
– Average = 4.8 ppm
4 5
Figure 7.74
NO and NO
2
at various locations in the upper 3 m (10 ft) of the radi-
ant section of the feld test furnace. (From Bussman, W., Baukal, C.,
Waibel, R. “Ghost NO
x
.” Presented at the Air & Waste Management
Association’s 98th Annual Conference & Exhibition, Control #28,
Minneapolis, MN, 2005.)
© 2011 by Taylor and Francis Group, LLC
Pollution Emissions 181
7.8.3.4 Conclusions
Due to the NO
x
stratifcation observed in the convection
section of the lab furnace, one cannot conclude with cer-
tainty that the ghost NO
x
anomaly was observed. The feld
tests, however, appear to indicate the ghost NO
x
anomaly.
This is at least in part likely due to catalytic decomposi-
tion of NO
2
through the convection section. Other causes
are also possible. It is recommended that additional tests
be performed. One test might include testing in another
furnace confguration. Another could be with different
burner types. Measurements of other species (e.g., N
2
O)
may also produce useful information that can be used to
isolate the source of the additional NO
x
.
References
1. Yeh, J. T. “Modeling Atmospheric Dispersion of
Pollutants.” In). Air Pollution Control and Design for
Industry, edited by P. N. Cheremisinoff. New York:
Marcel Dekker, 1993.
2. U.S. EPA. 40 CFR 60: Standards of Performance for New
Stationary Sources. Washington, DC: U.S. Environmental
Protection Agency, 2001.
3. U.S EPA. Title 40-Protection of Environment, Part 60-Stan-
dards of Performance for New Stationary Sources. http://
ecfr.gpoaccess.gov/cgi/t/text/text-idx?c=ecfr&tpl=/
ecfrbrowse/Title40/cfr60_main_02.tpl/ (accessed 12
February 2010).
4. Wheeler, A. J., and Ganji, A. R. Introduction to Engineering
Experimentation. 3rd ed. New York: Prentice Hall, 2010.
5. Cobb, G. W. Introduction to Design and Analysis of
Experiments. New York: Springer-Verlag, 1998.
6. Holman, J. P. Experimental Methods for Engineers. 7th ed.
New York: McGraw-Hill, 2000.
7. Montgomery, D. C. Design and Analysis of Experiments.
5th ed. New York: John Wiley & Sons, 2000.
8. Diamond, W. J. Practical Experiment Designs: For Engineers
and Scientists. 3rd ed. New York: John Wiley & Sons,
2001.
9. Finlayson-Pitts, B. J., and Pitts, J. N. Atmospheric Chemistry:
Fundamentals and Experimental Techniques. New York:
John Wiley & Sons, 1986.
10. Clement, R., and Kagel, R., eds. Emissions from Combustion
Processes: Origin, Measurement, Control. Boca Raton, FL:
CRC Press, 1990.
11. Cornforth, J. R. ed. “Instrumentation and Measurement.”
In Combustion Engineering and Gas Utilization. 3rd ed.
London: E&FN Spon, 1992.
12. Baumbach, G. Air Quality Control. Berlin: Springer,
1996.
13. Liu, D. H. F., and Lipták, B. G., eds. Environmental
Engineers’ Handbook. Boca Raton, FL: Lewis, 1997.
14. Schnelle, K. B., and Brown, C. A. Air Pollution Control
Technology Handbook. Boca Raton, FL: CRC Press, 2002.
15. Friedrich, R., and Reis, S., eds. Emissions of Air Pollutants.
New York: Springer, 2004.
16. Crawford, M. Air Pollution Control Theory. New York:
McGraw-Hill, 1976.
17. Baukal, C. E. Industrial Combustion Pollution and Control.
New York: Taylor & Francis, 2004.
18. United Nations, ed. Concepts and Methods of Environment
Statistics: Statistics of the Natural Environment. New York:
United Nations Publications, 1992.
19. Newman, O., and Foster, A., eds. Environmental Statistics
Handbook: Europe. Farmington Hills, MI: Gale Group, 1993.
20. Patil, G. P., and Rao, C. R., eds. Multivariate Environmental
Statistics. New York: Elsevier Science, 1994.
21. Rao, C. R., and Patil, G. P., eds. Environmental Statistics.
New York: Elsevier Science, 1994.
22. Ott, W. R. Environmental Statistics and Data Analysis. Boca
Raton, FL: CRC Press, 1995.
23. Hoshmand, A. R. Statistical Methods for Environmental and
Agricultural Sciences. 2nd ed. Boca Raton, FL: CRC Press,
1997.
24. Kottegoda, N. T., and Russo, R. Statistics, Probability and
Reliability Methods for Civil and Environmental Engineers.
New York: McGraw-Hill, 1997.
25. Piegorsch, W. W., Cox, L. H., and Nychka, D., eds. Case
Studies in Environmental Statistics. New York: Springer-
Verlag, 1998.
26. Novartis Foundation. Environmental Statistics: Analysing
(sic) Data for Environmental Policy. New York: John Wiley
& Sons, 1999.
27. Manly, B. F. Statistics for Environmental Science and
Management. Boca Raton, FL: CRC Press, 2000.
28. Berthouex, P. M., and Brown, L. C. Statistics for
Environmental Engineers. 2nd ed. Boca Raton, FL: Lewis
Publishers, 2002.
29. Townsend, J. Practical Statistics for Environmental and
Biological Scientists. New York: John Wiley & Sons, 2002.
30. Lodge, J. P., ed. Methods of Air Sampling and Analysis. 3rd
ed. Chelsea, MI: Lewis Publishers, 1977.
31. Mussatti, D. C., ed. EPA Air Pollution Control Cost Manual.
6th ed. Report EPA/452/B-02-001. Washington, DC: U.S.
Environmental Protection Agency, January 2002.
32. Mussatti, D. C., Groeber, M., Maloney, D., Koucky, W.,
and Hemmer, P. M. “Section 2: Generic Equipment and
Devices, Chapter 4: Monitors.” In Air Pollution Control
Cost Manual, edited by D. C. Mussatti. 6th ed. .Report
EPA/452/B-02-001. Washington, DC: U.S. Environmental
Protection Agency, January 2002.
33. American National Standards Institute. ISO 10780: 1994,
Stationary source emissions: Measurement of velocity and
volume fowrate of gas streams in ducts. Washington, DC:
American National Standards Institute, 2007.
34. Mullinger, P., and Jenkins, B. Industrial and Process
Furnaces. Oxford, UK: Butterworth-Heinemann, 2008.
35. Offce of the Federal Register. 60.2 Defnitions. U.S. Code of
Federal Regulations Title 40 Part 60. Washington, DC: U.S.
Government Printing Offce, 2001.
36. Bussman, W., and Baukal, C. “Ambient Condition Effects
on Process Heater Emissions.” Proceedings of the Int’l
Mechanical Engineering Congress & Exhibition, Paper
IMECE2008-68284, Boston, MA, November 2008.
© 2011 by Taylor and Francis Group, LLC
182 Industrial Combustion Testing
37. American Petroleum Institute. Burners for Fired Heaters in
General Refnery Services, API Recommended Practice 535. 2nd
ed. Washington, DC: American Petroleum Institute, 2006.
38. Farnsworth, R., and Rowling, H. “The Reduction of
Pollution from Industrial and Domestic Equipment: The
Oil Industry’s Contribution.” In Pollution Prevention,
edited by Peter Hepple. London, UK: The Institute of
Petroleum, The Elsevier Publishing Co., 1968.
39. Chellappan, C., and Waibel, R. “Reducing NO
x

