HANDBOOK OF HEAT TRANSFER Warren M. Rohsenow
Editor
Department of Mechanical Engineering Massachusetts Institute of Technology
James R Hartnett
Editor Energy Resources Center University of Illinois at Chicago
Young I. Cho Editor Department of Mechanical Engineering and Mechanics Drexel University
Third Edition
MCGRAWHILL New York San Fran©isco Washington, D.C. Auckland Bogot6 Caracas Lisbon London Madrid Mexi©oCity Milan Montreal New Delhi San Juan Singapore Sydney Tokyo Toronto
Library of Congress CataloginginPublication Data H a n d b o o k of heat transfer / editors, W.M. Rohsenow, J.P. H a r t n e t t , Y.I. Cho. m 3rd ed. p. cm. Includes bibliographical references and index. I S B N 0070535558 (alk. p a p e r ) 1. H e a t   T r a n s m i s s i o n m H a n d b o o k s , manuals, etc. 2. Mass t r a n s f e r m H a n d b o o k s , manuals, etc. I. Rohsenow, W. M. ( W a r r e n M.) II. H a r t n e t t , J. E (James E) III. Cho, Y. I. (Young I.) QC320.4.H36 1998 621.402'2dc21 9751381 CIP
McGrawHill A Division o[ The McGraw.Hill ~ i e s
Copyright © 1998 by The McGrawHill Companies, Inc. All rights reserved. Printed in the United States of America. Except as permitted under the United States Copyright Act of 1976, no part of this publication may be reproduced or distributed in any form or by any means, or stored in a data base or retrieval system, without the prior written permission of the publisher. 1 2 3 4 5 6 7 8 9 0
DOC/DOC
9 0 3 2 1 0 9 8
ISBN 0070535558
The sponsoring editor for this book was Robert Esposito, the editing supervisor was Stephanie S. Landis, and the production supervisor was Pamela A. Pelton. It was set in Times Roman by North Market Street Graphics. Printed and bound by R. R. Donnelley & Sons Company. McGrawHill books are available at special quantity discounts to use as premiums and sales promotions, or for use in corporate training programs. For more information, please write to the Director of Special Sales, McGrawHill, 11 West 19th Street, New York, NY 10011. Or contact your local bookstore.
Information contained in this work has been obtained by The McGrawHill Companies, Inc. ("McGrawHill") from sources believed to be reliable. However, neither McGrawHill nor its authors guarantee the accuracy or completeness of any information published herein, and neither McGrawHill nor its authors shall be responsible for any errors, omissions, or damages arising out of use of this information. This work is published with the understanding that McGrawHill and its authors are supplying information but are not attempting to render engineering or other professional services. If such services are required, the assistance of an appropriate professional should be sought.
This book is printed on acidflee paper.
CONTRIBUTORS
Bergles, Arthur E. Department of Mechanical Engineering, Rensselaer Polytechnical Institute (CHAP. 11, Techniques to Augment Heat Transfer), email:
[email protected] Bergman, Theodore L. Departmentof Mechanical Engineering, University of Connecticut (CHAP. 18, Heat Transfer in Materials Processing), email:
[email protected] Chauk, Shriniwas Departmentof Chemical Engineering, Ohio State University (CHAP. 13, Heat Transfer in Fluidized and Packed Beds) Chen, PingHai Department of Mechanical Engineering, National Taiwan University, Taiwan, ROC (CHAP. 16, Measurement of Temperature and Heat Transfer), email:
[email protected] Chiang, Hwai Derg IndustrialTechnology Research Institute, Taiwan, ROC (CHAP.16, Measurement of Temperature and Heat Transfer), email:
[email protected] Cho, Young I. Departmentof Mechanical Engineering and Mechanics, Drexel University (CHAP.1, Basic Concepts of Heat Transfer; CHAP.10, Nonnewtonian Fluids), email:
[email protected] Dong, Z.F. Departmentof Mechanical Engineering, Florida International University (CHAP. 5, Forced Convection, Internal Flows), email:
[email protected] Ebadian, M.A. Hemispheric Center for Environmental Technology, Florida International University (CHAP. 5, Forced Convection, Internal Flows), email:
[email protected] Fan, L.S. Departmentof Chemical Engineering, Ohio State University (CHAP.13, Heat Transfer in Fluidized and Packed Beds), email:
[email protected] Goldstein, Richard J. Department of Mechanical Engineering, University of Minnesota (cHar,. 16, Measurement of Temperature and Heat Transfer), email:
[email protected] Hartnett, J a m e s P. Energy Resources Center, University of Illinois, Chicago (CHAP. 1, Basic Concepts of Heat Transfer; CHAP.10, Nonnewtonian Fluids), email:
[email protected] Hewitt, Geoffrey F. Departmentof Chemical Engineering and Chemical Technology, Imperial College of Science, Technology and Medicine, London, UK (CHAP. 15, Boiling), email:
[email protected] Hollands, K. G.T. Department of Mechanical Engineering, University of Waterloo, Canada (CHAP. 4, Natural Convection), email:
[email protected] Howell, John R. Departmentof Mechanical Engineering, University of Texas at Austin (CHAP.7, Radiation), email:
[email protected] Inouye, Mamoru Flows)
Ames Research CenterNASA (retired) (CHAP.6, Forced Convection, External
Irvine, Thomas F., Jr. Departmentof Mechanical Engineering, State University of New York (CHAP.2, Thermophysical Properties), email:
[email protected] Kaviany, Massoud Department of Mechanics and Applied Mechanics Engineering, University of Michigan (CHAP.9, Heat Transfer in Porous Media), email:
[email protected] Majumdar, Arun Departmentof Mechanical Engineering, University of California, Berkeley (CHAP.8, Microscale Heat Transfer), email:
[email protected] i a r t o , Paul J. Departmentof Mechanical Engineering, Naval Postgraduate School (CHAP.14, Condensation), email:
[email protected] XVII
xviii
CONTRIBUTORS MengO;, M. Pinar Department of Mechanical Engineering, University of Kentucky (CHAP. 7, Radiation), email:
[email protected] Peterson, G. P. Bud Departmentof Mechanical Engineering, Texas A&M University (CHAP. 12, Heat Pipes), email:
[email protected] Parikh, Pradip G. Boeing Commercial Airplane Group (CHAP.6, Forced Convection, External Flows), email:
[email protected] Raithby, George D. Departmentof Mechanical Engineering, University of Waterloo, Canada (CHAP.4, Natural Convection), email:
[email protected] Rubesin, Morris W. Ames Research CenterNASA (retired) (CHAP. 6, Forced Convection, External Flows), email:
[email protected] Sekulic, Dusan P. Department of Mechanical Engineering, University of Kentucky (CHAP. 17, Heat Exchangers), email:
[email protected] Shah, Ramesh K. Delphi Harrison Thermal Systems, Lockport, NY (CHAP. 17, Heat Exchangers), email:
[email protected] Viskanta, Raymond Schoolof Mechanical Engineering, Purdue University (CHAP.18, Heat Transfer in Materials Processing), email:
[email protected] Yovanovich, M. Michael Departmentof Mechanical and Electrical Engineering, University of Waterloo, Canada (CHAP.3, Conduction), email:
[email protected] PREFACE
INTRODUCTION Since the publication of the second edition of Handbook of Heat Transfer, there have been many new and exciting developments in the field, covering both fundamentals and applications. As the role of technology has grown, so too has the importance of heat transfer engineering. For example, in the industrial sector heat transfer concerns are critical to the design of practically every process. The same is true of such vitally important areas as energy production, conversion, and the expanding field of environmental controls. In the generation of electrical power, whether by nuclear fission or combustion of fossil fuels, innumerable problems remain to be solved. Similarly, further miniaturization of advanced computers is limited by the capability of removing the heat generated in the microprocessors. Heat transfer problems at the macro scale, as exemplified by global warming, also offer tremendous challenges. As technology advances, engineers are constantly confronted by the need to maximize or minimize heat transfer rates while at the same time maintaining system integrity. The upper and lower boundariesmsystem size, pressure, and temperatureare constantly expanding, confronting the heat transfer engineer with new design challenges. In preparing this third edition, the goal of the editors was to provide, in a single volume, uptodate information needed by practicing engineers to deal with heat transfer problems encountered in their daily work. This new edition of the handbook contains information essential for design engineers, consultants, research engineers, university professors, students, and technicians involved with heat transfer technology.
COVERAGE The third edition of Handbook of Heat Transfer provides expanded treatment of the fundamental topics covered in earlier editions. More than half of the authors of these basic chapters on conduction, convection, radiation, condensation, and boiling are new, reflecting the fact that there are new leaders in the field. Those chapters in the second edition dealing with applications related to the socalled energy crisis (solar energy, energy storage, cooling towers, etc.) have been replaced by new chapters treating heat transfer problems encountered in materials processing, porous media, and micro scale systems. Sections on the following topics were retained and updated: thermophysical properties, heat transfer enhancement, heat exchangers, heat pipes, fluidized beds, nonnewtonian fluids, and measurement techniques.
UNITS It is recognized at this time that the English Engineering System of units cannot be completely replaced by the International System (SI). Transition from the English system of units to SI will proceed at a rational pace to accommodate the needs of the profession, industry, and the public. The transition period will be long and complex, and duality of units probably will xix
~t
PREFACE be demanded for at least one or two decades. Both SI and English units have been incorporated in this edition to the maximum extent possible, with the goal of making the handbook useful throughout the world. In general, numerical results, tables, figures, and equations in the handbook are given in both systems of units wherever presentation in dimensionless form is not given. In a few cases, some tables are presented in one system of units, mostly to save space, and conversion factors are printed at the end of such tables for the reader's convenience.
NOMENCLATURE An attempt has been made by the editors to use a unified nomenclature throughout the handbook. Given the breadth of the technical coverage, some exceptions will be found. However, with few exceptions, one symbol has only one meaning within any given section. Each symbol is defined at the end of each section of the handbook. Both SI and English units are given for each symbol in the nomenclature lists.
INDEX This edition provides a comprehensive alphabetical index designed to provide quick reference to information. Taken together with the Table of Contents, this index provides quick and easy access to any topic in the book.
ACKNOWLEDGMENTS The editors acknowledge the outstanding performance of the contributing authors. Their cooperation on the contents and length of their manuscripts and in incorporating all of the previously mentioned specifications, coupled with the high quality of their work, has resulted in a handbook that we believe will fulfill the needs of the engineering community for many years to come. We also wish to thank the professional staff at McGrawHill Book Company, who were involved with the production of the handbook at various stages of the project, for their cooperation and continued support. The outstanding editorial work of Ms. Stephanie Landis of North Market Street Graphics is gratefully acknowledged. The handbook is ultimately the responsibility of the editors. Care has been exercised to minimize errors, but it is impossible in a work of this magnitude to achieve an errorfree publication. Accordingly, the editors would appreciate being informed of any errors so that these may be eliminated from subsequent printings. The editors would also appreciate suggestions from readers on possible improvements in the usefulness of the handbook so that these may be included in future editions. W. M. Rohsenow J. E Hartnett Y. I. Cho
CONTENTS
Contributors xvii Preface xix Chapter 1. Basic Concepts of Heat Transfer
1.1
Heat Transfer Mechanisms / 1.1 Conduction / 1.1 Radiation / 1.3 Convection / 1.4 Combined Heat Transfer Mechanisms / 1.10 Conservation Equations / 1.11 The Equation of Continuity / 1.13 The Equation of Motion (Momentum Equation) / 1.14 The Energy Equation / 1.18 The Conservation Equations for Species / 1.21 Use of Conservation Equations to Set Up Problems / 1.22 Dimensionless Groups and Similarity in Heat Transfer / 1.23 Units and Conversion Factors / 1.29 Nomenclature / 1.31 References / 1.36
Chapter 2. Thermophysical Properties
2.1
Conversion Factors / 2.1 Thermophysical Properties of Gases / 2.3 Thermophysical Properties of Liquids / 2.26 Thermophysical Properties of Solids / 2.46 Thermophysical Properties of Saturated Refrigerants / 2.69 Acknowledgment / 2.73 Nomenclature / 2.73 References / 2.73 Selected Additional Sources of Thermophysical Properties / 2.?4
Chapter 3. Conduction and Thermal Contact Resistances (Conductances) Introduction / 3.1 Basic Equations, Definitions, and Relationships / 3.2 Shape Factors / 3.3 Shape Factors for Ellipsoids: Integral Form for Numerical Calculations / 3.11 Shape Factors for ThreeDimensional Bodies in Unbounded Domains / 3.15 ThreeDimensional Bodies with Layers: Langmuir Method / 3.19 Shape Factors for TwoDimensional Systems / 3.20 Transient Conduction / 3.23 Introduction / 3.23 Internal Transient Conduction / 3.23 Lumped Capacitance Model / 3.24
3.1
vi