Emissions from Ethylene Furnaces: Practical Application
of Lean Premix Combustion.” American Flame Research
Committee Meeting, October 2004.
40. Becker, H. A., and Brown, A. P. G. “Response of Pitot
Probes in Turbulent Streams. “Journal of Fluid Mechanics
62 (1974): 85–114.
41. Cho, H. S., and Becker, H. A. “Response of Static Pressure
Probes in Turbulent Streams.” Experiments in Fluids 3
(1985): 93–102.
42. U.S. EPA. Compilation of Air Pollutant Emission Factors, Vol.
I: Stationary Point and Area Sources. 5th ed. Washington, DC:
U.S. Environmental Protection Agency report AP-42, 1995.
43. U.S. EPA. Appendix B: Performance Specifcations. Code
of Federal Regulations 40 Part 60. Washington, DC: U.S.
Environmental Protection Agency, 2001.
44. U.S. EPA. Appendix F: Quality Assurance Procedures. Code
of Federal Regulations 40 Part 60. Washington, DC: U.S.
Environmental Protection Agency, 2001.
45. Parker, J. M. “Continuous Emission Monitoring (CEM).”
In Handbook of Air Pollution Control and Engineering,
edited by J. C. Mycock, J. D. McKenna, L. Theodore. Boca
Raton, FL: Lewis Publishers, 1995.
46. Slavejkov, A. G., and Baukal, C. E. “Flue Gas Sampling
Challenges in Oxygen-Fuel Combustion Processes.” In
Transport Phenomena in Combustion, edited by S. H. Chan,
vol. 2, 1230–37. Boca Raton, FL: Taylor & Francis, 1996.
47. Baukal, C. E., and Dalton, A. I. “Nitric Oxide
Measurements in Oxygen Enriched Air-Natural Gas
Combustion Systems.” Proceedings of the Fossil Fuel
Combustion Symposium, edited by S. N. Singh, New York:
ASME, 1990.
48. ANSI/ASME. Flue and Exhaust Gas Analysis: Instruments
and Apparatus. New York: American Society of Mechanical
Engineers standard PTC 19.10, 1981.
49. Hayes, R., Baukal, C. E., Singh, P., and Wright, D. “Fuel
Composition Effects on NO
x
.” Proceedings of the
Air & Waste Management Association’s 94th Annual
Conference and Exhibition, Orlando, FL, June 24–28,
2001.
50. Baukal, C. E., ed. The John Zink Combustion Handbook.
Boca Raton, FL: CRC Press, 2001.
51. Baukal, C. E., ed. Industrial Burners Handbook. Boca Raton,
FL: CRC Press, 2004.
52. Bussman, W., Baukal, C., Waibel, R. “Ghost NO
x
.”
Presented at the Air & Waste Management Association’s
98th Annual Conference & Exhibition, Control #28,
Minneapolis, MN, June 21–24, 2005.
53. Idelchik, I. E. Handbook of Hydraulic Resistance. New York:
Hemisphere Publishing Corporation, 1986.
© 2011 by Taylor and Francis Group, LLC
183
8
Combustion Noise
Mahmoud M. Fleifl, Carl-Christian Hantschk, and Edwin Schorer
CONTENTS
8.1 Fundamentals of Sound ............................................................................................................................................... 184
8.1.1 Introduction ....................................................................................................................................................... 184
8.1.2 Basics of Sound .................................................................................................................................................. 184
8.1.2.1 Sound Pressure Level and Frequency ............................................................................................. 185
8.1.2.2 Decibel ................................................................................................................................................. 186
8.1.2.3 Sound Power Level ............................................................................................................................ 186
8.1.2.4 Threshold of Hearing ........................................................................................................................ 187
8.1.2.5 Threshold of Pain ............................................................................................................................... 188
8.1.2.6 Correction Scales ................................................................................................................................ 188
8.1.3 Measurements ................................................................................................................................................... 189
8.1.3.1 Overall Sound Level and How to Add dB Values ......................................................................... 190
8.1.3.2 Atmospheric Attenuation ................................................................................................................. 193
8.2 Industrial Noise Pollution ........................................................................................................................................... 193
8.2.1 OSHA Requirements ........................................................................................................................................ 194
8.2.2 International Requirements ............................................................................................................................ 194
8.2.3 Noise Sources and Environment Interaction ................................................................................................ 194
8.3 Mechanisms of Industrial Combustion Equipment Noise ..................................................................................... 195
8.3.1 Combustion Roar and Combustion Instability Noise ................................................................................. 195
8.3.1.1 Flare Combustion Roar ..................................................................................................................... 195
8.3.1.2 Flare Combustion Instability Noise ................................................................................................ 198
8.3.1.3 Burner Combustion Noise ................................................................................................................ 199
8.3.1.4 Burner Combustion Instability Noise ............................................................................................. 199
8.3.2 Fan Noise............................................................................................................................................................ 200
8.3.3 Gas Jet Noise ...................................................................................................................................................... 200
8.3.3.1 Gas Jet Mixing Noise ......................................................................................................................... 200
8.3.3.2 Shock-Associated Noise .................................................................................................................... 201
8.3.4 Valve and Piping Noise .................................................................................................................................... 202
8.4 Noise Abatement Techniques ..................................................................................................................................... 202
8.4.1 Flare Noise Abatement Techniques ............................................................................................................... 202
8.4.2 Burner Noise Abatement Techniques ............................................................................................................ 204
8.4.3 Valve and Piping Noise Abatement Techniques .......................................................................................... 205
8.4.4 Fan Noise Abatement Techniques .................................................................................................................. 206
8.5 Analysis of Combustion Equipment Noise ............................................................................................................... 206
8.5.1 Multiple Burner Interaction ............................................................................................................................. 206
8.5.2 High Pressure Flare .......................................................................................................................................... 207
8.5.3 Atmospheric Attenuation Example ................................................................................................................ 207
Glossary ................................................................................................................................................................................... 208
References ................................................................................................................................................................................ 209
Bibliography ........................................................................................................................................................................... 210
© 2011 by Taylor and Francis Group, LLC
184 Industrial Combustion Testing
8.1 Fundamentals of Sound
8.1.1 introduction
The sensation of sound is a thing sui generis, not
comparable with any of our other sensations. No
one can express the relation between a sound and
a color or smell. Directly or indirectly, all ques-
tions connected with this subject must come for
decision to the ear, as the organ of hearing; and
from it there can be no appeal. But we are not
therefore to infer that all acoustical investigations
are conducted with the unassisted ear. When
once we have discovered the physical phenom-
ena which constitute the foundation of sound,
our explorations are in great measure transferred
to another feld lying within the dominion of the
principles of Mechanics. Important laws are in
this way arrived at, to which the sensations of the
ear cannot but conform.
—The Theory of Sound by Lord Rayleigh III
Noise is commonly referred to as unwanted sound.*
Noise is a common by-product of our mechanized
civilization and is an insidious danger in industrial
environments. Noise pollution is usually a local problem
and so is not viewed on the same scale of importance as
the more notorious industrial emissions like NOx, CO,
and particulates. Nonetheless, it is an environmental
pollutant of signifcant impact.
Serious concern is merited when a pollutant can result
in either environmental damage or human discomfort.
Considering the impact on people, noise is most often
a source of annoyance, but it can also have much more
detrimental effects, such as causing actual physical
injury. Noise-related injuries range from short-term dis-
comfort to permanent hearing loss. According to recent
statistics, more than 20 million Americans are exposed
to hazardous sound levels on a regular basis. There are
approximately 28 million Americans who have some
degree of hearing loss: about one-third of these—more
than nine million—have been affected, at least in part,
by exposure to excessive noise.
The sense of hearing is a fragile and vital function of
the human body. It is similar to vision, more so than the
other senses, because permanent and complete dam-
age can be sustained quite commonly in an industrial
environment. So it follows that noise pollution has been
recognized as a safety concern for a long time, and has
been appropriately regulated.
Although personnel safety may be the most impor-
tant, noise pollution has several other side effects that
* A glossary of terms concerning noise is given at the end of this
chapter before the references.
are signifcant. Combustion equipment designers are
often asked why they would want to constrain the com-
bustion process in order to reduce noise. Frequently
these questions come from persons working in plants
situated in remote areas. However, given the age and
economic drivers of the petroleum refning and chemi-
cal industries, currently it is common to fnd plants
located in densely populated areas. With industry that
is situated close to residential areas or busy commer-
cial facilities, high levels of noise become objectionable
to people in the neighborhood. This eventually leads
to government regulations to control noise emissions.
On the other hand, within the industrial site itself, the
immediate issue with noise is one of employee safety.
Furthermore, it is not surprising to fnd that employee
moral and performance improves when noise is
reduced.
Equipment too is affected by noise. In most cases
these effects are addressed in the realm of vibration
control and are beyond the scope of this chapter. Suffce
it to say, that noise is indeed a form of vibration and
eventually contributes to fatigue, which reduces equip-
ment life. Usually the effects of fatigue are long term
and are possibly accepted as normal wear and tear if
the equipment life cycle spans a reasonable duration.
In extreme cases, the effects of vibration may be more
rapidly manifested, such as in the case of cracking and
falling of the hard refractory linings in furnaces. This
chapter has been written as a practical guide as well as
a reference on noise for engineers involved in design,
operation, or maintenance of combustion equipment,
be it burners, furnaces, fares, or thermal oxidizers. In
addition, since this chapter provides a comprehensive
coverage of the fundamentals of sound, the creative
engineer will be able to extend his or her knowledge to
analysis of other noise producing industrial equipment
as well.
8.1.2 basics of Sound
If a tree falls in the forest and nobody is around to
hear it, does it still make a sound?
Webster’s dictionary defnes sound as “That which is
heard.” Obviously, an engineer will fnd this defni-
tion woefully inadequate for his or her purposes. So
we resort to the defnition provided by the Handbook of
Noise Measurement: “Sound is the vibration of particles
in a gas liquid or solid” [1].
Sound is propagated through any medium in waves
that take the form of pressure peaks (compressions) and
troughs (rarefactions) as illustrated in Figure 8.1. The
pressure wave travels through the medium at the speed
of sound. The auditory system in humans and most ani-
mals senses the impingement of these pressure waves
© 2011 by Taylor and Francis Group, LLC
Combustion Noise 185
on a tissue membrane and converts them to electrical
impulses that are then sent to the brain and interpreted
there.
Figure 8.2 is a cross-section of the human ear. Sound
is collected and funneled into the ear canal by the outer
ear. At the end of the ear canal, the sound impinges
upon a membrane called the ear drum. The bones of
the middle ear convey the ear drum’s vibration to the
inner ear. The inner ear consists of a fuid flled Basilar
membrane that has tiny hair cells on the inside. The hair
cells sense the vibration conveyed to the Basilar mem-
brane and convert this into electrical signals that are
then conveyed to the brain.
In reality, most naturally occurring sounds are
composites of different pressure levels at various fre-
quencies. On the other hand, a pure tone is a sound
at only one frequency. Any pure tone can be uniquely
identifed by two of its properties, namely, pressure
level and frequency. A tuning fork is a good example
of a pure tone generator. Such naturally occurring pure
tone generators are rare. Even musical instruments
create notes that have signifcant pressure levels at two
or more multiples, or harmonics, of the fundamental
frequency of the note.
8.1.2.1 Sound Pressure Level and Frequency
Pressure level defnes the loudness of the sound, while
frequency defnes the pitch or tone of the sound. Pressure
level is the amplitude of the compression, or rarefaction,
of the pressure wave. The common unit of pressure level
is Decibel, abbreviated to “dB.” Frequency is the num-
ber of pressure waves that pass by an arbitrary point of
reference in a given unit of time. As such, the measure
of sound frequency can be cycles per second (CPS), and
as with electricity, the commonly used unit is the Hertz.
One Hertz is one cycle per second.
Another important quantity in the description of
sound waves is the speed of sound, c, which describes the
velocity at which a sound wave propagates through the
ambient air or other fuid (e.g., water). In other words,
the speed of sound describes the distance a sound wave
propagates per unit of time. As a third quantity that
characterizes a sound wave, the wavelength λ must be
mentioned. It describes the distance between one wave
crest and the next (i.e., between one point of maximum
compression in the sound wave to the next).
The frequency f, the speed of sound c, and the wave-
length λ are related as follows:
c = λ × f. (8.1)
The speed of sound depends mainly on the fuid in
which the sound wave propagates and the temperature.
Table 8.1 gives some examples for the speed of sound in
different media and at different temperatures.
The typical range of human hearing extends from
20 Hz to 20 kHz. Young children can hear frequencies
slightly higher than 20 kHz but this ability dimin-
ishes with age. This trend of reduced high frequency
sensitivity continues with advancing age. Loss of
hearing in humans in the later stages of life typically
manifests itself as diminished sensitivity to frequen-
cies from 10 to 20 kHz. Mechanically, this is due to
the deterioration of the fne hair cells in the Basilar
membrane.
It is important to note that the ear is not equally sen-
sitive over the entire range from 20 Hz to 20 kHz. This
is vital to understanding how noise affects us and how
noise control is implemented. The human ear is much
less sensitive to sound at the extremes of low and high
frequencies and we will discuss this in more detail later
in the chapter.
Sound wave - a pressure wave
moving at the speed of sound
+
Pressure

Figure 8.1
Pressure peaks and troughs.
Eardrum
Nerve that sends
electrical signals
to the brain
Cochlea
Ear canal
Fluid filled
membrane
Hair cells
Figure 8.2
Cross-section of human ear.
© 2011 by Taylor and Francis Group, LLC
186 Industrial Combustion Testing
The wide range of frequencies in our hearing range
may be conveniently handled by breaking it up into
octave bands. Each octave band represents a doubling
in frequency. Table 8.2 shows the 10 octave bands that
cover the hearing range and the center frequencies
that can be used to represent the octave band. Each
octave band extends over seven fundamental musical
notes.
8.1.2.2 Decibel
The unit of sound level, the decibel, is diffcult to
visualize and merits some explanation. While it is
possible to quantify sound in units of either power
or pressure, neither unit is convenient to use. This is
because, in practice, we have to deal everyday with
sounds that extend over a very large range of power
or pressure values. For example, the sound power of
a whisper is 10
−9
watts, while the sound power of a
jet plane is 10
3
watts. The range of these two sound
sources spans 10
12
ratio. The decibel, a dimensionless
unit, was invented in order to represent these large
ranges conveniently.
In the 1960s Bell Laboratories coined the term decibel.
The “deci” stands for the base 10 Log scale on which the
decibel is based. See Figure 8.3 for how decibel relates to
watts. The “bel,” of course, stood for Bell Labs.
In Figure 8.3, the Y-axis is power in watts given in
scientifc notation. The Y-axis follows a base-10 scale.
The X-axis gives dB values and the line provides the
relationship: 120 dB is equal to 1 watt. As an illustration
of the log
10
relationship, note that 110 dB is equal to one-
tenth of a watt (0.1 watt) and 100 dB is equal to 100th of
a watt.
8.1.2.3 Sound Power Level
There is a subtle but important difference between the
terms Sound Power Level (PWL), and Sound Pressure
Level (SPL). Sound power level is used to indicate the
total energy emitting ability of a sound source. In
other words, sound power is an attribute of the source
itself. Sound pressure level, on the other hand, is used
to indicate the intensity of sound received at any point
of interest, from one or more sources. The illustration
in Figure 8.4 shows the formula to calculate the SPL to
be expected at a distance r from a spherically radiating
source of power level L
w
:
L
p
 = L
w
− 10 log
10
(4πr
2
) + 10.5, (8.2)
where
L
p
  =  sound pressure level in dB
L
w
 =  Sound power level in dB
r  =  Distance from source in ft.
Table 8.1
Speed of Sound in Different Media and at
Different Temperatures
Fluid
Temperature
[°C]
Speed of Sound
c [m/s]
Air, dry −40 307
20 320
0 332
20 344
40 356
Hydrogen 20 1316
Methane 20 448
Ethylene 20 331
Water 0 1403
40 1529
80 1555
100 1543
Steel 20 5180
Soft rubber 20 54
Table 8.2
The Ten Octave Bands
Full Octave Band Standards
Octave Band, Hz Center Frequency, Hz
22–44 31.5
44–88 63
88–177 125
177–355 250
355–710 500
710–1420 1000
1420–2840 2000
2840–5680 4000
5680–11,360 8000
11,360–22,720 16,000
P
o
w
e
r
,