CONTENTS Heisler and Grober ChartsSingleTerm Approximations / 3.24 Multidimensional Systems / 3.25 Transient OneDimensional Conduction in HalfSpaces / 3.26 External Transient Conduction from Long Cylinders / 3.28 Transient External Conduction from Spheres / 3.29 Instantaneous Thermal Resistance / 3.30 Transient External Conduction from Isothermal Convex Bodies / 3.31 Spreading(Constriction) Resistance / 3.34 Introduction / 3.34 Definitions of Spreading Resistance / 3.34 Spreading Resistance of Isoflux Arbitrary Areas on HalfSpace / 3.35 Circular Annular Contact Areas on HalfSpace / 3.36 Doubly Connected Isoflux Contact Areas on HalfSpace / 3.37 Effect of Contact Conductance on Spreading Resistance / 3.38 Spreading Resistance in Flux Tubes and Channels / 3.39 Effect of Flux Distribution on Circular Contact Area on HalfSpace / 3.39 Simple Correlation Equations of Spreading Resistance for Circular Contact Area / 3.40 Accurate Correlation Equations for Various Combinations of Contact Area, Flux Tubes, and Boundary Condition / 3.40 General Spreading Resistance Expression for Circular Annular Area on Circular Flux Tube / 3.41 Spreading Resistance Within TwoDimensional Channels / 3.41 Effect of Single and Multiple Layers (Coatings) on Spreading Resistance / 3.43 Circular Contact Area on Single Layer (Coating) on HalfSpace / 3.46 Circular Contact Area on Multiple Layers on Circular Flux Tube / 3.47 Transient Spreading Resistance / 3.48 Transient Spreading Resistance of Isoflux Hyperellipse Contact Area on HalfSpace / 3.49 Transient Spreading Resistance of Isoflux Regular Polygonal Contact Area on HalfSpace / 3.50 Transient Spreading Resistance Within SemiInfinite Flux Tubes and Channels / 3.50 Contact, Gap, and Joint Resistances and Contact Conductances / 3.51 Point and Line Contact Models / 3.51 Thermal Contact, Gap, and Joint Conductance Models / 3.55 Gap Conductance Model and Integral / 3.59 Acknowledgments / 3.60 Nomenclature / 3.60 References / 3.67
Chapter 4. Natural Convection Introduction / 4.1 Basics / 4.1 Equations of Motion and Their Simplification / 4.1 Problem Classification / 4.5 Heat Transfer Correlation Method / 4.6 External Natural Convection / 4.12 Flat Plates / 4.20 Cylinders / 4.26 Open Cavity Problems / 4.32 Cooling Channels / 4.32 Extended Surfaces / 4.36 Natural Convection Within Enclosures / 4.40 Introduction / 4.40 Geometry and List of Parameters for Cavities Without Interior Solids / 4.40 The Conduction Layer Model / 4.43 Horizontal Rectangular Parallelepiped and Circular Cylinder Cavities / 4.44 Heat Transfer in Vertical Rectangular Parallelepiped Cavities: 0 = 90 ° / 4.50 Heat Transfer in Inclined Rectangular Cavities / 4.55 Heat Tranfer in Enclosures with Interior Solids at Prescribed Temperature / 4.58 Partitioned Enclosures / 4.60
4.1
CONTENTS
vii
Transient Natural Convection / 4.63 External Transient Convection / 4.63 Internal Transient Convection / 4.66 Natural Convection with Internal Generation / 4.68 Internal Problems / 4.68 Convection in Porous Media / 4.69 Properties and Dimensionless Groups / 4.69 External Heat Transfer Correlations / 4.71 Internal Heat Transfer Correlations / 4.72 Mixed Convection / 4.73 External Flows / 4.73 Internal Flows / 4.78 Acknowledgments / 4.80 Nomenclature / 4.80 References / 4.87
Chapter 5. Forced Convection, Internal Flow in Ducts Introduction / 5.1 Scope of the Chapter / 5.1 Characteristics of Laminar Flow in Ducts / 5.1 Characteristics of Turbulent Flow in Ducts / 5.2 Hydraulic Diameter / 5.3 Fluid Flow Parameters / 5.3 Heat Transfer Parameters / 5.4 Thermal Boundary Conditions / 5.5 Circular Ducts / 5.5 Laminar Flow / 5.6 Turbulent Flow / 5.18 Transition Flow / 5.30 Concentric Annular Ducts / 5.32 Four Fundamental Thermal Boundary Conditions / 5.32 Laminar Flow / 5.33 Turbulent Flow / 5.50 Parallel Plate Ducts / 5.59 Laminar Flow / 5.59 Turbulent Flow / 5.65 Rectangular Ducts / 5.67 Laminar Flow / 5.67 Turbulent Flow / 5.72 Triangular Ducts / 5.73 Laminar Flow / 5.73 Turbulent Flow / 5.78 Elliptical Ducts / 5.82 Laminar Flow / 5.82 Turbulent Flow / 5.84 Curved Ducts and Helicoidal Pipes / 5.84 Fully Developed Laminar Flow / 5.85 Developing Laminar Flow / 5.90 Turbulent Flow in Coils with Circular Cross Sections / 5.90 Fully Developed Laminar Flow in Curved, Square, and Rectangular Ducts / 5.91 Fully Developed Turbulent Flow in Curved Rectangular and Square Ducts / 5.92 Laminar Flow in Coiled Annular Ducts / 5.92 Laminar Flow in Curved Ducts with Elliptic Cross Sections / 5.92 Longitudinal Flow Between Cylinders / 5.93 Laminar Flow / 5.93 Fully Developed Turbulent Flow / 5.97
5.1
viii
CONTENTS
Internally Finned Tubes / 5.99 Circular Ducts with Thin Longitudinal Fins / 5.100 Square Ducts with Thin Longitudinal Fins / 5.101 Rectangular Ducts with Longitudinal Fins from Opposite Walls / 5.101 Circular Ducts with Longitudinal Triangular Fins / 5.101 Circular Ducts with Twisted Tape / 5.102 Semicircular Ducts with Internal Fins / 5.104 Elliptical Ducts with Internal Longitudinal Fins / 5.104 Other Singly Connected Ducts / 5.105 Sine Ducts / 5.105 Trapezoidal Ducts / 5.106 RhombicDucts / 5.107 Quadrilateral Ducts / 5.107 Regular Polygonal Ducts / 5.107 Circular Sector Ducts / 5.108 Circular SegmentDucts / 5.108 Annular Sector Ducts / 5.110 StadiumShapedDucts / 5.111 MoonShaped Ducts / 5.113 Corrugated Ducts / 5.113 Parallel Plate Ducts with Spanwise Periodic Corrugations at One Wall / 5.115 Cusped Ducts / 5.116 Cardioid Ducts / 5.117 Unusual Singly Connected Ducts / 5.117 Other Doubly Connected Ducts / 5.117 Confocal Elliptical Ducts / 5.117 Regular Polygonal Ducts with Centered Circular Cores / 5.118 Circular Ducts with Centered Regular Polygonal Cores / 5.118 Isosceles Triangular Ducts with Inscribed Circular Cores / 5.120 Elliptical Ducts with Centered Circular Cores / 5.120 Concluding Remarks / 5.120 Nomenclature / 5.120 References / 5.125
Chapter 6. Forced Convection, External Flows
6.1
Introduction / 6.1 Definition of Terms / 6.2 TwoDimensional Laminar Boundary Layer / 6.2 Uniform FreeStream Conditions / 6.2 Surface with Streamwise Pressure Gradient / 6.28 TwoDimensional Turbulent Boundary Layer / 6.46 Turbulence Transport Mechanisms and Modeling / 6.46 Uniform FreeStream Conditions / 6.54 Transitional Boundary Layers / 6.72 Transitional Boundary Layers for Uniform FreeStream Velocity / 6.72 Complex Configurations / 6.74 Nomenclature / 6.75 References / 6.80
Chapter 7. Radiation Introduction / 7.1 Radiation Intensity and Flux / 7.2 Blackbody Radiation / 7.3 Nonblack Surfaces and Materials / 7.6
7.1
CONTENTS Radiative Exchange: Enclosures Containing a Nonparticipating Medium Black Surfaces / 7.12 Exchange Among Gray Diffuse Surfaces / 7.16 Radiative Exchange with a Participating Medium / 7.19 Fundamentals and Definitions / 7.19 Solution Techniques for the RTE / 7.24 Solutions to Benchmark Problems / 7.43 Radiative Properties for Participating Media / 7.44 Radiative Properties of Gases / 7.44 Radiative Properties of Particulates / 7.55 Radiative Properties of Porous Materials / 7.66 Radiative Properties of Semitransparent Materials / 7.69 Combined Modes with Radiation / 7.70 The General Energy Equation / 7.70 Interaction with Conduction and Convection / 7.71 Interaction with Combustion and Turbulence / 7. 71 Closing Remarks / 7.72 Appendix A: Radiative Property Tables / 7.73 Appendix B: Radiation Configuration Factors / 7.76 Nomenclature / 7.84 References / 7.86
/
ix
Z12
Chapter 8. Microscale Transport Phenomena
8.1
Introduction / 8.1 Time and Length Scales / 8.2 Kinetic Theory / 8.3 Formulation / 8.3 Thermal Conductivity of Crystalline and Amorphous Solids / 8.5 Boltzmann Transport Theory / 8.9 General Formulation / 8.9 Fourier and Ohm's Laws / 8.11 Hyperbolic Heat Equation / 8.12 Mass, Momentum, and Energy ConservationHydrodynamic Equations / 8.12 Equation of Radiative Transfer for Photons and Phonons / 8.15 Nonequilibrium Energy Transfer / 8.16 Joule Heating in HighField Electronic Devices / 8.17 Radiative Heating by Ultrashort Laser Pulses / 8.21 Summary / 8.23 Nomenclature / 8.24 References / 8.25
Chapter 9. Heat Transfer in Porous Media Introduction / 9.1 SinglePhase Flow / 9.4 Conduction Heat Transfer / 9.4 Convection Heat Transfer / 9.7 Radiation HeatTransfer / 9.13 TwoMedium Treatment / 9.32 TwoPhase Flow / 9.35 Momentum Equations for LiquidGas Flow / 9.36 Local Volume Averaging of Energy Equation / 9.38 Effective Thermal Conductivity / 9.41 Thermal Dispersion / 9.42 Phase Change / 9.44 Condensation at Vertical Impermeable Bounding Surfaces / 9.44 Evaporation at Vertical Impermeable Bounding Surfaces / 9.51
9.1
x
CONTENTS Evaporation at Horizontal Impermeable Bounding Surfaces / 9.52 Evaporation at Thin PorousLayerCoated Surfaces / 9.58 Melting and Solidification / 9.60 Nomenclature / 9.68 Glossary / 9.72 References / 9. 76
Chapter 10. Nonnewtonian Fluids
10.1
Introduction / 10.1 Overview / 10.1 Classification of Nonnewtonian Fluids / 10.1 Material Functions of Nonnewtonian Fluids / 10.2 Rheological Property Measurements / 10.3 Thermophysical Properties of Nonnewtonian Fluids / 10. 7 Governing Equations of Nonnewtonian Fluids / 10.8 Use of Reynolds and Prandtl Numbers / 10.9 Use of the Weissenberg Number / 10.11 Laminar Nonnewtonian Flow in a Circular Tube / 10.11 Velocity Distribution and Friction Factor / 10.11 Fully Developed Heat Transfer / 10.13 Laminar Heat Transfer in the Thermal Entrance Region / 10.13 Laminar Nonnewtonian Flow in a Rectangular Duct / 10.14 Velocity Distribution and Friction Factor / 10.14 Fully Developed Heat TransfermPurely Viscous Fluids / 10.17 Heat Transfer in the Thermal Entrance RegionPurely Viscous Fluids / 10.21 Laminar Heat Transfer to Viscoelastic Fluids in Rectangular Ducts / 10.23 Turbulent Flow of Purely Viscous Fluids in Circular Tubes / 10.29 Fully Established Friction Factor / 10.29 Heat Transfer / 10.30 Turbulent Flow of Viscoelastic Fluids in Circular Tubes / 10.31 Friction Factor and Velocity Distribution / 10.31 Heat Transfer / 10.35 Degradation / 10.38 Solvent Effects / 10.40 Failure of the ReynoldsColburn Analogy / 10.41 Turbulent Flow of Purely Viscous Fluids in Rectangular Ducts I 10.42 Friction Factor / 10.42 Heat Transfer / 10.43 Turbulent Flow of Viscoelastic Fluids in Rectangular Ducts / 10.43 Friction Factor / 10.43 Heat Transfer / 10.44 Anomalous Behavior of Aqueous Polyacrylic Acid Solutions I 10.45 Flow over Surfaces; Free Convection; Boiling / 10.45 Flow over Surfaces / 10.45 Free Convection / 10.45 Boiling / 10.46 Suspensions and Surfactants / 10.46 Flow of Food Products / 10.46 Electrorheological Flows / 10.46 Nomenclature / 10.46 References / 10.49
Chapter 11. Techniques to Enhance Heat Transfer Introduction / 11.1 General Background / 11.1 Classification of Heat Transfer Enhancement Techniques / 11.1 Performance Evaluation Criteria / 11.3
11.1
CONTENTS
xi
Treated and Structured Surfaces / 11.6 Boiling / 11.6 Condensing / 11.9 Rough Surfaces / 11.9 SinglePhase Flow / 11.9 Boiling / 11.15 Condensing / 11.15 Extended Surfaces / 11.16 SinglePhaseFlow / 11.16 Boiling / 11.21 Condensing / 11.24 Displaced Enhancement Devices / 11.29 SinglePhaseFlow / 11.29 Flow Boiling / 11.32 Condensing / 11.33 SwirlFlow Devices / 11.34 SinglePhase Flow / 11.34 Boiling / 11.38 Condensing / 11.40 SurfaceTension Devices / 11.41 Additives for Liquids / 11.41 Solid Particles in SinglePhase Flow / 11.41 Gas Bubbles in SinglePhase Flow / 11.41 Liquid Additives for Boiling / 11.42 Additives for Gases / 11.44 Solid Particles in SinglePhase Flow / 11.44 Liquid Drops in SinglePhase Flow / 11.45 Mechanical Aids / 11.45 Stirring / 11.45 Surface Scraping / 11.46 Rotating Surfaces / 11.46 Surface Vibration / 11.46 SinglePhaseFlow / 11.46 Boiling / 11.49 Condensing / 11.49 Fluid Vibration / 11.49 SinglePhase Flow / 11.49 Boiling / 11.51 Condensing / 11.52 Electric and Magnetic Fields / 11.52 Injection / 11.54 Suction / 11.55 Compound Enhancement / 11.55 Prospects for the Future / 11.56 Nomenclature / 11.57 References / 11.60
Chapter 12. Heat Pipes Introduction / 12.1 Fundamental Operating Principles / 12.2 Capillary Limitation / 12.3 Other Limitations / 12.8 Design and Manufacturing Considerations / 12.10 Working Fluid / 12.11 Wicking Structures / 12.1I Materials Compatibility / 12.12 Heat Pipe Sizes and Shapes / 12.12 Reliability and Life Tests / 12.13
12.1
xii
CONTENTS
Heat Pipe Thermal Resistance / 12.14 Types of Heat Pipes / 12.15 VariableConductance Heat Pipes / 12.15 MicroHeat Pipes / 12.16 Nomenclature / 12.17 References / 12.18
Chapter 13. Heat Transfer in Packed and Fluidized Beds
13.1
Introduction / 13.1 Hydrodynamics / 13.3 Packed Beds / 13.3 Fluidized Beds / 13.4 Heat Transfer in Packed Beds / 13.8 ParticletoFluid Heat Transfer / 13.9 Effective Thermal Conductivity / 13.9 WalltoBed Heat Transfer / 13.13 Relative Heat Transfer / 13.14 Heat Transfer in Fluidized Beds / 13.14 GasSolid Fluidized Beds / 13.14 LiquidSolid Fluidized Beds / 13.34 Concluding Remarks / 13.37 Nomenclature / 13.38 References / 13.41
Chapter 14. Condensation Introduction / 14.1 Modes of Condensation / 14.1 Condensation Curve / 14.2 Thermal Resistances / 14.2 Film Condensation on a Vertical Plate / 14.4 Approximate Analysis / 14.4 Boundary Layer Analysis / 14.10 Film Condensation on Horizontal Smooth Tubes / 14.15 Single Tube / 14.15 Tube Bundles / 14.17 Film Condensation on Horizontal Finned Tubes / 14.22 Single Tube / 14.22 Other Body Shapes / 14.25 Inclined Circular Tubes / 14.25 Inclined UpwardFacing Plates / 14.25 Horizontal UpwardFacing Plates and Disks / 14.26 Bottom of a Container / 14.27 Horizontal and Inclined DownwardFacing Plates and Disks / 14.27 General Axisymmetric Bodies / 14.28 Horizontal and Inclined Elliptical Cylinders / 14.29 Vertically Oriented Helical Coils / 14.29 Condensation with Rotation / 14.30 Zero Gravity / 14.31 InTube Condensation / 14.31 Flow Regimes / 14.31 Vertical Tubes / 14.33 Horizontal Tubes / 14.34 Pressure Losses / 14.38 Condenser Modeling / 14.40 Noncircular Passages / 14.41
14.1
CONTENTS
xiii
Direct Contact Condensation / 14.41 Condensation on Drops (Spray Condensers) / 14.41 Condensation on Jets and Sheets / 14.42 Condensation on Films / 14.43 Condensation on Vapor Bubbles / 14.44 Condensation of Mixtures / 14.45 Equilibrium Methods / 14.46 Nonequilibrium Methods / 14.48 Nomenclature / 14.49 References / 14.54
Chapter 15. Boiling
15.1
Introduction / 15.1 General Considerations / 15.1 Manifestations of Boiling Heat Transfer / 15.2 Stucture of This Chapter / 15.2 Phase Equilibrium / 15.3 SingleComponent Systems / 15.3 Multicomponent Systems / 15.5 Nucleation and Bubble Growth / 15.6 Equilibrium of a Bubble / 15.6 Homogeneous Nucleation / 15.7 Heterogeneous Nucleation / 15.9 Bubble Growth / 15.18 Bubble Release Diameter and Frequency / 15.26 PoolBoiling / 15.30 Pool Boiling Heat Transfer Before the Critical Heat Flux Limit / 15.31 The Critical Heat Flux Limit in Pool Boiling / 15.56 Heat Transfer Beyond the Critical Heat Flux Limit in Pool Boiling / 15.66 Cross Flow Boiling / 15.75 Heat Transfer Below the Critical Heat Flux Limit in Cross Flow Boiling / 15.77 Critical Heat Flux in Cross Flow Boiling / 15.81 Heat Transfer Beyond the Critical Heat Flux Limit in Cross Flow Boiling / 15.83 Forced Convective Boiling in Channels / 15.84 Heat Transfer Below the Critical Heat Flux Limit in Forced Convective Boiling in Channels / 15.89 Critical Heat Flux in Forced Convective Boiling in Channels / 15.112 Heat Transfer Beyond the Critical Heat Flux Limit in Forced Convective Boiling in Channels / 15.132 Thin Film Heat Transfer / 15.137 Evaporating Liquid Films: Laminar Flow / 15.138 Evaporating Liquid Films: Turbulent Flow / 15.140 Evaporating Liquid Films: Multicomponent Mixtures / 15.140 Evaporating Liquid Films with Nucleate Boiling / 15.141 Heat Transfer to a Nonevaporating (Subcooled) Falling Liquid Film / 15.141 Film Breakdown / 15.142 Rewetting of Hot Surfaces / 15.143 Nomenclature / 15.145 References / 15.152