W
dB
1×10
1
1×10
0
1×10
–1
1×10
–2
1×10
–3
1×10
–4
1×10
–5
1×10
–6
1×10
–7
1×10
–8
1×10
–9
1×10
–10
1×10
–11
1×10
–12
0 10 20 30 40 50 60 70 80 90 100 110 120 130
Figure 8.3
Relationship of decibels to watts.
© 2011 by Taylor and Francis Group, LLC
Combustion Noise 187
Equation 8.2 is only valid exactly if the following
conditions are fulflled:
1. Noise radiation from the source is uniform and
equal in every direction.
2. The source is small compared to the distance r.
3. There is no relevant noise contribution from
other sources at the point of interest.
4. There is no relevant infuence on the noise at the
point of interest from sound refecting surfaces
nearby.
If the above conditions are suffciently fulflled, then
Equation 8.2 can not only be used to calculate the SPL
expected at a certain distance r from the source if
the sound power level L
w
is known, but also to back-
calculate the PWL of a source from a measurement of
the sound pressure level at a known distance r from
the source. The latter procedure is a technique that is
standardized in ISO 3744 [2].
If the above conditions are not fulflled, then the SPL
predicted at a certain distance from a sound source can
differ signifcantly from the level actually measured
there, or the PWL of a source back-calculated from a
measurement of the SPL can differ signifcantly from
the real SPL of the source. In this context, it is important
to keep in mind that the PWL is an intrinsic property
of the source that is independent from the environment
while the SPL not only depends on the distance to the
source, but is also infuenced by other sound sources,
the environment (refecting surfaces. etc.), and the sound
propagation conditions (temperature, atmospheric pres-
sure, wind, barriers, etc.).
The following are useful equations that may be used
to calculate sound pressure and power levels, in dB,
from the equivalent pressure and power units.
L
p
(dB) = 20 log
10
P/(2 × 10
−5
),
L
w
(dB) = 10 log
10
W/(1 × 10
−12
),
where
L
p
  =  sound pressure level in dB,
L
w
 =  sound power level in dB,
P  =  pressure in N/m
2
,
W  =  sound power level in watts.
8.1.2.4 Threshold of Hearing
Figure 8.5 shows a curve, which is a map of the thresh-
old of hearing in humans. The Y-Axis is SPL and the
X-Axis is frequency. Any SPL at a frequency that falls
below the curve will be inaudible to humans. For
example a SPL of 30 dB at 63 Hz, being below the curve,
will be inaudible. Whereas, a SPL of 70 dB at the same
63 Hz will be audible. Humans are most sensitive to
sounds in the so called midfrequencies from 1 kHz
to about 5 kHz. This is generally the region in which
most of our everyday hearing activities take place.
The human voice, most musical instruments, and so
on produce sounds predominantly in the midfrequen-
cies. Additionally, at a constant level, sound with a low
or very high frequency will not have the same loud-
ness sensation as that in the medium frequency range.
For example, a 100 Hz tone at L
p
 = 50 dB gives the same
loudness as a 1000 Hz tone at L
p
 = 40 dB [3].
Noise source with a
given sound power level
(L
w
) radiating outward.
Sound pressure
wave at a
distance r feet
from the source.
r
Te sound
pressure level at
this point can be
calculated as
L
p
= L
w
+ 10 log
10
where r is in feet.
1
+10.5
4πr
2
Figure 8.4
Calculating sound pressure level at a distance r.
Human threshold of hearing
Frequency, Hz
1
6
130
110
90
70
50
30
10
–10
–30
3
1
.
5
6
3
1
2
5
2
5
0
5
0
0
1
0
0
0
2
0
0
0
4
0
0
0
8
0
0
0
1
6
0
0
0
2
0
0
0
0
S
o
u
n
d

p
r
e
s
s
u
r
e

l
e
v
e
l
,

d
B
Figure 8.5
Threshold of hearing in humans.
© 2011 by Taylor and Francis Group, LLC
188 Industrial Combustion Testing
8.1.2.5 Threshold of Pain
Figure 8.6 also shows the threshold of pain super-
imposed. Fortunately for us, the threshold of pain is
relatively fat. Generally a SPL of over 120 dB at any fre-
quency will cause pain. An important observation that
can be derived from the two curves is that if a sound is
audible at very low or very high frequencies, persons
subject to this sound are very close to experiencing
pain.
8.1.2.6 Correction Scales
Sound meters have the capability to measure with equal
sensitivity over the entire audible range. However, since
humans do not hear with equal sensitivity at all frequen-
cies, the sound meter measurement needs to be modi-
fed to quantify what really affects us. This can be done
using a correction curve. The most common correction
is the A-scale correction curve, which resembles an ide-
alized inverse of the threshold of hearing curve (refer
to Figure 8.7). An A-weighted sound level correlates
reasonably well with hearing-damage risk in industry
and with subjective annoyance for a wide category of
industrial and community noises. After applying the
A-scale correction, dBA is usually used as the unit of
SPL. Figure 8.8 shows a typical burner noise curve as
measured by the noise meter (fat scale) and the result
after applying A-scale correction.
The other, less used correction scales are named, as
may be expected B, C, and D. Referring to Figure 8.9 we
can see that the C-scale is essentially fat over the range
of interest and the B-scale lies somewhere between the
A and C scales. With the understanding we have today
of the infuence of low-frequency sounds, we fnd that
the B and C scales, although meant to be used with low-
frequency sounds, do not apply adequate correction in
Threshold of pain
Threshold of hearing
Frequency, Hz
130
110
90
70
50
30
10
–10
–30
1
6
3
1
.
5
6
3
1
2
5
2
5
0
5
0
0
1
0
0
0
2
0
0
0
4
0
0
0
8
0
0
0
1
6
,
0
0
0
2
0
,
0
0
0
S
o
u
n
d

p
r
e
s
s
u
r
e

l
e
v
e
l
,

d
B

Figure 8.6
Threshold of hearing and threshold of pain for humans.
Treshold of hearing
A-scale correction factor
Frequency, Hz
3
1
.
5
S
o
u
n
d

p
r
e
s
s
u
r
e

l
e
v
e
l
,

d
B
80
60
40
20
0
–20
–40
6
3
1
2
5
2
5
0
5
0
0
1
0
0
0
2
0
0
0
4
0
0
0
8
0
0
0
1
6
,
0
0
0
Figure 8.7
A-weighted scale.
A-weighted
Frequency, Hz
3
1
.
5
85
80
75
70
65
60
55
50
45
40
S
o
u
n
d

p
r
e
s
s
u
r
e

l
e
v
e
l
,

d
B
6
3
1
2
5
2
5
0
5
0
0
1
0
0
0
2
0
0
0
4
0
0
0
8
0
0
0
1
6
,
0
0
0
Figure 8.8
Typical burner noise curve.
+20
+10
0
–10
–20
–30
–40
–50
–60
–70
10 2 5 10
2
10
3
10
4
2 5
Frequency, Hz
R
e
l
a
t
i
v
e