Chapter 16. Measurement of Temperature and Heat Transfer Introduction / 16.1 Temperature Measurement / 16.2 Basic Concepts and Definitions / 16.2 Standards and Temperature Scales / 16.3 Sensors / 16.8 Local Temperature Measurement / 16.51 Calibration of Thermometers and Assurance of Measurements / 16.54
16.1
xiv
CONTENTS
Heat Flux Measurement / 16.58 Basic Principles / 16.58 Methods / 16.59 Thermal Resistance Gauges / 16.60 Measurement by Analogy / 16.64 Introduction / 16.64 Sublimation Technique / 16.65 Electrochemical Technique / 16.66 Acknowledgments / 16.68 Nomenclature / 16.68 List of Abbreviations / 16.71 References / 16.71
17.1
Chapter 17. Heat Exchangers Introduction / 17.1 Classification of Heat Exchangers / 17.2 ShellandTube Exchangers / 17.2 Newer Designs of ShellandTube Exchangers / 17.14 Compact Heat Exchangers / 17.15 Exchanger Heat Transfer and Pressure Drop Analysis / 17.25 Heat Transfer Analysis / 17.27 The eNTU, PNTU, and MTD Methods / 17.30 Fin Efficiency and Extended Surface Efficiency / 17.34 Extensions of the Basic Recuperator Thermal Design Theory / 17.47 eNTUo and AFI Methods for Regenerators / 17.55 SinglePhase Pressure Drop Analysis / 17.62 SinglePhase Surface Basic Heat Transfer and Flow Friction Characteristics Experimental Methods / 17.69 Analytical Solutions / 17. 76 Experimental Correlations / 17.84 Influence of TemperatureDependent Fluid Properties / 17.88 Influence of Superimposed Free Convection / 17.89 TwoPhase Heat Transfer and Pressure Drop Correlations / 17.89 Flow Patterns / 17.89 TwoPhase Pressure Drop Correlations / 17.95 Heat Transfer Correlations for Condensation / 17.97 Heat Transfer Correlations for Boiling / 17.103 Thermal Design for SinglePhase Heat Exchangers / 17.105 Exchanger Design Methodology / 17.105 Extended Surface Heat Exchangers / 17.105 ShellandTube Heat Exchangers / 17.111 Thermal Design for TwoPhase Heat Exchangers / 17.120 Condensers / 17.120 Vaporizers / 17.125 FlowInduced Vibration / 17.127 Tube Vibration / 17.127 Acoustic Vibrations / 17.128 Design Guidelines for Vibration Mitigation / 17.136 Flow Maldistribution / 17.136 GeometryInduced Flow Maldistribution / 17.136 Flow Maldistribution Induced by Operating Conditions / 17.141 Mitigation of Flow Maldistribution / 17.145 Fouling and Corrosion / 17.146 Fouling / 17.147 Corrosion / 17.152 Concluding Remarks / 17.153 Nomenclature / 17.154 References / 17.162
17.66
CONTENTS
Chapter 18. Heat Transfer in Materials Processing Introduction / 18.1 Heat Transfer Fundamentals Relevant to Materials Processing / 18.2 Conduction HeatTransfer / 18.2 Conduction Heat Transfer in BeamIrradiated Materials / 18.2 Conduction Heat Transfer with Thermomechanical Effects I 18.9 SinglePhase Convective Heat Transfer I 18.12 TwoPhase Convective Heat Transfer / 18.26 Radiation Heat Transfer I 18.35 SystemLevel Thermal Phenomena / 18.43 Heating of a Load Inside Industrial Furnaces / 18.43 Quenching / 18.51 Processing of Several Advanced Materials / 18.57 Concluding Remarks / 18.61 Nomenclature / 18.61 References / 18.65
Index follows Chapter 18
xv
18.1
CHAPTER 1
BASIC CONCEPTS OF HEAT TRANSFER Y. I. Cho Drexel University
E. N. Ganic University of Sarajevo
J. P. Hartnett University of Illinois, Chicago
W. M. Rohsenow Massachusetts Institute of Technology
HEAT TRANSFER MECHANISMS Heat is defined as energy transferred by virtue of a temperature difference. It flows from regions of higher temperature to regions of lower temperature. It is customary to refer to different types of heat transfer mechanisms as modes. The basic modes of heat transfer are conduction, radiation, and convection.
Conduction Conduction is the transfer of heat from one part of a body at a higher temperature to another part of the same body at a lower temperature, or from one body at a higher temperature to another body in physical contact with it at a lower temperature. The conduction process takes place at the molecular level and involves the transfer of energy from the more energetic molecules to those with a lower energy level. This can be easily visualized within gases, where we note that the average kinetic energy of molecules in the highertemperature regions is greater than that of those in the lowertemperature regions. The more energetic molecules, being in constant and random motion, periodically collide with molecules of a lower energy level and exchange energy and momentum. In this manner there is a continuous transport of energy from the hightemperature regions to those of lower temperature. In liquids the molecules are more closely spaced than in gases, but the molecular energy exchange process is qualitatively similar to that in gases. In solids that are nonconductors of electricity (dielectrics), heat is conducted by lattice waves caused by atomic motion. In solids that are good 1.1
1.2
CHAPTER ONE
conductors of electricity, this lattice vibration mechanism is only a small contribution to the energy transfer process, the principal contribution being that due to the motion of free electrons, which move in a similar way to molecules in a gas. At the macroscopic level the heat flux (i.e., the heat transfer rate per unit area normal to the direction of heat flow) q" is proportional to the temperature gradient: q"=k
dT dx
(1.1)
where the proportionality constant k is a transport property known as the thermal conductivity and is a characteristic of the material. The minus sign is a consequence of the fact that heat is transferred in the direction of decreasing temperature. Equation 1.1 is the onedimensional form of Fourier's law of heat conduction. Recognizing that the heat flux is a vector quantity, we can write a more general statement of Fourier's law (i.e., the conduction rate equation) as q" =  k VT
(1.2)
where V is the threedimensional del operator and T is the scalar temperature field. From Eq. 1.2 it is seen that the heat flux vector q" actually represents a current of heat (thermal energy) that flows in the direction of the steepest temperature gradient. If we consider a onedimensional heat flow along the x direction in the plane wall shown in Fig. 1.1a, direct application of Eq. 1.1 can be made, and then integration yields kA
q=~
(T2  T1)
(1.3)
where the thermal conductivity is considered constant, Ax is the wall thickness, and T1 and T2 are the wallface temperatures. Note that q/A = q", where q is the heat transfer rate through an area A. Equation 1.3 can be written in the form 7"2 Ta q  Ax/kA
T 2  T1 
Rth

thermal potential difference thermal resistance
(1.4)
where zLv,/kA assumes the role of a thermal resistance Rth. T h e relation of Eq. 1.4 is quite like Ohm's law in electric circuit theory. The equivalent electric circuit for this case is shown in Fig. 1.1b. The electrical analogy may be used to solve more complex problems involving both series and parallel resistances. Typical problems and their analogous electric circuits are given in many heat transfer textbooks [14]. In treating conduction problems it is often convenient to introduce another property that is related to the thermal A conductivity, namely, the thermal diffusivity (x,
FTemperature profile
q
(x 
TI ~
~x (a)
T2
q
T~
Ax kA
(b)
FIGURE 1.1 Onedimensional heat conduction through a plane wall (a) and electric analog (b).
k pc
(1.5)
where p is the density and cv is the specific heat at constant pressure. As mentioned above, heat transfer will occur whenever there exists a temperature difference in a medium. Similarly, whenever there exists a difference in the concentration or density of some chemical species in a mixture, mass transfer must occur. Hence, just as a temperature gradient constitutes the driving potential for heat transfer, the existence of a concentration gradient for some species in a mixture provides the driving potential for transport of that species. Therefore,
BASIC CONCEPTS OF HEAT TRANSFER
1.3
the term mass transfer describes the relative motion of species in a mixture due to the presence of concentration gradients. Since the same physical mechanism is associated with heat transfer by conduction (i.e., heat diffusion) and mass transfer by diffusion, the corresponding rate equations are of the same form. The rate equation for mass diffusion is known as Fick's law, and for a transfer of species 1 in a binary mixture it may be expressed as dC1 jl " D ~
(1.6)
where C1 is a mass concentration of species 1 in units of mass per unit volume. This expression is analogous to Fourier's law (Eq. 1.1). Moreover, just as Fourier's law serves to define one important transport property, the thermal conductivity, Fick's law defines a second important transport property, namely the binary diffusion coefficient or mass diffusivity D. The quantity jl [mass/(time x surface area)] is defined as the mass flux of species 1, i.e., the amount of species 1 that is transferred per unit time and per unit area perpendicular to the direction of transfer. In vector form Fick's law is given as jl =  D V C 1
(1.7)
In general, the diffusion coefficient D for gases at low pressure is almost composition independent; it increases with temperature and varies inversely with pressure. Diffusion coefficients are markedly concentration dependent and generally increase with temperature.
Radiation Radiation, or more correctly thermal radiation, is electromagnetic radiation emitted by a body by virtue of its temperature and at the expense of its internal energy. Thus thermal radiation is of the same nature as visible light, x rays, and radio waves, the difference between them being in their wavelengths and the source of generation. The eye is sensitive to electromagnetic radiation in the region from 0.39 to 0.78 ~tm; this is identified as the visible region of the spectrum. Radio waves have a wavelength of 1 x 10 3 to 2 x 101° ~tm, and x rays have wavelengths of 1 × 105 to 2 x 102 ktm, while the bulk of thermal radiation occurs in rays from approximately 0.1 to l00 ktm. All heated solids and liquids, as well as some gases, emit thermal radiation. The transfer of energy by conduction requires the presence of a material medium, while radiation does not. In fact, radiation transfer occurs most efficiently in a vacuum. On the macroscopic level, the calculation of thermal radiation is based on the StefanB o l t z m a n n law, which relates the energy flux emitted by an ideal radiator (or blackbody) to the fourth power of the absolute temperature: eb = t~T 4
(1.8)
Here ~ is the StefanBoltzmann constant, with a value of 5.669 × 108 W/(m2.K4), or 1.714 x 10 9 Btu/(h.ft 2"°R4). Engineering surfaces in general do not perform as ideal radiators, and for real surfaces the above law is modified to read e = et~T 4
(1.9)
The term e is called the emissivity of the surface and has a value between 0 and 1. When two blackbodies exchange heat by radiation, the net heat exchange is then proportional to the difference in T 4. If the first body "sees" only body 2, then the net heat exchange from body 1 to body 2 is given by q = aAI(T~  T~)
(1.10)
1.4
CHAFFERONE When, because of the geometric arrangement, only a fraction of the energy leaving body 1 is intercepted by body 2,
q = ~A1F~_2(T 4  T 4)
(1.11)
where FI_ 2 (usually called a shape factor or a view factor) is the fraction of energy leaving body 1 that is intercepted by body 2. If the bodies are not black, then the view factor F~_ 2 must be replaced by a new factor ~1 2 which depends on the emissivity ~ of the surfaces involved as well as the geometric view. Finally, if the bodies are separated by gases or liquids that impede the radiation of heat through them, a formulation of the heat exchange process becomes more involved (see Chap. 7).
Convection Convection, sometimes identified as a separate mode of heat transfer, relates to the transfer of heat from a bounding surface to a fluid in motion, or to the heat transfer across a flow plane within the interior of the flowing fluid. If the fluid motion is induced by a pump, a blower, a fan, or some similar device, the process is called forced convection. If the fluid motion occurs as a result of the density difference produced by the temperature difference, the process is called free or natural convection. Detailed inspection of the heat transfer process in these cases reveals that, although the bulk motion of the fluid gives rise to heat transfer, the basic heat transfer mechanism is conduction, i.e., the energy transfer is in the form of heat transfer by conduction within the moving fluid. More specifically, it is not heat that is being convected but internal energy. However, there are convection processes for which there is, in addition, latent heat exchange. This latent heat exchange is generally associated with a phase change between the liquid and vapor states of the fluid. Two special cases are boiling and condensation.
Heat Transfer Coefficient. In convective processes involving heat transfer from a boundary surface exposed to a relatively lowvelocity fluid stream, it is convenient to introduce a heat transfer coefficient h, defined by Eq. 1.12, which is known as Newton's law ofcooling: q"= h ( T ~  Tf)
Fluid flow
(1.12)
Here T~ is the surface temperature and Tf is a characteristic fluid temperature. For surfaces in unbounded convection, such as plates, tubes, bodies of revolution, etc., immersed in a large body of fluid, it is customary to define h in Eq. (1.12) with Tr as the temperature of the fluid far away from the surface, often identified as T~ (Fig. 1.2). For bounded convection, including such cases as fluids flowing in tubes or channels, across tubes in bundles, etc., Tyis usually taken as the enthalpymixedmean temperature, customarily identified as Tin. The heat transfer coefficient defined by Eq. 1.12 is sensitive to the geometry, to the physical properties of the fluid, and to the fluid velocity. However, there are some special situations in which h can depend on the temperature difference AT Tw  TI. For example, if the surface is hot enough to boil a liquid surrounding it, h will typically vary as ATE; or in the case of natural convection, h varies as some weak power of A T B typically as AT TM or AT 1/3.It is important to note that Eq. 1.12 as a definition of h is valid in these cases too, although its usefulness may well be reduced. As q "  q/A, from Eq. 1.12 the thermal resistance in convection heat transfer is given by
o,I,o
~T~ F I G U R E 1.2 Velocity and temperature distributions in flow over a flat plate.