r
e
s
p
o
n
s
e
,

d
B
2
A
B
C
D
5 2
Figure 8.9
Weighting scales.
© 2011 by Taylor and Francis Group, LLC
Combustion Noise 189
the lower frequencies. Lastly, the D-scale is different
from the others, in that it has a pronounced correction
in the range of 2 kHz to 5 kHz. The D-scale was devised
for the aircraft industry and is rarely used otherwise.
8.1.3 Measurements
A simple schematic of a noise meter is shown in
Figure 8.10. The microphone is a transducer that trans-
forms pressure variations in air to a corresponding
electrical signal. Since the electrical signal generated
by the microphone is relatively small in magnitude, a
preamplifer is needed to boost the signal before it can
be analyzed, measured, or displayed. Special weighting
networks are used to shape the signal spectrum and
apply the various correction scales discussed above.
The weighted signal then passes through a second out-
put amplifer into a meter. The meter and associated
electronic circuits detect the approximate rms-value of
the signal and display it in dB.
Noise meters range from the simplest—microphone
and needle gauge—to sophisticated digital signal pro-
cessing (DSP) equipped analyzers. The more sophis-
ticated analyzers are equipped with fast Fourier
transform (FFT) capabilities that aid accurate narrow
band analysis. In general spectrum analyzers, allow the
user to map the SPL at different frequencies, or in other
words, generate a curve of the sound over different fre-
quencies. However, there is a great deal of difference
between instruments than make one measurement per
octave band and those that slice the octave band up into
several intervals and make a measurement at each inter-
val. Typically, instruments are capable of:
1. Octave band measurements
2. One-third octave band measurements
3. Narrow band measurements
Table 8.3 shows the usual octave and one-third octave
bands. As the name suggests, a one-third octave band
instrument makes three measurements in each octave as
opposed to the single measurement of the octave band
instrument. A narrow band instrument, on the other
hand, uses DSP to implement analysis FFT, and in the
current state of the art, FFT analysis allows the analyzed
frequency range to be sliced up into a large number of
smaller intervals, limited in number only by the mea-
sured time interval length and the available computer
power.
Figure 8.11 is a comparison of the same sound spec-
trum as analyzed using three different band inter-
vals—octave band, one-third octave, and narrow
band. This comparison shows that the additional reso-
lution provided by narrower band methods can be of
vital importance. In this example the level at 1 kHz, as
recorded by the octave band instrument is 90 dB, on
the one-third octave instrument it is 85 dB, and on the
narrow band instrument it is 70 dB. The lower reso-
lution measurements produce higher measurements
due to the spillover infuence of the nearby peak at
1.8 kHz. In addition, in implementing noise control for
this source, it is very valuable to know that it is the
narrow peak at 1.8 kHz that is driving the maximum
noise. This knowledge helps to zero in on the source,
which, for example, may be a fan at 3600 rpm and with
30 blades. This results in a tonal frequency (blade pas-
sing frequency) of 3600/60 × 30 = 1800 Hz (see Section
8.3.2 for the calculation of the tonal frequency of fan
noise).
However, as with all things, there is a cost associated
with high performance. For most purposes, a one-third
octave analysis is usually quite adequate. The advan-
tage of making broad band analyses using octave or
one-third octave band flter sets are that less time is
A
B
C
D
Band
filters
Weighting networks
Amplifier
Microphone
Amplifier
Meter
Flat
Slow
Fast Rectifier
Output
A
B
C
D
Figure 8.10
Schematic of a noise meter.
© 2011 by Taylor and Francis Group, LLC
190 Industrial Combustion Testing
needed to obtain data and the instrumentation required
to measure the data is less expensive.
In making sound measurements, several factors
regarding the nature of the source should be consid-
ered. Whether the source is a true point source in space
radiating spherically, or whether it is a hemispherical
source close to one fat surface or a quarter sphere
between two fat surfaces, and so on, it will make a
difference in how the measurement needs to be per-
formed. However, a detailed discussion of measure-
ment issues is beyond the scope of this chapter and
the reader may use some of the more comprehensive
works in the list of references at the end of this chapter.
The American Petroleum Institute has issued a rec-
ommended practice for measuring noise from fred
process heaters [4].
8.1.3.1 Overall Sound Level and How to Add dB Values
As mentioned before, most sounds are composites of
several different levels at different frequencies. This is
especially true of industrial noise. A typical burner noise
curve is shown in Figure 8.12. As can be seen, there are
signifcant levels in two frequency zones, both of which
Table 8.3
Octave and One-Third Octave Bands
Octave One-Third Octave
Band
Lower
Band Limit Center
Upper
Band Limit
Lower
Band Limit Center
Upper
Band Limit
12 11 16 22 14.1 16 17.8
13 17.8 20 22.4
14 22.4 25 28.2
15 22 31.5 44 28.2 31.5 35.5
16 35.5 40 44.7
17 44.7 50 56.2
18 44 63 88 56.2 63 70.8
19 70.8 80 89.1
20 89.1 100 112
21 88 125 177 112 125 141
22 141 160 178
23 178 200 224
24 177 250 355 224 250 282
25 282 315 355
26 355 400 447
27 355 500 710 447 500 562
28 562 630 708
29 708 800 891
30 710 1,000 1,420 891 1,000 1,122
31 1,122 1,250 1,413
32 1,413 1,600 1,788
33 1,420 2,000 2,840 1,778 2,000 2,239
34 2,239 2,500 2,818
35 2,818 3,150 3,548
36 2,840 4,000 5,680 3,548 4,000 4,467
37 4,467 5,000 5,623
38 5,623 6,300 7,079
39 5,680 8,000 11,360 7,079 8,000 8,913
40 8,913 10,000 11,220
41 11,220 12,500 14,130
42 11,360 16,000 22,720 14,130 16,000 17,780
43 17,780 20,000 22,390
Note: The advantages of making broad analyses of sound using octave or one-third octave band
flter sets are that less time is needed to obtain data and the instrumentation required to
measure the date is less expensive. The main disadvantage is the loss of detailed informa-
tion about the sound which is available from narrow band (FFT) analyzers.
© 2011 by Taylor and Francis Group, LLC
Combustion Noise 191
will contribute to the apparent intensity experienced
by a person working in the vicinity of the burner. It is
diffcult to describe this sound without using either a
diagram like the one shown or a table listing various
SPLs occurring in the different octave bands. The over-
all sound level, a single number, has been devised to
represent such composite sound curves conveniently.
If a single number is to be used to represent the whole
curve, then it should adequately represent the peaks in
the curve, since the peaks have the most infuence on
the listener. Consequently, it is not practical to use the
average of the various levels in the octave bands, since
this number would be less than the levels at the peaks.
Therefore, one must not confuse the average with the
overall sound level.
The overall sound level is calculated by adding the
individual levels in the various octave bands. In col-
umns 1 and 2 of Table 8.4, the burner sound curve has
been split up into its component levels in each octave
band. In column 3, the A-weighted correction has
similarly been split up and listed. Column 4 gives the
A-corrected values for the sound curve by simply add-
ing column 3 to column 2. Now, the values in column 4
must be added to obtain the A-weighted overall sound
level.
Since the decibel is based on a Log
10
scale, simple
addition cannot be used. For example, if two values
of equal magnitude are added, say 100 dB and 100 dB,
the result is 103 dB. The formula to be used is as
follows:

L
i n
L
i
total =1 to
0.1
= 10 log( 10 ), ∑

where
L
total
  =  the total level
L
i
  =  each individual level
n  =  the number of levels to be added.
Octave-band spectrum
100
(a)
(b)
(c)
90
80
70
60
50
100
90
80
70
60
50
100
90
80
70
60
50
10 100 1K 10K
Third-octave band spectrum
Narrow-band spectrum
Figure 8.11
Comparison of same sound spectrum analyzed using three different
band intervals.
85
80
75
70
65
60
55
50
45
40
Frequency, Hz
S
o
u
n
d

p
r
e
s
s
u
r
e

l
e
v
e
l
,

d
B
3
1
.
5
6
3
1
2
5
2
5
0
5
0
0
1
0
0
0
2
0
0
0
4
0
0
0
8
0
0
0
1
6
0
0
0
Figure 8.12
Typical burner noise curve.
Table 8.4
A-Weighting of the Burner Sound
Curve from Figure 8.12
Frequency
Hz SPL dB
A-Scale
CF dB
SPL
dBA
31.5 72 −39 33
63 75 −26 49
125 79 −16 63
250 79 −9 70
500 72 −3 69
1000 69 0 69
2000 68 1 69
4000 78 1 79
8000 83 −1 82
16,000 80 −7 73
© 2011 by Taylor and Francis Group, LLC
192 Industrial Combustion Testing
And subtraction can be performed by using

L
L L
diff
0.1 0.1
= 10 log(10 10 ).
2 1

However, some simple rules of thumb may be used to
perform quick estimates. They are as follows:
1. When adding dB values of equal magnitude or
differ by 1, the sum is 3 dB added to the greater
number.
2. When the two values are different by 2 or
3 dB, then the sum is 2 dB added to the greater
number.
3. When adding two values that differ by 4–9 dB,
then the sum is 1 dB added to the greater
number.
4. For values that differ by 10 dB or more, the sum
is just the larger number.
5. Always start with the smallest number in the
list and add it to the next larger number.
To understand why these rules work refer to the chart
in Figure 8.3. From the chart it can be seen that one watt
is equal to 120 dB.
1 watts = 120 dB
1 watts = 120 dB
2 watts = 123 dB
On the chart, 2 watts registers 123 dB on the line.
Similarly, the reason numbers that are 10 dB or more in
difference are neglected is because:
1.0 watts = 120 dB
0.1 watts = 110 dB
1.1 watts = 120 dB
Since the 110 dB contributes only a 10th of a watt of
power, we neglect it in approximation. The example
becomes more vivid when we add two numbers that
differ by 20 dB or more.
1.00 watts = 120 dB
0.01 watts = 100 dB
1.01 watts = 120 dB
Rule number 5 is especially necessary when adding a
list that contains several numbers that are almost equal
in value and one or more that are 10 dB greater, such as
in a list that contains six values of 90 dB and one value
of 100 dB. If we begin to add from the 100 dB value we
will arrive at a wrong result. Reader beware, the rules
provided are approximations. For exact calculations, the
formulas should be used.
Table 8.5 shows the effect of applying the addi-
tion rules to the values generated by breaking up the
burner noise curve. At the end of the addition list, 1 dB
has been added to compensate for any errors due to
approximation.
As an alternative to using the above rules of thumb
for level addition, the nomogram in Figure 8.13 can be
used to determine the sum of the two levels L
1

and L
2
:
For a certain difference between the two levels L
1

and L
2

in the lower scale of the nomogram, the corresponding
Table 8.5
Addition Rules
Frequency
Hz
SPL
dB
A-Scale
CF dB
SPL
dBA
31.5 72 −39 33
49
63
71
73
75
76
80
84
82
63 75 −26 49
125 79 −16 63
250 79 −9 70
500 72 −3 69
1,000 69 0 69
2,000 68 1 69
4,000 78 1 79
8,000 83 −1 82
16,000 80 −7 73
20
0 0.5 1 1.5 2 2.5 3
10 9 8 7 6 5 4 3 2 1 0
L
1
–L
2
ΔL
Figure 8.13
Nomogram for level addition.
© 2011 by Taylor and Francis Group, LLC
Combustion Noise 193
level increase ΔL to be added to the higher one of the two
levels can be found in the upper part of the nomogram.
8.1.3.2 Atmospheric Attenuation
When a sound wave travels through still air it is absorbed
or attenuated by the atmosphere. Over a couple of hun-
dred feet, the atmosphere does not signifcantly attenu-
ate the sound, however, over a few thousand feet the
sound level can be substantially reduced. The amount
of sound that is attenuated in still air depends largely
on the atmospheric temperature and relative humidity.
Figure 8.14 are plots showing the atmospheric attenua-
tion for aircraft-to-ground propagation in SPL per 1000
feet (300 m) distance for center frequencies of 500, 1000,
2000, 4000, and 8000 Hz. Notice that the atmospheric
attenuation is more signifcant at higher frequencies
than lower ones. For example, suppose that we are
1000 feet (300 m) away from a noise source and that the
atmospheric temperature and relative humidity is 80°F
(27°C) and 10%. The plots in Figure 8.14 show that the
atmospheric attenuation for 500 Hz is approximately 2
dB whereas for 8000 hertz the attenuation is 55 dB. For
an example, see Section 8.5.3.
Atmospheric attenuation, outdoors, can also be
affected by turbulence, fog, rain, or snow. Typically,
the more turbulence in the air, the more the sound is
attenuated. There appears to be conficting evidence
as to whether or not fog attenuates sound. It is recom-
mended that no excess attenuation be assigned to fog or
light precipitation.
8.2 Industrial Noise Pollution
Thus far, sound has been discussed. So what is noise?
An all-encompassing defnition would be that noise is
any undesirable sound. By saying this, the concept is
introduced that what is considered to be noise is some-
what relative and depends on several temporal and
circumstantial factors.
For example, it is not unusual for a person to encoun-
ter SPLs of 100 to 110 dB at a sporting event, in a stadium
full of cheering fans, and yet not be perturbed by it. On
the contrary, the barely 45 dB sound of a dripping faucet
may cause considerable annoyance in the quiet of the
night. Table 8.6 gives some typical noise levels for vari-
ous scenarios.
Industrial noise pollution is a major concern for soci-
ety as a whole. In a recent survey, the effects of exposure
to noise in refnery workers was studied extensively. A
cross-section of workers in different divisions/units
was chosen. It was found that noise levels averaged
87 to 88 dBA in the aromatic and paraffn facilities and
89 dBA in alkylation facilities. In comparison, work-
ers in the warehouse, health clinics, laboratories, and
offces were, generally, found to be exposed to much
lower levels.
Noise can damage hearing, and can cause physical or
mental stress (increased pulse rate, high blood pressure,
Temperature°C
–10
355/710 Hz, 4th octave band, GMF 500 Hz
710/1400 Hz, 5th octave band, GMF 1000 Hz
1400/2800 Hz, 6th octave band, GMF 2000 Hz
10
0
10
0
20
10
0
0 10
–10
2800/5600 Hz
7th octave band
GMF 4000 Hz
0 10 20 30 40
100
80
60
40
20
0
20 30 40 50 60
Temperature, °F
Temperature°C
70 80 90 100
0
0
10
20
30
10 20 30 40
–10 0 10 20 30 40
160
140
120
100
80
d
B
/
1
0
0
0

m
d
B
/
1
0
0
0

m
60
40
20
0
50 60
0
50
40
A
t
m
o
s
p
h
e
r
i
c

a
t
t
e
n
u
a
t
i
o
n
,

d
B
/
1
0
0
0

f
t
A
t
m
o
s
p
h
e
r
i
c

a
t
t
e
n
u
a
t
i
o
n
,

d
B
/
1
0
0
0

f
t
A
t
m
o
s
p
h
e
r
i
c

a
t
t
e
n
u
a
t
i
o
n
,

d
B
/
1
0
0
0

f
t
30
20
10
0
10 20 30 40 50 60 70 80 90 100
Temperature, °F
Temperature,°C
Temperature, °F
70 80 90 100
0 10 20 30 40
40
20
0
40
d
B
/
1
0
0
0