1 Rth
hA
which is actually the resistance at a surfacetofluid interface.
BASICCONCEPTSOF HEATTRANSFER
1.5
At the wall, the fluid velocity is zero, and the heat transfer takes place by conduction. Therefore, we may apply Fourier's law to the fluid at y = 0 (where y is the axis normal to the flow direction, Fig. 1.2):
q " =  k ~9~YTly=0
(1.13)
where k is the thermal conductivity of fluid. By combining this equation with Newton's law of cooling (Eq. 1.12), we then obtain h

q" Tw T:
_
k(OT/Oy)ly=0 rw T:
_
(1.14)
so that we need to find the temperature gradient at the wall in order to evaluate the heat transfer coefficient. Similar results may be obtained for convective mass transfer If a fluid of species concentration C1= flows over a surface at which the species concentration is maintained at some value Cl.w ~ C1,~, transfer of the species by convection will occur. Species 1 is typically a vapor that is transferred into a gas stream by evaporation or sublimation at a liquid or solid surface, and we are interested in determining the rate at which this transfer occurs. As for the case of heat transfer, such a calculation may be based on the use of a convection coefficient [3, 5]. In particular we may relate the mass flux of species 1 to the product of a transfer coefficient and a concentration difference
J1 = hD(Cl,w

Cl,oo)
(1.15)
Here hD is the convection mass transfer coefficient and it has a dimension of Lit. At the wall, y = 0, the fluid velocity is zero, and species transfer is due only to diffusion; hence jlD
OC1 I
~y
y=0
(1.16)
Combining Eqs. 1.17 and 1.18, it follows that
hD =  D(OC,/Oy)ly=o
(1.17)
C1, w  Cl,oo
Therefore conditions that influence the surface concentration gradient (~Cl/OY)ly=Owill also influence the convection mass transfer coefficient and the rate of species transfer across the fluid layer near the wall. For convective processes involving highvelocity gas flows (high subsonic or supersonic flows), a more meaningful and useful definition of the heat transfer coefficient is given by
q"= h(Tw  Taw)
(1.18)
Here Taw,commonly called the adiabatic wall temperature or the recovery temperature, is the equilibrium temperature the surface would attain in the absence of any heat transfer to or from the surface and in the absence of radiation exchange between the surroundings and the surface. In general the adiabatic wall temperature is dependent on the fluid properties and the properties of the bounding wall. Generally, the adiabatic wall temperature is reported in terms of a dimensionless recovery factor r defined as V2 Taw = Tf+ r 2Cp
(1.19)
The value of r for gases normally lies between 0.8 and 1.0. It can be seen that for lowvelocity flows the recovery temperature is equal to the freestream temperature TI. In this case,
] .6
CHAPTER ONE
Eq. 1.15 reduces to Eq. 1.12. From this point of view, Eq. 1.18 can be taken as the generalized definition of the heat transfer coefficient.
Boundary Layer Concept.
The transfer of heat between a solid body and a liquid or gas flow is a problem whose consideration involves the science of fluid motion. On the physical motion of the fluid there is superimposed a flow of heat, and the two fields interact. In order to determine the temperature distribution and then the heat transfer coefficient (Eq. 1.14) it is necessary to combine the equations of motion with the energy conservation equation. However, a complete solution for the flow of a viscous fluid about a body poses considerable mathematical difficulty for all but the most simple flow geometries. A great practical breakthrough was made when Prandtl discovered that for most applications the influence of viscosity is confined to an extremely thin region very close to the body and that the remainder of the flow field could to a good approximation be treated as inviscid, i.e., could be calculated by the method of potential flow theory. The thin region near the body surface, which is known as the boundary layer, lends itself to relatively simple analysis Potential by the very fact of its thinness relative to the dimensions of flow , ." " region 1 1 1 the body. A fundamental assumption of the boundary layer u~l~ ....~~ approximation is that the fluid immediately adjacent to the ,, ~ bo~r~ ~ y ~ , body surface is at rest relative to the body, an assumption "~ , , , " v"j~lu req~n~n,r[fffflllll//'" that appears to be valid except for very lowpressure gases, when the mean free path of the gas molecules is large rela,y////I///" tive to the body [6]. Thus the hydrodynamic or velocity boundary layer 5 may be defined as the region in which the fluid velocity changes from its freestream, or potential flow, value to zero at the body surface (Fig. 1.3). In reality there is F I G U R E 1.3 Boundary layer flow past an extemal no precise "thickness" to a boundary layer defined in this surface. manner, since the velocity asymptotically approaches the freestream value. In practice we simply imply that the boundary layer thickness is the distance in which most of the velocity change takes place. The viscous forces within the boundary layer region are described in terms of the shear stress x between the fluid layers. If this stress is assumed to be proportional to the normal velocity gradient, we have the defining equation for viscosity du
T.=~t dy
(1.20)
The constant of proportionality la is called the dynamic viscosity (Pas), and Eq. 1.20 is sometimes referred to as Newton's law of shear [7] for a simple flow in which only the velocity component u exists. The ratio of the viscosity l.t to the density p is known as the kinematic viscosity (m2/s) and is defined as v
~t (1.21) P Flow inside a tube is a form of boundary layer problem in which, near the tube entrance, the boundary layer grows in much the same manner as over an external surface until its growth is stopped by symmetry at the centerline of the tube (Fig. 1.4). Thus the tube radius becomes the ultimate boundary layer thickness. When there is heat transfer or mass transfer between the fluid and the surface, it is also found that in most practical applications the major temperature and concentration changes occur in a region very close to the surface. This gives rise to the concept of the thermal boundary layer ~)rand the concentration boundary layer ?h~.The influence of thermal conductivity k and mass diffusivity D is confined within these regions. Outside the boundary layer region the flow is essentially nonconducting and nondiffusing. The thermal (or concentration) boundary layer may be smaller than, larger than, or the same size as the velocity boundary layer. The development of the thermal boundary layer in the entrance region of a tube is shown in Fig. 1.5.
BASIC CONCEPTS OF HEAT TRANSFER
1.7
layer inlet flow
~
>
  ...._
>
~ 
8
   . . . . . ~ . ~ , .
~
~
~," L. F" F I G U R E 1.4
flow
~r~d" r.3I
Entrance length
Velocity profile for laminar flow in a tube.
Inletflow at uniform temperature
/    Thermal boundary layer
~ T.~
_
•
/ I 
'
•
/

r
i
.,/
Tm
~T(r,x) I_ F"
Entrance length
.3 r1
F I G U R E 1.5 The development of temperature profile in the entrance region of a tube.
It is important to notice the similarity between Eqs. 1.1, 1.6, and 1.20. The heat conduction equation, Eq. 1.1, describes the transport of energy; the diffusion law, Eq. 1.6, describes the transport of mass; and the viscous shear equation, Eq. 1.20, describes the transport of momentum across fluid layers. We note also that the kinematic viscosity v, the thermal diffusivity o~, and the diffusion coefficient D all have the same dimensions L2/t. As shown in Table 1.10, a dimensionless number can be formed from the ratio of any two of these quantities, which will give relative speeds at which momentum, energy, and mass diffuse through the medium. Laminar and Turbulent Flows. There are basically two different types of fluid motion, identified as laminar and turbulent flow. In previous sections we referred basically to laminar flow. In the case of flow over a flat plate (Fig. 1.6), the flow near the leading edge is smooth and streamlined. Locally within the boundary layer the velocity is constant and invariant with time. The boundary layer thickness grows with increasing distance from the leading edge, and at some critical distance the inertial effects become sufficiently large compared to the viscous damping action that small disturbances in the flow begin to grow. As these disturbances
Laminar "~Transition ~~
Turbulent
I F I G U R E 1.6 Laminar, transition, and turbulent boundary layer flow regimes in flow over a flat plate.
1.8
CHAPTER ONE
become amplified, the regularity of the viscous flow is disturbed and a transition from laminar to turbulent flow takes place. (However, there still must be a very thin laminar sub,,~mV,vrwnaW T layer next to the wall, at least for a smooth plate.) These disturbances may originate from the free stream or may be induced by surface roughness. ~In the turbulent flow region a very efficient mixing takes t place, i.e., macroscopic chunks of fluid move across streamFIGURE 1.7 Property variation with time at some lines and transport energy and mass as well as momentum point in a turbulent boundary layer. vigorously. The most essential feature of a turbulent flow is the fact that at a given point in it, the flow property X (e.g., velocity component, pressure, temperature, or a species concentration) is not constant with time but exhibits very irregular, highfrequency fl__uctuations (Fig. 1.7). At any instant, X may be represented as the sum of a timemean value X and a fluctuating component X'. The average is taken over a time that is large compared with the period of typical fluctuation, and if X is independent of time, the timemean flow is said to be steady. The existence of turbulent flow can be advantageous in the sense of providing increased heat and mass transfer rates. However, the motion is extremely complicated and difficult to describe theoretically [3, 8]. In dealing with turbulent flow it is customary to speak of a total shear stress and total fluxes normal to the main flow direction (the main flow is in the x direction, and the y axis is normal to the flow direction), which are defined as a~
"r.,= la ~y  pU"v"
(1.22)
aT  pCpv'T') q;'= (k ~y
(1.23)
jl.,=(D ~OC' ~C~)
(1.24)
where the first term on the right side of Eqs. 1.221.24 is the contribution due to molecular diffusion and the second term is the contribution due to turbulent mixing. For example, u'v' is the time average of the product of u' and v'. A simple conceptual model for turbulent flow deals with eddies, small portions of fluid in the boundary layer that move about for a short time before losing their identity [8]. The transport coefficient, which is defined as eddy diffusivity for momentum transfer ~M, has the form 8~ M bYy= u'v'
(1.25)
Similarly, eddy diffusivities for heat and mass transfer, ~n and ~m, respectively, may be defined by the relations bT
eI4 ~ =v'T" bC1
Em  ~ y
" I) t f t l
(1.26)
(1.27)
Hence the total shear stress and total fluxes may be expressed, with the help of the relations of Eqs. 1.5 and 1.21, as b~ x, : p(v + ,M) by
(1.28)
BASIC CONCEPTS OF HEAT TRANSFER
1.9
n
3T
qt = pcp(ct + if,) by
(1.29)
3C1
jl,t = (]D + ifm) ~y
(1.30)
In the region of a turbulent boundary layer far from the surface (the core region), the eddy diffusivities are much larger than the molecular diffusivities. The enhanced mixing associated with this condition has the effect of making velocity, temperature, and concentration profiles more uniform in the core. This behavior is shown in Fig. 1.8, which gives the measured velocity distributions for laminar and turbulent flow where the mass flow is the same in both cases [7]. It is evident from Fig. 1.8 that the velocity gradient at the surface, and therefore the surface shear stress, is much larger for turbulent flow than for laminar flow. Following the sameargument, the tempera(a) (b) ture or concentration gradient at the surface, and therefore F I G U R E 1.8 Velocity distribution in a tube: (a) lamthe heat and mass transfer rates, are much larger for turbuinar; (b) turbulent. lent than for laminar flow. When the flow in the tube is turbulent, the mean velocity is about 83 percent of the center velocity. For laminar flow, the profile has a parabolic shape and the mean velocity is onehalf the value at the center. A fundamental problem in performing a turbulent flow analysis involves determining the eddy diffusivities as a function of the mean properties of the flow. Unlike the molecular diffusivities, which are strictly fluid properties, the eddy diffusivities depend strongly on the nature of the flow; they can vary from point to point in a boundary layer, and the specific variation can be determined only from experimental data. For flow in circular tubes, the numerical value of the Reynolds number (defined in Table 1.10), based on mean velocity at which transition from laminar to turbulent flow occurs, was established as being approximately 2300, i.e.,
Reef=( VmD
cr
=2300
(1.31)
There exists, however, as demonstrated by numerous experiments [7], a lower value for Recr that is approximately at 2000. Below this value the flow remains laminar even in the presence of very strong disturbances. If the Reynolds number is greater than 10,000, the flow is considered to be fully turbulent. In the 2300 to 10,000 region, the flow is often described as transition flow. It is possible to shift these values by minimizing the disturbances in the inlet flow, but for general engineering application the numbers cited are representative. For a flow over a flat plate, the transition to turbulent flow takes place at distance x, measured from the leading edge, as determined by Recr : aP
:
SX 10Sto 106
(1.32)
aP
Ox < OIj ~  > 0 I u~(x) I I
Flow reversal J
"~
" Vortices
F I G U R E 1.9 Velocity profile associated with separation on a circular cylinder in cross flow.
but the values are dependent on the level of turbulence in the main stream. Here V= is the freestream velocity. A particularly interesting phenomenon connected with transition in the boundary layer occurs with blunt bodies, e.g., spheres or circular cylinders. In the region of adverse pressure gradient (i.e., 3P/bx > 0 in Fig. 1.9) the boundary layer separates from the surface. At this location the shear stress goes to zero, and beyond this point there is a reversal of flow in the vicinity of the wall, as shown in Fig. 1.9. In this
1.10
CHAPTER ONE separation region, the analysis of the flow is very difficult and emphasis is placed on the use of experimental methods to determine heat and mass transfer.
Nonnewtonian Fluids. In previous parts of this section we have mentioned only newtonian fluids. Newtonian fluids are those that have a linear relationship between the shear stress and the velocity gradient (or rate of strain), as in Eq. 1.20. The shear stress x is equal to zero when du/dy equals zero. The viscosity, given by the ratio of shear stress to velocity gradient, is independent of the velocity gradient (or rate of strain), but may be a function of temperature and pressure. Gases, and liquids such as water, usually exhibit newtonian behavior. However, many fluids, such as colloidal suspensions, polymeric solutions, paint, grease, blood, ketchup, slurry, etc., do not follow the linear shear stressvelocity gradient relation; these are called nonnewtonian fluids. Chapter 10 deals with the hydrodynamics and heat transfer of nonnewtonian fluids.
Combined Heat Transfer Mechanisms In practice, heat transfer frequently occurs by two mechanisms in parallel. A typical example is shown in Fig. 1.10. In this case the heat conducted through the plate is removed from the plate surface by a combination of convection and radiation. An energy balance in this case gives ~ksA
d~,. J = hZ(Tw T,o) + oA~ (T 4 T4a) uYl
(1.33)
W
where Ta is the temperature of the surroundings, ks is the thermal conductivity of the solid plate, and ~ is the emissivity of the plate (i.e., in this special case ~12 = ~, as the area of the plate is much smaller than the area of the surroundings [3]). The plate and the surroundings are separated by a gas that has no effect on radiation. There are many applications where radiation is combined with other modes of heat transfer, and the solution of such problems can often be simplified by using a thermal resistance Rth for radiation. The definition of Rth is similar to that of the thermal resistance for convection and conduction. If the heat transfer by radiation, for the example in Fig. 1.10, is written
TwT~ q=
gth
(1.34)
j F,ow, u _ L
//j
Yt
A
/4 qco~ hA(TwToo)
/S
T eat conducted through wail FIGURE 1.10 Combination of conduction, convection, and radiation heat transfer.
BASIC CONCEPTS OF HEAT TRANSFER
1.11
the resistance is given by Rth
(1.35)
T w  Ta o A e ( T 4 _ T 4)
Also, a heat transfer coefficient hr c a n be defined for radiation:
hr
1
 ° e ( T 4  T4) : oe(Tw + Ta)(T 2 + TZa)
Rthm
(1.36)
T w  Ta
Here we have linearized the radiation rate equation, making the heat rate proportional to a temperature difference rather than to the difference between two temperatures to the fourth power. Note that hr depends strongly on temperature, while the temperature dependence of the convection heat transfer coefficient h is generally weak.
CONSERVATION EQUATIONS Each time we try to solve a new problem related to momentum, heat, and mass transfer in a fluid, it is convenient to start with a set of equations based on basic laws of conservation for physical systems. These equations include: 1. 2. 3. 4.
The The The The
continuity equation (conservation of mass) equation of motion (conservation of momentum) energy equation (conservation of energy, or the first law of thermodynamics) conservation equation for species (conservation of species)
These equations are sometimes called the equations o f change, inasmuch as they describe the change of velocity, temperature, and concentration with respect to time and position in the system. The first three equations are sufficient for problems involving a pure fluid (a pure substance is a single substance characterized by an unvarying chemical structure). The fourth equation is added for a mixture of chemical species, i.e., when mass diffusion with or without chemical reactions is present. • The control volume. When deriving the conservation equations it is necessary to select a control volume. The derivation can be performed for a volume element of any shape in a given coordinate system, although the most convenient shape is usually assumed for simplicity (e.g., a rectangular shape in a rectangular coordinate system). For illustration purposes, different coordinate systems are shown in Fig. 1.11. In selecting a control volume we
(x, y, z)
(r,
9 I I I
Iz
Iz
I I
I
, ~i . x
.
.
.
Y
(a) FIGURE 1.11 (c) spherical.
. I x.
=y .
~ .
(r,O, ¢,)
z)
1 I i =Y
°'~ ~ /
/
/'
I
ll 1I
~ ._~_r,Jf/
x,~"~¢"''"~'
x
(b)
"y
(c)
Coordinate systems: (a) rectangular, (b) cylindrical,
1.12
CHAPTERONE have the option of using a volume fixed in space, in which case the fluid flows through the boundaries, or a volume containing a fixed mass of fluid and moving with the fluid. The former is known as the eulerian viewpoint and the latter is the lagrangian viewpoint. Both approaches yield equivalent results. • The partial time derivative OB/Ot. T h e partial time derivative of B(x, y, z, t), where B is any continuum property (e.g., density, velocity, temperature, concentration, etc.), represents the change of B with time at a fixed position in space. In other words, 3B/Ot is the change of B with t as seen by a stationary observer. • Total time derivative dB/dt. T h e total time derivative is related to the partial time derivative as follows: dB
OB
tit
dx OB
dy OB
+ ¥ ffx + ¥
dz OB
(1.37)
+ d;
where dx/dt, dy/dt, and dz/dt are the components of the velocity of a moving observer. Therefore, dB/dt is the change of B with time as seen by the moving observer. • Substantial time derivative DB/Dt. This derivative is a special kind of total time derivative where the velocity of the observer is just the same as the velocity of the stream, i.e., the observer drifts along with the current: DB Dt

i)B ~)B i)B ~)B ~7 + u ~x + v oyX+ w ~ z
(1.38)
where u, v, and w are the components of the local fluid velocity V. The substantial time derivative is also called the derivative following the motion. T h e sum of the last three terms on the right side of Eq. 1.38 is called the convective contribution because it represents the change in B due to translation. The use of the operator D / D t is always made when rearranging various conservation equations related to the volume element fixed in space to an element following the fluid motion. The operator D / D t may also be expressed in vector form: D Dt
   + ( V . V) 3t
(1.39)
Mathematical operations involving V are given in many textbooks. Applications of V in various operations involving the conservation equations are given in Refs. 6 and 10. Table 1.1 gives the expressions for D / D t in different coordinate systems. TABLE 1.1
Substantial Derivative in Different Coordinate Systems
Rectangular coordinates (x, y, z): D /) /)
/)
/)
Dt  i)t + u~x + V~y + w ~)z
Cylindrical coordinates (r, 0, z): D
/)
/)
v0 /)
Dt  i)t + v, ~r + r  ~ + Vz Oz
Spherical coordinates (r, 0, ~)" D /) /)
v0 /)
v,
/)
Dt  ~)t + v, ~r + ~r  ~ + rsin0 /){~
BASIC CONCEPTS OF HEAT TRANSFER
1.13
The Equation of ContinuiW For a volume element fixed in space,
3p _  ( V  p V ) 3t net rate of mass effiux per unit volume
(1.40)
The continuity equation in this form describes the rate of change of density at a fixed point in the fluid. By performing the indicated differentiation on the right side of Eq. 1.40 and collecting all derivatives of p on the left side, we obtain an equivalent form of the equation of continuity: Dp Dt p(V.
V)
(1.41)
The continuity equation in this form describes the rate of change of density as seen by an observer "floating along" with the fluid. For a fluid of constant density (incompressible fluid), the equation of continuity becomes: V V = 0
(1.42)
Table 1.2 gives the equation of continuity in different coordinate systems.
TABLE 1.2 Equation of Continuity in Different Coordinate Systems Rectangular coordinates (x, y, z): c3p C3 C3
C3
aS+ ~ (p") + b;y(pv)+ ~ (pw): o
Cylindrical coordinates (r, 0, z): c3p 1 C3 1,9 C3 c3t + r~r (prvr) +r  ~ (pv0) + ~z (pVz) = 0 Spherical coordinates (r, 0, ¢): 1 C3 c3p 1 C3 1 C3 (pv,) =0 C3t + ~ ~r (pr2vr) + ~ sin r 0 C30(pve sin 0) + ~r sin 0 m c3(1)
Incompressible flow Rectangular coordinates (x, y, z): c3u c3v
c3w
a x + ~ y + ~ =° Cylindrical coordinates (r, O, z): 1 C3 1 c3ve r c3r (rVr) +   r   ~
c3Vz =0
Spherical coordinates (r, 0, ~): 1 C3 1 C3 1 c3v, r 2 c3r (r2vr) + r sin 0 C30 (v0 sin 0) + r sin 0 C3~  0
1.14
CHAPTERONE
The Equation of Motion (Momentum Equation) The momentum equation for a stationary volume element (i.e., a balance over a volume element fixed in space) with gravity as the only body force is given by 3pV Ot
=  ( V . pV)V 
VP
rate of increase of momentum per unit volume
rate of momentum gain by convection per unit volume
pressure force on element per unit volume
+
V. x
+
rate of momentum gain by viscous transfer per unit volume
pg
(1.43)
gravitational force on element per unit volume
Equation 1.43 may be rearranged, with the help of the equation of continuity, to give DV
p   ~ =  V P + V . x + pg
(1.44)
The last equation is a statement of Newton's second law of motion in the form mass x acceleration = s u m o f forces.
These two forms of the equation of motion (Eqs. 1.43 and 1.44), correspond to the two forms of the equation of continuity (Eqs. 1.40 and 1.41). As indicated, the only body force included in Eqs. 1.43 and 1.44 is gravity. In general, electromagnetic forces may also act on a fluid. The scalar components of Eq. 1.44 are listed in Table 1.3 and the components of the stress tensor x are given in Table 1.4. For the flow of a newtonian fluid with varying density but constant viscosity/.t, Eq. 1.44 becomes DV 1 p  ~ =  V P + ff ktV(V • V) +/.tV2V + pg
(1.45)
If p and ILt are constant, Eq. 1.44 may be simplified by means of the equation of continuity (V • V = 0) for a newtonian fluid to give DV p~ =  V P + ~.I,V2'V ] pg
(1.46)
This is the famous NavierStokes equation in vector form. The scalar components of Eq. 1.46 are given in Table 1.5. For V • x = 0, Eq. 1.44 reduces to Euler's equation: DV p   ~ =  V P + pg
(1.47)
which is applicable for describing flow systems in which viscous effects are relatively unimportant. As mentioned before, there is a subset of flow problems, called natural convection, where the flow pattern is due to buoyant forces caused by temperature differences. Such buoyant forces are proportional to the coefficient of thermal expansion 13, defined as: 13=p
~
,
(1.48)
where T is absolute temperature. Using an approximation that applies to low fluid velocities and small temperature variations, it can be shown [911] that V P  pg = p[Sg(T Too)
(1.49)
BASIC CONCEPTS OF HEAT TRANSFER
1.15
Then Eq. 1.44 becomes DV p   ~ = V . x  p~g(T Too)
(1.50)
buoyant force on element per unit volume
The above equation of motion is used for setting up problems in natural convection when the ambient temperature T= may be defined.
The Energy Equation For a stationary volume element through which a pure fluid is flowing, the energy equation reads 3
~)"~"p ( u + 1//2V2) =  V rate of gain of energy per unit volume
p V ( u + 1/2V2) 
rate of energy input per unit volume by convection
V"
q"
+
rate of energy input per unit volume by conduction
V. PV
+
rate of work done on fluid per unit volume by pressure forces
p(V • g) rate of work done on fluid per unit volume by gravitational forces
V. (x. V) rate of work done on fluid per unit volume by viscous forces
+
q"
(1.51)
rate of heat generation per unit volume ("source term")
where u is the internal energy. The left side of this equation, which represents the rate of accumulation of internal and kinetic energy, does not include the potential energy of the fluid, since this form of energy is included in the work term on the right side. Equation 1.51 may be rearranged, with the aid of the equations of continuity and motion, to give [10, 191 Du
p ~
=  V • q '  P ( V . V) + VV:'I: + q'"
(1.52)
A summary of VV:'~ in different coordinate systems is given in Table 1.6. For a newtonian fluid, VV:x = BO
(1.53)
and values of dissipation function • in different coordinate systems are given in Table 1.7. Components of the heat flux vector q "   k V T are given in Table 1.8 for different coordinate systems. Often it is more convenient to work with enthalpy rather than internal energy. Using the definition of enthalpy, i  u + P/p, and the mass conservation equation, Eq. 1.41, Eq. 1.52 can be rearranged to give Di DP p  ~ = V . k V T +   ~ + . ~ + q"
(1.54)
1.16
CHAPTER ONE TABLE 1.3
Equation of Motion in Terms of Viscous Stresses (Eq. 1.44)* Rectangular coordinates (x, y, z)
x direction
(xx
=  ~ + ffx+~y+ az ]+pgx
p ¥+.~+v~+w~ y direction
(3V 3V 312 312) 3P ~3"[,xy 3T,yy 3"r,zy I p ~..tu~xqV~ytw~Z :    ~ y  b \ 3x q~yb 3Z / qrDgy z direction
P ~+~ffx +~ffy +WTz =Tz + ~ x +~y + az ]+pgz Cylindrical coordinates (r, 0, z) r direction
3Vr 3V r 1)0 3V r p ~+Vr~r + r 30
])2 r + Vz ~Z
=  ~ r +
~
1 3'l;0r "g00 3Zzr1 (rT'rr) + r 30 _ mr + 3ZJ + Pgr
0 direction / 3Vo
3V 0
V0 3V 0
VrVO
P ~  ~ + V r  ~ r +  r   ~ + ~ r
+ Vz ~Z =   r  ~ +
~r (r21:rO)+r   ~
+ 3Z ] + Pgo
Z direction
[3Vz 3vz vo 3Vz 3v~ 3P [ 1 3 13Xoz 3Xzz] P l~ "~"Vr ~r t r  ~ + v z "3 Z ] =  "~Z t Lr ~r ( r T,rz) +   r   ~ + 3 Z J + Pg z Spherical coordinates (r, 0, ~) r direction V, 3V r 3Vr 3V r VO 3V r P  ~ + V r   ~ r +  r  ~  +  r sin 0 30
+
VO 2 + V~ ~
3P
J
3r
r
[~_ 3 ~r (r2~rr) +
1
3
r s i n 0 30
(T,Orsin 0) +
1
(~l:~r
rsin0
30
Zoo +rZ~ ] + Pgr
0 direction / 3V 0 3V 0 ]20 3V 0 p ~ '  ~ "~ I/r ~r ~  I
I]~
av 0
VrV0
r frO r sin 0 3~  I   r +
5; (r~/+
~r
v~ cot 0)r
=  l_r 3P30
1 3 1 3%0 Xr0 X. cot0  (x00 sin 0) + r sin 0 3¢ + r  ~ r sin 0 30
+ Pgo
# direction
{ 3V¢~
3V,
V0 3V¢~
=
V, 3V¢~ r sin 0 3~
VC~Vr r
vov, cot 0 r
)
1 3P [ 1 3 1 31:o~ 1 3x. Xr# 2X0~cot0] r sin 0 3~ + [ 7 ~ r (r2"l:r~)+   r  ~ + r sin  0 3~ + r + r + pg*
* Components of the stress tensor (x) for newtonian fluids are given in Table 1.4. This equation may also be used for describing nonnewtonian flow. However, we need relations between the components of x and the various velocity gradients; in other words, we have to replace the expressions given in Table 1.4 with other relations appropriate for the nonnewtonian fluid of interest. The expressions for x for some nonnewtonian fluid models are given in Ref. 10. See also Chap. 10.
BASIC CONCEPTS OF HEAT TRANSFER TABLE 1.4
1.17
Components of the Stress Tensor x for Newtonian Fluids* Rectangular coordinates (x, y, z)
bu
2
3v
2
Xxx=. 2 ffXX~ (V • V)
]
Xzz=. 2~z  ~ (V • V) bu
3v
T,xy " T,yx " ~J,
Jr "~X
"Cyz = "Czy= ~t
+
"Czx= "Gz= kt
+
bw
~v. v):Gx +c:oy +
bz
Cylindrical coordinates (r, 0, z) I
~V r
]
2
[(l vo
~  g 2 7~+ Xzz = g/2
~(v.v)
]
2
OVz
bTz~ (v. v)]
~r~=~Or=~t r ~
1
+
ro,,o
r~ 10vz 1
r~z
~,]
"~zr T'rz = ~'l'[~T "~ 3Z J
~
(V.V)=I
1 ~vo
r ~r ( r v r ) +  r  ~
3Vz + ~z
Spherical coordinates (r, 0, ~) 3Vr
2
]
T,rr=~l, 2 "~F '~ (V • V )
~g 2 7~+ [ (
x~=~t 2
1
~(v.v) 3v,
r sin 0 /)~
[ ~ (~)
"Or0 "l~0r ~LI,rfirr
vr
VoCOtO) 2
+ ~ + ~ r
(V'V)
]
l Ovr1
+  r   ~
rsino~( v° ) x0~ = %0 = l a [ ~

r
~
1
~vo]
+ r sin 0 /)~
1 ()Vr ~ (~)1 %r='r'r~=l't[ rsinO 3~ + r  ~ r
1 ~
(V. V) = ~ ~ r
1
~
1
~v,
(r2vr) + rsinO 30 (vo sin O) + rsinO ~
* It should be noted that the sign convention adopted here for components of the stress tensor is consistent with that found in many fluid mechanics and heat transfer books; however, it is opposite to that found in some books on transport phenomena, e.g., Refs. 10, 11, and 14.