m
d
B
/
1
0
0
0

m
d
B
/
1
0
0
0

m
Relative
humidity, %
20
10
20
30
50
70
90
Relative
humidity,
%
10
5600/11,200 Hz
8th octave band
GMF 8000 Hz
20
30
50
60
70
90
0
60
40
20
0
Relative
humidity, %
10
20
30
50
70
90
Figure 8.14
Atmospheric attenuation.
© 2011 by Taylor and Francis Group, LLC
194 Industrial Combustion Testing
nervousness, sleep disorders, lack of concentration, and
irritability). Irreparable damage can be caused by sin-
gle transient sound events with peak levels exceeding
140 dBA (e.g., shots or explosions). Long duration noise
exceeding 85 dBA can lead to short-term reversible hear-
ing impairment and long-term exposure to higher levels
than 85 dBA can cause permanent hearing loss.
The following is a mathematical model based on
empirical data (ISO 1999) used to calculate the maxi-
mum permissible continuous noise level at the work
place that will not lead to permanent hearing loss:
L
A,m
 < 85 + 10 log(24/T
n
) dBA, (8.3)
where T
n
 = daily noise exposure time in hours.
Wearing ear protection devices at continuous noise
levels greater than 85 dBA can prevent or reduce the
danger of permanent hearing damage.
8.2.1 OSHa requirements
Title 29 CFR, Section 1910.95 of the Occupational Safety
and Health Act (OSHA) pertains to the protection of
workers from potentially hazardous noise. Table 8.7
shows OSHA permissible noise exposure levels.
Required by OSHA, the employer must provide pro-
tection against the effects of noise exposure when the
sound levels exceed those shown in Table 8.7. When the
daily noise exposure consists of two or more periods of
noise exposure at different levels, their combined effect
should be considered rather than the individual effects
of each. According to OSHA the exposure factor (EF) is
defned as [5]:
EF = C
1
/T
1
 + C
2
/T
2
 + C
3
/T
3
 + … + C
n
/T
n
, (8.4)
where C
n
is the total time of exposure at a specifc noise
level and T
n
is the total time of exposure permitted at
that level and shown in Table 8.7. If the EF exceeds 1.0,
the employee’s exposure is above OSHA limits. If OSHA
identifes such a situation, a citation may be issued and a
grace period defned in which the employer must correct
the violation or face penalties of even $10,000 per day.
8.2.2 international requirements
Regulations aimed at protecting individuals from indus-
trial noise pollution have been enforced in almost all
industrialized countries. The noise caused in industries
and the work place is generally taken as a serious issue.
Most countries have adopted 85 dBA as the limit for
the permissible noise. At any workplace with sound
levels exceeding 85 dBA, ear protection devices must
be worn, and workers exposed to this level should have
their hearing level checked periodically.
8.2.3 Noise Sources and environment interaction
The main individual sources of noise in chemical and
petrochemical plants are: burners (process furnaces,
steam boilers, and fares), fans, compressors, blowers,
pumps, electric motors, steam turbines, gears, valves,
exhausts to open air, conveyors and silos, airborne
splash noise from cooling towers, coal mills, and load-
ing and unloading of raw and fnished materials.
Although noise pollution caused by industrial
sectors are minor as compared to the noise caused by
road and rail traffc, industrial noise receives more
attention due to public representation. The ISO 1996
provides a set of international regulations for noise
protection in residential neighborhoods located near
industrial areas.
National or local authorities must enforce noise-
limiting values that should not be exceeded in the
neighborhood. The magnitude of limiting value, addi-
tional charges for tonality and impulsive noise, and the
legalities change not only from country to country, but
sometimes within different states and regions in the
same country. In general, nighttime noise limits are
10 to 15 dB lower than that for the daytime.
Table 8.6
Sound Levels of Various Sources
Threshold of hearing 0 dBA
Rustle of leaves 10 dBA
Normal conversation (at 1 m) 30 dBA
Min. level in Chicago at night 40 dBA
City street, very busy traffc 70 dBA
Noisiest spot at Niagara Falls 85 dBA
Threshold of pain 120 dBA
Jet engine (at 50 m) 130 dBA
Rocket (at 50 m) 200 dBA
Table 8.7
OSHA Permissible Noise Exposure
Duration per
Day (Hours)
Sound Pressure Level,
dBA (Slow Response)
8.0 90
6.0 92
4.0 95
3.0 97
2.0 100
1.5 102
1.0 105
0.5 110
0.25 or less 115
Note: Exposure to impulsive or impact
noise should not exceed 140 dBA.
© 2011 by Taylor and Francis Group, LLC
Combustion Noise 195
8.3 Mechanisms of Industrial
Combustion Equipment Noise
There are four major mechanisms of noise production in
combustion equipment. They can be categorized as pre-
dominantly high frequency or low frequency sources.
They are:
Low Frequency Noise Sources
Combustion roar and instability
Fan noise
High Frequency Noise Sources
Gas jet noise
Piping and valve noise
8.3.1 Combustion roar and Combustion
instability Noise
In order to understand combustion roar, the mixing
process taking place between the fuel and the oxidant
on a very minute scale must be considered. It is known
that a well-blended mixture of fuel and air will combust
very rapidly if the mixture is within the fammability
limits for that fuel. On the other hand, a raw fuel stream
that depends on turbulence and momentum to mix in
the ambient fuid, to create a fammable mixture, tends
to create a slower combustion process. In either case,
when regions in the mixing process achieve a famma-
ble mixture and encounter a source of ignition, combus-
tion takes place. The closer a mixture is to stoichiometry
when it encounters the ignition source, the more rapid
will be the combustion. Combustion occurring close
to stoichiometry converts more of the energy release
into noise. Figure 8.15 shows an example of a test fare
operated at the same fow rate of fuel, but with different
degrees of premixing. In the left picture, the fuel fow is
well mixed with air and thus quickly burns in a crisp
smokeless fame. In the picture on the right, the same
fuel fow rate is burned with very poor premixing and
a slowly burning large fame is generated. The differ-
ence in overall PWL of the noise generated by the com-
bustion process itself is signifcant: For the operational
mode shown in the left picture, it is about 5 dB higher
than for the operational mode in the right picture.
See the discussion on thermoacoustic effciency
(TAE) later in this section for more details. The noise
coming from each small region of rapidly combusting
mixture adds up to create what we call combustion
roar. Therefore, combustion roar is largely a function
of how rapidly the fuel is being burned. In addition, in
the context of combustion equipment like burners and
fares, usually the larger the fuel release, the more the
turbulence in the combustion process. Since turbulence
directly infuences the mixing rate, high turbulence
processes also produce more combustion roar. So it is
more accurate to state that the level of combustion roar
generated from a combustion process is a function of
the amount of fuel being burned and how rapidly we
arrange to burn it.
8.3.1.1 Flare Combustion Roar
It has been recognized for a long time that the noise
emitted from a normal operating fare has two mech-
anisms at work, namely, combustion roar and gas jet
noise. Combustion roar typically resides in the lower
frequency region of the audible frequency spectrum
while gas jet noise occurs in the higher frequencies as
illustrated in Figure 8.16.
5

m
5

m
5

m
5

m
Figure 8.15
Test fare at John Zink test site in Tulsa. Combustion of identical fuel
fow rates with different degrees of mixing. Total noise emissions of
the fame in the left picture are about 5 dB higher than for the fame
in the right picture.
Region of
combustion
roar
Region of
gas jet
noise
Noise frequency (Hertz)
N
o
i
s
e

l
e
v
e
l
Figure 8.16
Illustration showing typical noise signature emitted from a fare
operating under normal operating conditions.
© 2011 by Taylor and Francis Group, LLC
196 Industrial Combustion Testing
As mentioned before, the amount of combustion roar
emitted from a fare, generally, depends on how fast the
waste gas stream mixes with the ambient air. A waste
gas stream that exits a fare tip with a low velocity and
low levels of turbulence will mix slowly with the ambi-
ent air and burn relatively quietly. These types of fames
are called buoyancy-dominated fames. A waste gas
stream that exits a fare tip with a high velocity and high
levels of turbulence, however, will burn much faster and
create substantially more combustion noise for the same
heat release rate. These high velocity fames are momen-
tum-dominated fames. Increasing the rate at which the
waste gas burns results in “bigger explosions” of the
air–fuel mixture. These bigger explosions create larger
disturbances in the atmosphere resulting in higher
levels of combustion roar.
The combustion roar emitted from a fare fame is
not highly directional and is considered to be a mono-
pole source. That is, it is analogous to a spherical bal-
loon whose surface is expanding and shrinking at
various frequencies and emitting uniform spherical
waves.
High levels of turbulence in a fare fame are usually
desirable because it helps reduce radiation and increase
the smokeless capacity of the fare. Unfortunately, it
increases the combustion roar. Unlike the solution for
fare radiation reduction, it is not practical to increase
the height of a fare stack or boom to reduce combus-
tion noise. This is because even doubling the fue
stack height will only reduce the SPL at the fare base
by about 6 dB (see Equation 8.2). In addition, combus-
tion roar is low frequency sound and so can travel a
great distance without being substantially attenuated
by the atmosphere. The signature of low frequency
combustion roar noise typically consists of a broad-
band spectrum with a single peak.
The combustion roar emitted from a stable burning
fare typically peaks at a frequency of about 63 Hz. The
combustion noise spectrum can be estimated from the
overall sound pressure level (OASPL) by using the val-
ues in Table 8.8 [6]. It can be noted that, at frequencies
above about 500 Hz, the noise contribution from com-
bustion is relatively insignifcant. A typical method for
estimating the PWL emitted from a fare fame is to
relate the energy released from the combustion of the
waste gas stream (chemical energy) to the noise energy
liberated by the combustion. The ratio of noise energy
to chemical energy released from the combustion is
called the TAE. For a stable-burning fare, the TAE typ-
ically varies between 1 × 10
−9
to 3 × 10
−6
. The value of
the TAE largely depends on the turbulent mixing of the
waste gas with ambient air and is usually determined
experimentally.
In Figure 8.17, the sound power emitted by a num-
ber of industrial fares at different loads is shown. The
data are obtained from feld measurements on several
types of fares under various operating conditions [7].
The noise intensity is given as the sound power level L
w
according to Equation 8.5 and is plotted versus the heat
release rate of the fare

Q
combust
, for which a logarithmic
scale is used:
L
W
w

|
.

`
}


10
1 10
12
log
Watts
acoust
×
. (8.5)
It is important to emphasize that in this context “oper-
ating condition” does not only refer to the heat release
rate of the fare alone but also to all other parameters
characterizing its operation, as for example, fow rates
of auxiliary equipment (steam or air injection, etc.).
Accordingly, fares shown in Figure 8.17 can operate
under different conditions at the same heat release rate.
Table 8.8
Calculation of the Typical Combustion Noise
Spectrum of a Stable Burning Flare from the
Overall Sound Pressure Level OASPL
Frequency (Hz) Resultant Noise Spectrum (dB)
31.5 OASPL – 5
63 OASPL – 4
125 OASPL – 9
250 OASPL – 15
500 OASPL – 20
1,000 OASPL – 21
2,000 OASPL – 24
4,000 OASPL – 28
8,000 OASPL – 34
175
165
155
145
135
125
115
10 100 1000 10.000
EF
EF-SSA
EF-SSS
EF-SSS-NC
GF-SSS-NC
TAE = 1.9E-8
TAE = 4.5E-5
Q
combust
, megawatts
L
w
A