1.18
CHAPTER ONE
TABLE 1.5 (Eq. 1.46)
Equation of Motion in Terms of Velocity Gradients for a Newtonian Fluid with Constant p and la
Rectangular coordinates (x, y, z) x direction
(~u ~u ~u ~.) ~P l~u ~,, ~u~ p¥ + . ~ + ~ y + W ~ ~+,~~Vx~+~y7+ ~z~i+pgx y direction
(~ ~v ~ ~) ~P l~ P ~ + . ~ x + ~ y +WTz :  ~ + , [ ~ +
~
~
~y~+~z~I+Pgr
z direction
(~w ~w ~w ~w~ ~P /~w ~w ~w~ p ~+u~x+V~y+W~z ,]:  ~z + ~t[~x2+~+3y2 3z 2 ) + pgz Cylindrical coordinates (r, 0, z) r direction
/aVr
aVr
vo aVr
Pk~ + Vr~r + . r . 30 . .
]202
aVr~
aP
r + Vz 3z ] :  ~
+
[a (1 a ) 1 a2Vr 2 av o a2Vr] . . . g ~r ~r (rVr) + .r 2 . 302 r 2 30 + 3z2J + Pgr
0 direction
{3vo 3vo vo 3vo P ~   ~ + V r  ~ r +  r  ~ +
v,vo
3vo \
1 3P
~r + ~ ~z) :  r  ~ + ~' Tr
bTr (rv0) + 7 ~
+7 ~
+ az ~ J + pg0
z direction Plff't' + Vr"~r +  r   ~ + Vz" 3Z) = ~
+ ~tLrffrr \ r 3 r ) + ~    ~
+ 3Z2j + Pgz
Spherical coordinates (r, 0, ¢)* r direction
0Vr
p \ ~dt + V r ~ + ar 
r frO+ r sin 0 30
r
) = ~
"~~l,~VZvr
r2
r 2 a0 
0cot0 r2
2 av, ) r 2sin0 3¢ +Pgr
0 direction
/aVe
ave
ve aVe
v,
ave
vrvo
v~cot0~ 1 aP [ 2 3Vr a0 ) = . r 30 . +kt . V2VO+rE .
P~~ + Vr~r +  r 00+ r sin 0 3¢ + . r . . . r direction
{av~ av~ ve av, v~ av, V$Vr vev$ p~~ + Vr~r +  r   ~ + r s i n 0 a¢ + r + r cot 0
Vo r 2sin 20
2 cos 0 3v, r 2 sin 2 0 3¢ ) + Pg0
)
1 3P ( v_______L__ ~ 2 aVr r sin 0 0 ~ + g V2v*  F sin 2 0 + r 2 sin 0 3¢
* For spherical coordinates the laplacian is V2=~~rr r2~r + r 2 s i n e  ~ sine
+ r2sin2 e
2 cos 0 av0 r 2sin 20 3 ¢ ) + pg¢
BASIC CONCEPTS OF HEAT TRANSFER
1.19
TABLE 1.8 Summary of Dissipation Term VV:x in Different Coordinate Systems Rectangular coordinates (x, y, z): VV:~,  T,xx
(~) (~~)(o~)(~~ ~:)(~: ~~)(~: ~:) d T,yy
"]"CZZ
"]"Cxy
I
at T,yz
d
d "Czx
"t
Cylindrical coordinates (r, O, z): fDVr~ (1DV 0 _.~_) [DVz~ [D (_~_)1DVrl (1DV z DVO~ (Dr z DVr~ VV:~~'~rrt~r ) "l'r,oo ~Jt+ Xzzt~z j + Xro r~r +r DOJ4"'~Oz ~ dr" DZ / I'r'rz ~r "Jr"DZ /
Spherical coordinates (r, O, ~): (~)
(1Dvo_~ _~) ( l rsin 0 Dv,D~ v, vocotO) [DvolDvr _~) "at"%O + + T,~ + r + ~ r + T'r°t~r + r ~DO 
VV:'~ " T,rr
[DvO + \or + ~ImT_.
1 D V r V   ~ 7 ) (+ x0, 1Dr° rsinl DVOo D~ r sin 0 Dt~ ~ ~ +
v,
C2tr0 )
TABLE 1.7 The Viscous Dissipation Function • Rectangular coordinates (x, y, z):
~rr~u~ ;~v~ ~w~l (~v ~uv (~w ~v~ (~ ~w~ ~(~u ~v ~w~ +tOy/ + \ O z / J + ~x + Dy J + ~y +~z] + ~z + Dx )  7 ~x +~y +~z )
¢ = Lt Ox )
Cylindrical coordinates (r, 0, z):
2r(~vrV O= l\Dr] +
(2;~vo ~r)2+\DZ]J (DVZ~21 [ D (_~)1 Dv,12 + r~r +r ~+
DOJ
i1~ ~vol~ roar ~Vz~ ~ra~
l Ovo ~Vzl~
Dz J +\Dz +~r]3 [rorr (rVr) + r   ~ +~zJ
+ Lr~+
Spherical coordinates (r, O, ~): 2[(DVr~2
(1Dv 0
¢II = L\ Dr ] +
~
~r)2 ( 1 DV$ Vr vo cot O) 2] + r sin 0 D4~ r r
+ 
[ D (_~) 1 Dvr]2 [~_0__~( + r~r +r  ~ J +
v, ) ~
+
1
Dvo]2 [ 1 Dvr D (_~)] 2 + r sin 0 Dt~ +r ~r
r sin 0 D~
2 1 1~/ )~r (r2v') + 1 D (vo sin O) + 1 Dk'$] 2 r sin 0 DO r sin 0 Dt~' TABLE 1.8 Scalar Components of the Heat Flux Vector q" Rectangular (x, y, z) DT q'; =  k ~x
Cylindrical (r, 0, z) ~T q~=k Or
OT q'y"=k Dy
q~ = k  ~ r DO
1 DT q~ = k r DO
DT q'z'=k Dz
~T qz' =k Dz
1 ~T q~ =  k ~r sin 0 D~
1 DT
Spherical (r, 0, 0) OT q~ =k D~
1.20
CHAPTER ONE
For most engineering applications it is convenient to have the equation of thermal energy in terms of the fluid temperature and heat capacity rather than the internal energy or enthalpy. In general, for pure substances [11], Di Dt 
()
()
o,
Oi DP Oi DT 1 (1~T)~+ce ~ r~ + ~ e Dt  p
Dt
where 13is defined by Eq. 1.48. Substituting this into Eq. 1.54 we have the following general relation: DT DP q,,, p C p   ~ t  = V , k V T + T [ 3  ~ + laO +
(1.56)
For an ideal gas, ~ = l/T, and then DT DP pCp ~ = V . k V T + ~ + gO + q"
(1.57)
Note that Cp need not be constant. We could have obtained Eq. 1.57 directly from Eq. 1.54 by noting that for an ideal gas, di = Cp d T where Cp is constant and thus Di DT Dt  Cp Dt
For an incompressible fluid with specific heat c =cp = cv we go back to Eq. 1.52 (du = c d T ) to obtain DT pc   ~ = V . k V T + gO + q"
(1.58)
Equations 1.52, 1.54, and 1.56 can be easily written in terms of energy (heat) and momentum fluxes using relations for fluxes given in Tables 1.4, 1.6, and 1.8. The energy equation given by Eq. 1.58 (with q'"= 0 for simplicity) is given in Table 1.9 in different coordinate systems. For solids, the density may usually be considered constant and we may set V = 0, and Eq. 1.58 reduces to /)T
pc   ~ = V . k V T + q"
(1.59)
which is the starting point for most problems in heat conduction. The Energy E q u a t i o n f o r a Mixture. The energy equations in the previous section are applicable for pure fluids. A thermal energy equation valid for a mixture of chemical species is required for situations involving simultaneous heat and mass transfer. For a pure fluid, conduction is the only diffusive mechanism of heat flow; hence Fourier's law is used, resulting in the term V. kVT. More generally this term may be written Vq', where q" is the diffusive heat flux, i.e., the heat flux relative to the mass average velocity. More specifically, for a mixture, q" is now made from three contributions: (1) ordinary conduction, described by Fourier's law, kVT, where k is the mixture thermal conductivity; (2) the contribution due to interdiffusion of species, given by ~,i jiig and (3) diffusional conduction (also called the diffusionthermo effect or Dufour effect [6, 12]). The third contribution is of the second order and is usually negligible: q" =  k V T + ~ i
]iii
(1.60)
BASIC CONCEPTS OF HEAT TRANSFER TABLE 1.9
1.21
The Energy Equation* (for Newtonian Fluids of Constant p and k)
Rectangular coordinates (x, y, z):
pc~ ff+~gx +~Ty +WTz :k[~x~+ ~Dy+2 ~Z2 ] +2~ ~
(0u
+~,~/+~~z/j+~ ~+ ~x/+ ~+ Ox/+ ~+~~y/l
Cylindrical coordinates (r, 0, z): pep "~'Jt" Vr'~r [
r~
"lr"Vz
f['DVr \2
~Z = k
312~.1,[/~) "t
~r r ~r +  ~   ~ + Dz2 ]
)12 IDVzi21 [[Dvo l D v z ) 2 (DVz DVr~2 [1DVr D (~)] 2} ~ D0 "~"vr 1.\ DZ } J I~].l/~ 317 ~ "t~ Dr + Dz ] + ~ + rffrr
[I(DVo
Spherical coordinates (r, 0, 00): (aT
pCp ~
aT
+ Vr ~
1 O2T] + r2sin20 D~2]
vo aT v, ~~) [1 O ( a T ) 1 O(aT) "Jr" k r2 + sin 0 r ~ rsin0 5 ~~r ~r r2sin0 D0 ~
"Jr"~
f[DVr \2
+g
(_~DV 0
~r)2 ( 1
DV~ r sin 0 Dc~
Vr r
v0 cot 0) 2} r
{[ D (~e_) 1 DVr]2 [ 1 DVr D (~_)] 2 [Sinr0 _ ~ ( v , ) r~r +r ~1 + rsin 0 D00 + r ~ + ~
1 Dye]2) + r sin 0 DO
* The terms contained in braces [ }are associatedwithviscousdissipationand may usuallybe neglected except in systemswith large velocity gradients. Here ji is a diffusive mass flux of species i, with units of mass/(area x time), as mentioned before. Substituting Eq. 1.60 in, for example, Eq. 1.54, we obtain the energy equation for a mixture: Di DP P Dt  Dt + V • k V T 
V •
(~i
) jiii + gO + q"
(1.61)
For a nonreacting mixture the term V • (~'.i jiii) is often of minor importance. But when endothermic or exothermic reactions occur, this term can play a dominant role. For reacting mixtures the species enthalpies ii = i °i +
f;
o Cp'i d T
must be written with a consistent set of heats of formation i,°. at T O[13]. T h e C o n s e r v a t i o n E q u a t i o n for S p e c i e s
For a stationary control volume, the conservation equation for species is ~)Ci = V. (CiV) V . ji + ri" i)t rate of storage net rate of net rate of diffusion productionrate of species i per convectionof species of species i per of species i per unit volume i per unit volume unit volume unit volume
(1.62)
1.22
CHAPTERONE Using the mass conservation equation, the above equation can be rearranged to obtain DmL=  V . ji + r ~" P Dt
(1.63)
where mi is mass fraction of species i, i.e., where m i = C i / p , where p is the density of the mixture, ~ i Ci = P, and Ci is a partial density of species i (i.e., a mass concentration of species i). The conservation equation for species can also be written in terms of mole concentration and mole fractions, as shown in Refs. 10, 12, and 13. The mole concentration of species i is ci = C~/Mo where M~ is the molecular weight of the species. The mole fraction of species i is defined as X i " Ci/C, where c = ~i Ci" As is obvious, ~~im~ = 1 and ~ i Xi ~ 1. Equations 1.62 and 1.63 written in different coordinate systems are given in Ref. 10.
Use of Conservation Equations to Set Up Problems For a problem involving fluid flow and simultaneous heat and mass transfer, equations of continuity, momentum, energy, and chemical species (Eqs. 1.41, 1.44, 1.54, and 1.63) are a formidable set of partial differential equations. There are four i n d e p e n d e n t variables: three space coordinates (say, x, y, z) and a time coordinate t. If we consider a pure fluid, there are five equations: the continuity equation, three momentum equations, and the energy equation. The five accompanying d e p e n d e n t variables are pressure, three components of velocity, and temperature. Also, a thermodynamic equation of state serves to relate density to the pressure, temperature, and composition. (Notice that for natural convection flows the momentum and energy equations are coupled.) For a mixture of n chemical species, there are n species conservation equations, but one is redundant, as the sum of mass fractions is equal to unity. A complete mathematical statement of a problem requires specification of boundary and initial conditions. Boundary conditions are based on a physical statement or principle (for example: for viscous flow the component of velocity parallel to a stationary surface is zero at the wall; for an insulated wall the derivative of temperature normal to the wall is zero; etc.). A general solution, even by numerical methods, of the full equations in the four independent variables is difficult to obtain. Fortunately, however, many problems of engineering interest are adequately described by simplified forms of the full conservation equations, and these forms can often be solved easily. The governing equations for simplified problems are obtained by deleting superfluous terms in the full conservation equations. This applies directly to laminar flows only. In the case of turbulent flows, some caution must be exercised. For example, on an average basis a flow may be twodimensional and steady, but if it is unstable and as a result turbulent, fluctuations in the three components of velocity may be occurring with respect to time and the three spatial coordinates. Then the remarks about dropping terms apply only to the timeaveraged equations [7, 12]. When simplifying the conservation equation given in a full form, we have to rely on physical intuition or on experimental evidence to judge which terms are negligibly small. Typical resulting classes of simplified problems are: Constant transport properties Constant density Timewise steady flow (or quasisteady flow) Twodimensional flow Onedimensional flow Fully developed flow (no dependence on the streamwise coordinate) Stagnant fluid or rigid body
BASIC CONCEPTS OF HEAT TRANSFER
1.23
Terms may also be shown to be negligibly small by orderofmagnitude estimates [7, 12]. Some classes of flow that result are: Creeping flows: inertia terms are negligible. Forced flows: gravity forces are negligible. Natural convection: gravity forces predominate. Lowspeed gas flows: viscous dissipation and compressibility terms are negligible. Boundarylayer flows: streamwise diffusion terms are negligible.
DIMENSIONLESS GROUPS AND SIMILARITY IN HEAT TRANSFER Modern engineering practice in the field of heat transfer is based on a combination of theoretical analysis and experimental data. Often the engineer is faced with the necessity of obtaining practical results in situations where, for various reasons, physical phenomena cannot be described mathematically or the differential equations describing the problem are too difficult to solve. An experimental program must be considered in such cases. However, in carrying the experimental program the engineer should know how to relate the experimental data (i.e., data obtained on the model under consideration) to the actual, usually larger, system (prototype). A determination of the relevant dimensionless parameters (groups) provides a powerful tool for that purpose. The generation of such dimensionless groups in heat transfer (known generally as dimensional analysis) is basically done (1) by using differential equations and their boundary conditions (this method is sometimes called a differential similarity) and (2) by applying the dimensional analysis in the form of the Buckingham pi theorem. The first method (differential similarity) is used when the governing equations and their boundary conditions describing the problem are known. The equations are first made dimensionless. For demonstration purposes, let us consider the relatively simple problem of a binary mixture with constant properties and density flowing at low speed, where body forces, heat source term, and chemical reactions are neglected. The conservation equations are, from Eqs. 1.42, 1.46, 1.58, and 1.63, Mass
V •V = 0
(1.64)
DV p~   V P + IaV2V
Momentum
DT
Thermal energy
pc ~
Dml Dt
Species
= kV2T + ~t~
_
(1.65)
(1.66) (1.67)
DV2m 1
Using L and V as characteristic length and velocity, respectively, we define the dimensionless variables x*
V* 
X
L v V
y*  y L
z* z L
(1.68) (1.69)
1.24
CHAPTERONE
t
t* 
P* 
and also
(1.70)
L/V P
(1.71)
pV 2
TT~ T* = ~
(1.72)
T=T~
(1.73)
m * = m l  ml,w
ml**  ml.w where the subscript oo refers to the external freestream condition or some average condition and the subscript w refers to conditions adjacent to a bounding surface across which transfer of heat and mass occurs. If we introduce the dimensionless quantities (Eqs. 1.681.73), into Eqs. 1.641.67, we obtain, respectively,
V* • V* = 0
(1.74)
DV* V'P* Dt* 
D T* Dt*
1 Re Pr
Dm*
1
Dt*
Re Sc
  ~
Jr"
1
(1.75)
V*2V *
V * 2 T * .[.
2 Ec Re
O*
V*2m*
(1.76)
(1.77)
Obviously, the solutions of Eqs. 1.741.77 depend on the coefficients that appear in these equations. Solutions of Eqs. 1.741.77 are equally applicable to the model and prototype (where the model and prototype are geometrically similar systems of different linear dimensions in streams of different velocities, temperatures, and concentration), if the coefficients in these equations are the same for both model and prototype. These coefficients, Pr, Re, Sc, and Ec (called dimensionless parameters or similarity parameters), are defined in Table 1.10. Focusing attention now on heat transfer, from Eq. 1.14, using the dimensionless quantities, the heat transfer coefficient is given as
k 3T*] h
L
3y* y,=o
(1.78)
or, in dimensionless form, h L _ 3T*]
k
= Nu
(1.79)
~y* r=0
where the dimensionless group Nu is known as the Nusselt number. Since Nu is the dimensionless temperature gradient at the surface, according to Eq. 1.76 it must therefore depend on the dimensionless groups that appear in this equation; hence Nu = fl(Re, Pr, Ec)
(1.80)
For processes in which viscous dissipation and compressibility are negligible, which is the case in many industrial applications, we have Nu = f2(Re, Pr)
(forced convection)
(1.81)
0 tI
0
• ,,,.i
r~
0
0
II, I d
"u
•I=I
"~
n~
..
•
~
=
~:
•
~:
•~
.
.
.
I.~
~1~~ ~010
.
~
~
~,
~
~,
~
.
~
~
.,~
~
~
~., • ,..~
'~
o
g.. .c:
~~
¢~
°~
o=
~
°~o oI~
o=
~
~
o
~ o~
o
o
~
~
°
o l=l
o
~o
l:l
~
~
~
~
~
"
.0
•
°
~1~ ~1
0
,o
~
,.~
~
~
~
~
c~
~
:~
~
0
~:~
~
~
,~
o.
~1
o
~
~
0
,~
0
0
,~
0
~
1
0
°'~
~
...~
~
~,
~
0
~
~
.,,,~
~
,.~
..o
",~
..o~
"~
~
..o
O~o
"~
..o
~
~
..o~
~
~.
~
o o
.,.~
~
~~
~o
I~
o
o
o°
~
'
II1~11

:o ~
"~
IIv
• ,~
~
°
~
¢~
~
~J rO
~o
~1~ ~
~
~
•o ,~~ ~,~
.,..~
0
1.25
1.26
% r~ 0
.,=,~
0
0 Ill ==1
0 ~
0
0
0
~'~
•~ ~o ~ ~.~ ~ ~.~
0 .,=~
o
II
0
.,.,~
0
0 o 0 >
0 0 0 o 0
0
~
~
0
0
~
~.~ 0
~.~ ~
~o
0 .,..~ o
0 o ¢) o
0
~'~
~~ .,.,~
II
0
~.o .~
~s
II o
0 o
0 o
o
o3 "0
II
0 °,.,~
2~
II
0
a3
a3
o o
0 o o o
r~ o
r~ r~ a3
0
o
0
:~
°.,~
o~ a3
0 ~0
0
a3 lq
r~
~
8~ 0
.,..~
0
0
0
.£ 1i
0,.~
~
0
=
0
0
0 . ,.,,~
II
z~
0
o
II
r~
= C
C
0
0
0
o
0
o
r~ 0 o
.,..~
0 0
,. , ~
~
"~,
E
,.c:
o
o
wn~
"~
o
w ~J .,.q w
I,
~E o,.cl
o
~
8*
~
•'  ~ '~.
.~.
1.27
1.28
CHAPTERONE In the case of buoyancyinduced flow, Eq. 1.65 should be replaced with the simplified version [16] of Eq. 1.50, and, following a similar procedure, we should obtain Nu = f3(Gr, Pr)
(natural convection)
(1.82)
where Gr is the Grashof number, defined in Table 1.10. Also, using the relation of Eq. 1.17 and dimensionless quantities,
h a  D /)m* II
(1.83)
L /)Y* ly*:o L  Om* hD D 3y* Iy*:0 = Sh
or
(1.84)
This parameter, termed the Sherwood number, is equal to the dimensionless mass fraction (i.e., concentration) gradient at the surface, and it provides a measure of the convection mass transfer occurring at the surface. Following the same argument as before (but now for Eq. 1.77), we have Sh = fa(Re, Sc)
(forced convection, mass transfer)
(1.85)
The significance of expressions such as Eqs. 1.801.82 and 1.85 should be appreciated. For example, Eq. 1.81 states that convection heat transfer results, whether obtained theoretically or experimentally, can be represented in terms of three dimensionless groups instead of seven parameters (h, L, V, k, Cp, It, and p). The convenience is evident. Once the form of the functional dependence of Eq. 1.81 is obtained for a particular surface geometry (e.g., from laboratory experiments on a small model), it is known to be universally applicable, i.e., it may be applied to different fluids, velocities, temperatures, and length scales, as long as the assumptions associated with the original equations are satisfied (e.g., negligible viscous dissipation and body forces). Note that the relations of Eqs. 1.80 and 1.85 are derived without actually solving the system of Eqs. 1.641.67. References 3, 7, 12, 15, 16, and 18 cover the above procedure in more detail and also include many different cases. It is important to mention here that once the conservation equations are put in dimensionless form it is also convenient to make an orderofmagnitude assessment of all terms in the equations. Often a problem can be simplified by discovering that a term that would be very difficult to handle if large is in fact negligibly small [7, 12]. Even if the primary thrust of the investigation is experimental, making the equations dimensionless and estimating the orders of magnitude of the terms is good practice. It is usually not possible for an experimental test to include (simulate) all conditions exactly; a good engineer will focus on the most important conditions. The same applies to performing an orderofmagnitude analysis. For example, for boundarylayer flows, allowance is made for the fact that lengths transverse to the main flow scale with a much shorter length than those measured in the direction of main flow. References 7, 12, and 17 cover many examples of the orderofmagnitude analysis. When the governing equations of a problem are unknown, an alternative approach of deriving dimensionless groups is based on use of dimensional analysis in the form of the Buckingham pi theorem [9, 11, 14, 16, 18]. The success of this method depends on our ability to select, largely from intuition, the parameters that influence the problem. For example, knowing in advance that the heat transfer coefficient in fully developed forced convection in a tube is a function of certain variables, that is, h = f(V, p, kt, Cp, k, D), we can use the Buckingham pi theorem to obtain Eq. 1.81, as shown in Ref. 11. However, this method is carried out without any consideration of the physical nature of the process in question, i.e., there is no way to ensure that all essential variables have been included. However, as shown above, starting with the differential form of the conservation equations we have derived the similarity parameters (dimensionless groups) in rigorous fashion.
BASIC CONCEPTS OF HEAT TRANSFER
1.29
In Table 1.10 those dimensionless groups that appear frequently in the heat and mass transfer literature have been listed. The list includes groups already mentioned above as well as those found in special fields of heat transfer. Note that, although similar in form, the Nusselt and Biot numbers differ in both definition and interpretation. The Nusselt number is defined in terms of thermal conductivity of the fluid; the Biot number is based on the solid thermal conductivity.
UNITS AND CONVERSION FACTORS The dimensions that are used consistently in the field of heat transfer are length, mass, force, energy, temperature, and time. We should avoid using both force and mass dimensions in the same equation, since force is always expressible in dimensions of mass, length, and time, and vice versa. We do not make a practice of eliminating energy in terms of force times length, because the accounting of work and heat is practically always kept separate in heat transfer problems. In this handbook both SI (the accepted abbreviation for Systdme International d'Unit~s, or International System of Units) and English engineering units* are used simultaneously throughout. The base units for the English engineering units are given in the second column of Table 1.11. The unit of force in English units is the pound force (lbf). However, the use of the pound mass (Ibm) and pound force in engineering work causes considerable confusion in the proper use of these two fundamentally different units. TABLE 1.11 Quantity
C o n v e r s i o n Factor gc for the C o m m o n Unit Systems SI
English engineering*
cgs*
Metric engineering
Mass
kilogram, kg
p o u n d mass, Ibm
gram, g
kilogram mass, kg
Length Time Force
meter, m second, s newton, N 1 kg.m/(N.s2) ~
foot, ft second, s, or hour, h pound force, lbf 32.174lbm.ft/(lbf.s2)
centimeter, cm second, s dyne, dyn 1 g.crn/(dyn.s2)
meter, m second, s kilogram force, kgf 9.80665kg.rn/(kgf.s2)
gc
* In this system of units the temperature is given in degrees Fahrenheit (°F). * Centimetergramsecond: this system of units has been used mostly in scientific work. * Since 1 kg.m/s 2 = 1 N, then gc = 1 in the SI system of units.
The two can be related as 1 Ibm × 32.174 ft/s 2 1 lbf =
whence
gc
gc = 32.174 lbmft/(lbf.s 2)
Thus, gc is merely a conversion factor and it should not be confused with the gravitational acceleration g. The numerical value of gc is a constant depending only on the system of units involved and not on the value of the gravitational acceleration at a particular location. Values
* Also associated with this system of units are such names as U.S. Customary Units, British engineering units, engineering units, and footpoundsecond system of units. The name English engineering units, or, for short, English units, is selected in this handbook because it has been used by practicing engineers more frequently than the other names mentioned.
1.30
CHAPTER ONE
TABLE 1.12
SI Base and Supplementary Units
Quantity
Unit
Length Mass Time Electric current Thermodynamic temperature Amount of substance Luminous intensity Plane angle* Solid angle*
meter (m) kilogram (kg) second (s) ampere (A) kelvin (K) mole (mol) candela (cd) radian (rad) steradian (sr)
* Supplementary units.
of gc corresponding to different systems of units found in engineering literature are given in Table 1.11. The SI base units are summarized in Table 1.12. The SI units comprise a rigorously coherent form of the metric system, i.e., all remaining units may be derived from the base units using formulas that do not involve any numerical factors. For example, the unit of force is the newton (N); a 1N force will accelerate a 1kg mass at 1 m/s 2. Hence 1 N = 1 kg.m/s 2. The unit of pressure is the N/m 2, often referred to as the pascal. In the SI system there is one unit of energy (thermal, mechanical, or electrical), the joule (J); 1 J = 1 N.m. The unit for energy rate, or power, is joules per second (J/s), where one J/s is equivalent to one watt (1 J/s = 1 W). In the English system of units it is necessary to relate thermal and mechanical energy via the mechanical equivalent of heat J~ Thus
Jc x thermal energy = mechanical energy The unit of heat in the English system is the British thermal unit (Btu). When the unit of mechanical energy is the poundforcefoot (lbcft), then Jc = 778.16 lbfft/Btu as I Btu  778.16 lbf.ft. Happily, in the SI system the units of heat and work are identical and
Jc is unity. Since it is frequently necessary to work with extremely large or small numbers, a set of standard prefixes has been introduced to simplify matters (Table 1.13). Symbols and names for all units used in the handbook are given in Table 1.14. Conversion factors for commonly
TABLE 1.13
SI Prefixes (Decimal Multiples and Submultiples in SI Are Formed by Adding the Following Prefixes to the SI Unit) Factor
Prefix
Symbol
Factor
Prefix
10TM 1015 1012
exapetateragigamegakilohectodeka
E P T G M k h da
101
deci
10 2 10 3
centimilli
106 109
micronano
10 12 10 15
picofemto
1018
atto
10 9 10 6 10 3
102 10
Symbol
BASIC CONCEPTS OF HEAT TRANSFER TABLE 1.14
Symbols and Names for Units Used in the Handbook
Symbol
Name
Symbol
Name
A Btu C °C cal cm deg dyn °F ft g H h hp in
ampere British thermal unit coulomb (= A.s) degree Celsius calorie centimeter degree dyne degree Fahrenheit foot gram henry (= V.s/A) hour horsepower inch joule (= N.m) kelvin (thermodynamic temperature)
kg kgf Ibm lbe m min mol N Pa pdl °R rad s sr T V W
kilogram mass kilogram force pound mass pound force meter minute mole newton Pascal (= N/m 2) poundal degree Rankine radian (plane angle) second steradian (solid angle) tesla (= V.s/m 2) volt watt (= J/s)
J
K
1.31
used q u a n t i t i e s in h e a t transfer, f r o m SI to English e n g i n e e r i n g units and vice versa, are given in Table 1.15. C o n v e r s i o n factors for mass, density, pressure, energy, specific energy, specific heat, therm a l conductivity, d y n a m i c viscosity, and k i n e m a t i c viscosity in different s y s t e m s of units are also given in Chap. 2 (Tables 2.12.9).