,

d
B
Figure 8.17
Sound power level L
w

calculated from measured noise data, plotted
versus heat release rate

Q
combust
for different types of industrial fares
under various operating conditions. (From Müller-BBM GmbH.
“Noise Emissions of Different Flare Systems—Field Measurements
Taken in Various Refneries and Petrochemical Plants.” Proprietary
data, 1900–2004.) EF = elevated (single-point) fare, GF = (enclosed)
ground fare, SSA = smoke suppression by air, SSS = smoke suppres-
sion by steam, NC = equipped with advanced noise control.
© 2011 by Taylor and Francis Group, LLC
Combustion Noise 197
In Figure 8.17 lines of constant TAE are straight
lines. The corresponding lines for the maximum and
minimum TAE observed in the measurements—that is
TAE = 1.9 × 10
−8
and 4.5 × 10
−5
—are shown as the dashed
and dashed-dotted lines.
In designing fares, the combustion noise emitted
from fares operating under various conditions is usu-
ally measured in order to determine the TAE. This infor-
mation can then be used in computer programs that are
used to model the level of combustion roar emitted from
a fare. Figure 8.18 is a photograph of a John Zink engi-
neer collecting noise levels from a fare using a real-time
noise level meter.
A fare fame that is highly turbulent, such as the
high-pressure fare as shown in Figure 8.18, may have a
TAE in the order of 1 × 10
−6
. However, a fame with low
levels of turbulence, such as the butane cigarette lighter
as shown in Figure 8.19 may have a TAE in the order
of 1 × 10
−9
. For every order of magnitude that the TAE
changes, the SPL will change by 10 dB.
Example: Consider a fare burning a waste gas stream
with a heat release of 5000 × 106 BTU/hr (1465 million
watts). A noise measurement shows that the sound pres-
sure level 400 feet (120 m) from the fame is 100 dB. What
is the TAE of the fare fame?
The sound power emitted from the fame, W, can be
determined as follows:

W
anti
L r
p
( )
log
log ( )
watts
×
+ −

1 10
10 4 1
12
10
10
2
π 00 5
10
.
,
|
.

`
}

]
]
]
(8.6)
where L
p
is the sound pressure level in dB (100 dB for
the example) and r is the distance from the fame in feet
(400 feet or 120 m for the example). Substituting these
values into Equation 8.6 gives W = 1792 watts. The TAE
is then calculated to be
TAE
Acoustical Power
Thermal Power
watts

1792
14665 10
1 2 10
6
6
×
×

watts
. .
(8.7)
Since the TAE is in the order of magnitude of 1 × 10
−6
,
one would expect that the fame would be highly turbu-
lent and momentum-dominated.
A central problem of the approach described above
certainly lies in applying it to a whole fare system
instead of the combustion process alone. Doing so
means that TAE becomes a “lumped parameter” incor-
porating all effects that have an impact on the acous-
tical behavior of the fare (i.e., noise emissions from
valves, injectors, smoke suppression devices etc.) as
well as any noise control measures installed. Other
approaches that are similar in that respect can be found
in the literature [8].
Another problem with using the TAE concept to
predict combustion noise is that it does not take the
frequency characteristics of the noise emissions into
account. Frequency characteristics are very important
for the design of adequate noise control measures and
will also determine the human perception of the noise
emissions since the ear’s sensitivity for noise is fre-
quency-dependent (see Section 8.1).
Figure 8.18
Photograph of an engineer collecting fare noise data.
Figure 8.19
Shadow photograph of a burning butane lighter.
© 2011 by Taylor and Francis Group, LLC
198 Industrial Combustion Testing
Additional challenges arise for spatially extended fare
systems that can no longer be treated as point sources.
Prediction of the noise emissions of a grade-mounted
multipoint fare system requires the correct modeling
not only of the individual burners and their combustion
and jet noise emissions, but also of the arrangement of
the burners in the fare pit and the effect of the wind
fence. Figure 8.20 shows the calculated sound pressure
feld contour plots for an LRGO fare system at different
operating conditions.
In summary, it can be concluded that the TAE con-
cept applied to fare systems will usually only allow a
very rough estimate of the actual noise emissions and
the associated effect of these emissions on persons in
the neighborhood of the fare. In some cases, the results
can even be completely off the real situation. A key issue
in developing reliable noise prediction tools for fares
that are more generally applicable in a broad range of
operating conditions lies in a proper treatment of the
individual sources that contribute to the overall noise
emissions of a fare system.
8.3.1.2 Flare Combustion Instability Noise
If a fame lifts too far above a fare tip, it can become
unstable. An unstable fame will periodically lift and
then reattach to the fare tip and create a low frequency
rumbling noise. Typically, this low rumbling noise
occurs in the frequency range of 5–10 Hz and is usually
called combustion instability. Being as low in frequency
as it is, combustion instability noise is usually inaudible
and can travel over several miles without being substan-
tially attenuated by the atmospheric air. When there are
reports of shaking the walls and windows of buildings
in the vicinity of a fare, it is usually due to combustion
instability.
Combustion instability noise can occur if too much
steam or air is added in the base of the fame. So, over-
aerating the waste gas stream in a fare, causes the fame
to periodically lift from the fare tip. This periodic lift-
ing and reattachment of the fame from the fare tip
is the mechanism that drives the low frequency rum-
bling noise. Usually, combustion instability noise can
> 60.0 dB(A)
POI@100 m
POI@50 m
POI@30 m
POI@1 m
POI@100 m
POI@50 m
POI@30 m
POI@1 m
POI@100 m
POI@50 m
POI@30 m
POI@1 m
> 65.0 dB(A)
> 70.0 dB(A)
> 75.0 dB(A)
> 80.0 dB(A)
> 85.0 dB(A)
> 90.0 dB(A)
> 95.0 dB(A)
> 100.0 dB(A)
> 105.0 dB(A)
> 110.0 dB(A)
> 115.0 dB(A)
Point source
Vert. area source
Barrier
Receiver
Calculation area
Figure 8.20
(See color insert following page 424.) Predicted sound pressure feld contour plots for a multipoint LRGO fare system. Left: frst stage in
operation; Middle: stages 1 and 2 in operation; Right: all stages in operation.
© 2011 by Taylor and Francis Group, LLC
Combustion Noise 199
be reduced by lowering the steam fow rate to a steam-
assisted fare or by lowering the blower airfow rate to
an air-assisted fare. Figure 8.21 is a graph showing
an example of a typical steam assisted fare operating
at both a normal condition as well as an oversteamed
condition [9]. Note that the combustion noise frequency
shifts substantially to a lower region and the level dra-
matically increases when the fare is oversteamed.
8.3.1.3 Burner Combustion Noise
Like fares, burner combustion noise is an unwanted
sound associated with combustion roar and combustion
instability. In many situations, the combustion noise can
be the dominant source of noise emitted from a burner.
Combustion roar and combustion instability are quite
complex by nature. The literature contains a variety of
combustion noise and combustion instability prediction
techniques for burners operating in a furnace. Most of
these prediction techniques are based on experimental
studies that attempt to correlate the acoustic power radi-
ated by the burner/furnace geometry, laminar burning
velocity of the air–fuel mixture, and various turbulence
parameters such as the turbulent length scale and inten-
sity. This section will not attempt to discuss these pre-
diction techniques in detail but will give a broad and
general discussion of combustion roar and combustion
instability noise using some of the results from these
studies.
Figure 8.12 is a plot showing a typical noise spec-
trum emitted from a burner operating under normal
conditions in a furnace. Notice that the noise spectrum
has two peak frequencies associated with it. The high fre-
quency noise contribution is from the fuel gas jets while
the low frequency contribution is from the combus-
tion roar. Like the combustion roar emitted from fares,
burner combustion roar is associated with a smooth
broadband spectrum having relatively low conversion
effciency from chemical energy to noise: in the range
of 1 × 10
−9
–1 × 10
−6
. However, the combustion noise spec-
trum associated with a burner and a fare are not similar.
The reason is that a fame burning in the open atmo-
sphere will behave differently compared to a fame that
is burning in an enclosed chamber such as a furnace.
The combustion roar associated with fares typically
peaks at a frequency of approximately 63 Hz while
combustion roar associated with burners can vary in
the 200–500 Hz range. Burner noise can have a spec-
trum shape and amplitude that can vary with many fac-
tors. Several of these factors include the internal shape
of the furnace, the design of the burner muffer, plenum
and tile, the acoustic properties of the furnace lining,
the transmission of the noise into the fuel supply piping,
and the transmissive and refective characteristics of the