TABLE 1.15
Conversion Factors for Commonly Used Quantities in Heat Transfer
Quantity
SI )English
English ~ SI*
Area
1 m2= 10.764 ft 2 = 1550.0 in 2
ft 2 = 0.0929 m 2 in 2 = 6.452 x 104 m 2
Density
1 kg/m 3 = 0.06243 lbm/ft3
Ibm/ft3 = 16.018 kg/m 3 slug/ft 3 = 515.379 kg/m 3
Energy t
1 J = 9.4787 x 104 Btu = 6.242 x 10 TM eV
Btu = 1055.056 J cal = 4.1868 J lbcft = 1.3558 J hp.h = 2.685 x 10 6 J
Energy per unit mass
1 J/kg = 4.2995 × 104 Btu/lbm
Btu/lbm = 2326 J/kg
Force
1 N = 0.22481 lbf
lbf = 4.448 N pdl = 0.1382 N
Heat flux
1 W/m 2 = 0.3171 Btu/(h.ft z)
Btu/(h'ft 2) = 3.1525 W/m: kcal/(h.m 2) = 1.163 W/m 2 cal/(s.cm 2) = 41.870 x 10 3 W / m
Heat generation per unit volume
1 W/m 3 = 0.09665 Btu/(h.ft 3)
Btu/(h'ft 3) = 10.343 W/m 3
Heat transfer coefficient
1 W/(m2K)=0.17612 Btu/(h.ft2.°F)
Btu/(h.ft 2°F) = 5.678 W/(maK) kcal/(h.m:.°C) = 1.163 W/(m2.K) cal/(s.cm 2.°C) = 41.870 x 103 W/(mR.K)
2
1.32
CHAPTER ONE
TABLE 1.15
C o n v e r s i o n Factors for C o m m o n l y U s e d Q u a n t i t i e s in H e a t T r a n s f e r Quantity
(Continued)
SI ) E n g l i s h
E n g l i s h ) SI*
H e a t t r a n s f e r rate
i W = 3.4123 B t u / h
1 B t u / h = 0.2931 W
Length
1 m = 3.2808 ft = 39.370 in
1 1 1 1 1
1 1 1
ft = 0.3048 m in = 2.54 cm = 0.0254 m y a r d = 0.9144 m s t a t u t e mile = 1609 m m i l = 0.001 in = 2.54 x 10 5 m lightyear = 9.46 x 1015 m a n g s t r o m = 101° m m i c r o n = 106m
Mass
i kg = 2.2046 Ibm
1 Ibm = 0.4536 kg i slug = 14.594 kg
Mass flow rate
1 kg/s = 7936.6 lbm/h = 2.2046 lbm/s
1 lbm/h = 0.000126 kg/s 1 lbm/s = 0.4536 kg/s
Power
1 W = 3.4123 B t u / h
1 1 1 1
B t u / h = 0.2931 W Btu/s = 1055.1 W lbcft/s = 1.3558 W hp = 745.7 W
P r e s s u r e a n d stress*
1 N/m2= = = =
1 1 1 1 1
lbJft2 = 47.88 N / m 2 lbf/in2 = 6894.8 N / m 2 psi = 1 lbJin2 = 6894.8 N / m 2 s t a n d a r d a t m o s p h e r e = 1.0133 x 105 N / m 2 bar = 1 x 105 N / m 2
Specific h e a t
1 J / ( k g . K ) = 2.3886 x 10 4 Btu/(lbm.°F)
1 B t u / ( l b m . ° F ) = 4187 J / ( k g . K )
Surface t e n s i o n
1 N / m = 0.06852 lbf/ft
1 lbf/ft = 14.594 N / m 1 d y n / c m = 1 x 10 3 N / m
Temperature
T(K) = = = T(°C) =
0.020886 l b J f t 2 1.4504 x 104 l b J i n 2 4.015 x 1 0 3 in w a t e r 2.953 x 104 in H g
T(°C) + 273.15 T(°R)/1.8 [T(°F) + 459.67]/1.8 [ T ( ° F )  32]/1.8
T ( ° R ) = 1.8T(K) = T(°F) + 459.67 T(°F) = 1.8T(°C) + 32 = 1 . 8 [ T ( K )  273.15] + 32
1K=I°C = 1.8°R = 1.8OF
1OR = 1OF
Thermal conductivity
1 W / ( m . K ) = 0.57782 B t u / ( h . f t . ° F )
1 B t u / ( h  f t  ° F ) = 1.731 W / ( m . K ) 1 k c a l / ( h . m . ° C ) = 1.163 W / ( m  K ) 1 c a l / ( s . c m . ° C ) = 418.7 W / ( m . K )
T h e r m a l diffusivity
1 m2/s  10.7639 ft2/s
1 ft2/s = 0.0929 m2/s 1 ft2/h = 2.581 x 10 5 m2/s
T h e r m a l resistance
1 K / W = 0.52750 ° F . h / B t u
1 ° F . h / B t u = 1.8958 K / W
Velocity
1 m/s = 3.2808 ft/s
1 ft/s = 0.3048 m/s 1 k n o t = 0.5144 rn/s
Viscosity ( d y n a m i c ) ~
1 N.s/m2 = 0.672 lbm/(ft's)
1 l b m / ( f t ' s ) = 1.4881 N . s / m 2 1 c e n t i p o i s e = 10 2 poise = 1 x 1 0 3 N . s / m 2
Temperature difference
= 2.089 x 10 2 l b f ' s / f t 2 Viscosity ( k i n e m a t i c )
I mZ/s = 10.7639 ftZ/s
 1 K/1.8 = 1°C/1.8
1 ft2/s = 0.0929 m2/s = 929 stoke 1 m2/s = 10,000 stoke
BASIC C O N C E P T S O F H E A T T R A N S F E R
TABLE 1.15
1.33
Conversion Factors for Commonly Used Quantities in Heat Transfer (Continued)
Quantity
SI + English
English + SI*
Volume
I m 3 = 35.3134 ft 3
1 ft 3 = 0.02832 m 3 1 in 3 = 1.6387 x 105 m 3 1 gal (U.S. liq.) = 0.003785 m 3 1 gal (U.K. liq.) = 0.004546 m 3 1 m 3 = 1000 liter 1 gal (U.S. liq.) = 4 quarts = 8 pints = 128 ounces 1 quart = 0.946 x 10 3 m 3
Volume flow rate
1 m3/s = 35.3134 ft3/s = 1.2713 x 105 ft3/h
1 ft3/h = 7.8658 x 106 m3/s 1 ft3/s = 2.8317 x 102 m3/s 1 gal (U.S. liq.)/min = 6.309 x 10 5 m3/s = 0.2271 m3/hr
* Some units in this column belong to the cgs and mks metric systems. , Definition of the units of energy based on thermal phenomena: 1 Btu = energy required to raise 1 Ibm of water I°F at 68°F 1 cal = energy required to raise 1 g of water I°C at 20°C * The SI unit for the quantity pressure is the pascal (Pa); 1 Pa = 1 N/m 2. Also expressed in equivalent units of kg/(s.m).
NOMENCLATURE Symbol, Definition, SI Units, English Units A
heat transfer area: m 2, ft 2
a
acceleration: m/s 2, ft/s 2
a
s p e e d of sound: m/s, ft/s
C
mass c o n c e n t r a t i o n of species: kg/m 3, lbm/ft 3
c
specific heat: J/(kg.K), Btu/(lbm'°F)
Cp
specific heat at constant pressure: J/(kg.K), Btu/(lbm'°F)
cv
specific heat at constant volume: J/(kg.K), Btu/(lbrn" °F)
D
tube inside diameter, diameter: m, ft
D
diffusion coefficient: mE/s, ft2/s
Ec
E c k e r t n u m b e r (see Table 1.10)
e
emissive power: W / m 2, Btu/(h.ft 2)
eb
b l a c k b o d y emissive power: W / m 2, Btu/(h.ft 2)
F
force: N, lbf
F12
view factor ( g e o m e t r i c shape factor for radiation f r o m o n e b l a c k b o d y to another)
~12
real b o d y view factor ( g e o m e t r i c shape and emissivity factor for radiation f r o m o n e gray b o d y to a n o t h e r )
f
f r e q u e n c y of vibration (see Table 1.10): s 1 d e n o t e s function of Eqs. 1.801.82 and 1.85
1.34
CHAPTERONE Gr
g g gc h ho i itg
J
J k L M m
Nu P Pr AP q q,, q,, q tp! Rth
Re F r r .t
Sc Sh St T AT t U
I1
V V V
Vr Vz
Grashof number (see Table 1.10) gravitational acceleration: m]s 2, ft/s 2 gravitational acceleration (vector): m/s 2, ft/s 2 conversion factor (see Table 1.11): lbm'ft/(lbcs 2) heat transfer coefficient: W/(m2.K), Btu/(h.ft 2"°F) mass transfer coefficient: m/s, ft/s enthalpy per unit mass: J/kg, Btu/lbm latent heat of evaporation: J/kg, Btu/lbm heat of formation: J/kg, Btu/lbm mass diffusion flux of species: kg/(sm2), lbm/(h'ft 2) mass diffusion flux of species (vector): kg/(s.m2), lbm/(h'ft 2) thermal conductivity: W/(m.K), Btu/(h.ft.°F) length: m, ft mass: kg, Ibm mass fraction of species (Eq. 1.63) Nusselt number (see Table 1.10) pressure: Pa (N/m2), lbf/ft 2 Prandtl number (see Table 1.10) pressure drop: Pa (N/m2), lbf/ft 2 heat transfer rate: W, Btu/h heat flux (vector): W/m 2, Btu/(h.ft 2) heat flux: W/m E, Btu/(hft 2) volumetric heat generation: W/m 3, Btu/(hft 3) thermal resistance: K/W, h. °F/Btu Reynolds number (see Table 1.10) radial distance in cylindrical or spherical coordinate: m, ft recovery factor (Eq. 1.19) volumetric generation rate of species: kg/(s.m3), lbm/(hft 3) Schmidt number (see Table 1.10) Sherwood number (see Table 1.10) Stanton number (see Table 1.10) temperature: °C, K, °E °R temperature difference: °C, °F time: s velocity component in the axial direction (x direction) in rectangular coordinates: m/s, ft/s internal energy per unit mass: J/kg, Btu/lbm velocity: m/s, ft/s velocity (vector): m/s, ft/s velocity component in the y direction in rectangular coordinates: m/s, ft/s velocity component in the r direction: m/s, ft/s velocity component in the z direction: m/s, ft/s
BASIC CONCEPTS OF H E A T T R A N S F E R
Vo V~ W X
y Z
velocity component in the 0 direction: m/s, ft/s velocity component in the ¢~direction: m/s, ft/s velocity component in the z direction in rectangular coordinates: m/s, ft/s rectangular coordinate: m, ft rectangular coordinate: m, ft rectangular or cylindrical coordinate: m, ft
Greek
8 8o 8r E.H E.M E.m
0 g V
P
thermal diffusivity: m2/s, ft2/s coefficient of thermal expansion: K 1, °R1 hydrodynamic boundary layer thickness: m, ft concentration boundary layer thickness: m, ft thermal boundary layer thickness: m, ft emissivity eddy diffusivity of heat: m2/s, ft2/s eddy diffusivity of momentum: m2/s, ft2/s eddy diffusivity of mass: m2/s, ft2/s angle in cylindrical and spherical coordinates: rad, deg molecular mean free path: m, ft dynamic viscosity: Pas, lbm/(Sft) kinematic viscosity: m2/s, ft2/s density: kg/m 3, lbm/ft 3 surface tension (see Table 1.10): N/m, lbf/ft StefanBoltzmann radiation constant: W/(m2.K4), Btu/(h.ft 2.°R4) shear stress: N/m:, lbf/ft 2 shear stress tensor: N/m R, lbf/ft 2 dissipation function (see Table 1.7): s2 angle in spherical coordinate system: rad, deg
Subscripts d aw cr
f g i l m r s
sat t w
surroundings adiabatic wall critical fluid gas (vapor) species i liquid mean radiation (Eq. 1.36) solid saturation total wall
1.35
1.36
CHAPTERONE x
x component
y
y component
z
z component
0
0 component ~ component
Miscellaneous Subscripts 1
species 1 in binary mixture of I and 2
oo
freestream condition
Superscripts •
fluctuating component (for example, X' is the fluctuating component of X) time average (for example, X is the time average of X)
Mathematical Operation Symbols d/dx
derivative with respect to x: m 1, ft 1
i)/i)t
partial time derivative operator:
d/dt
total time derivative operator: s1 (Eq. 1.37)
D/Dt
substantial time derivative operator: s1 (Eq. 1.38)
V
del operator (vector): m i, ft 1
V2
laplacian operator: m 2, ft 2
S 1
REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15.
E Kreith and W. Z. Black, Basic Heat Transfer, Harper & Row, New York, 1980. J.P. Holman, Heat Transfer, 8th ed., McGrawHill, New York, 1997. E E Incropera and D. P. DeWitt, Fundamentals of Heat Transfer, 4th ed., Wiley, New York, 1996. M.N. Ozisik, Basic Heat Transfer, McGrawHill, New York, 1977. R.E. Treybal, MassTransfer Operations, 3d ed., McGrawHill, New York, 1980. W. M. Kays and M. E. Crawford, Convective Heat and Mass Transfer, 2d ed., McGrawHill, New York, 1980. H. Schlichting, BoundaryLayer Theory, 7th ed., McGrawHill, New York, 1979. J.O. Hinze, Turbulence, 2d ed., McGrawHill, New York, 1975. J.H. Lienhard, A Heat Transfer Textbook, PrenticeHall, Englewood Cliffs, NJ, 1981. R. B. Bird, W. E. Stewart, and E. N. Lightfoot, Transport Phenomena, Wiley, New York, 1960. W. M. Rohsenow and H. Y. Choi, Heat, Mass, and Momentum Transfer, PrenticeHall, Englewood Cliffs, NJ, 1961. D. K. Edwards, V. E. Denny, and A. E Mills, Transfer Processes: An Introduction to Diffusion, Convection, and Radiation, 2d ed., Hemisphere, Washington, DC, and McGrawHill, New York, 1979. W. C. Reynolds and H. C. Perkins, Engineering Thermodynamics, 2d ed., McGrawHill, New York, 1977. A. S. Foust, L. A. Wenzel, C. W. Clump, L. Mans, and L. B. Andersen, Principles of Unit Operations, 2d ed., Wiley, New York, 1980. E M. White, Viscous Fluid Flow, McGrawHill, New York, 1974.
BASIC CONCEPTS OF HEAT TRANSFER
1.3']?
16. B. Gebhart, Heat Transfer, 2d ed., McGrawHill, New York, 1971. 17. E. R. G. Eckert and R. M. Drake Jr., Analysis of Heat and Mass Transfer, McGrawHill, New York, 1972. 18. V. P. Isachenko, V. A. Osipova, and A. S. Sukomel, Heat Transfer, Mir Publishers, Moscow, 1977. 19. S. Whitaker, Elementary Heat Transfer Analysis, Pergamon, New York, 1976.
CHAPTER 2
THERMOPHYSICAL PROPERTIES Thomas F. Irvine Jr. State University of New York at Stony Brook W h e n organizing a chapter of thermophysical properties with limited space, some difficult decisions have to be made. Since this is a h a n d b o o k for heat transfer practitioners, emphasis has b e e n placed on transport rather than t h e r m o d y n a m i c properties. The primary exception has b e e n the inclusion of densities and isobaric specific heats, which are n e e d e d for the calculation of Prandtl n u m b e r s and thermal diffusivities. In the spirit of today's c o m p u t e r usage, a n u m b e r of gas properties are given in equation rather than tabular form. However, they are accompani ed by skeleton tables to allow for program checks. Because new refrigerants are being considered and used in technical applications, a number of transport and t h e r m o d y n a m i c propert y tables are included for these substances. W h e n e v e r possible, the properties in this chapter are divided into those for gases, liquids, and solids. There are unavoidable overlaps to this a r r a n g e m e n t when the tables account for phase changes such as in the case of water.
CONVERSION FACTORS TABLE 2.1
Conversion Factors for Units of Density kg/m 3
lbm/ft3
kg/m 3 1 lbm/~3 16.0185 lbm/(U.K, gal) 99.7763 lbm/(U.S, gal) 119.826 slug]~ 3 515.38 g/cm3 1000 t/m 3 1000 U.K. ton/yd 3 1328.94 U.S. ton/yd 3 1186.5
0.06243 1 6.22884 7.48052 32.1740 62.428 62.428 82.963 74.075
lbm/(U.K, gal) lbm/(U.S, gal) 0.01002 0.16054 1 1.20094 5.1653 10.0224 10.0224 13.319 11.892
8.3454.3 0.13368 0.83268 1 4.3011 8.34540 8.34540 11.0905 9.9022
slug/ft3
g/cm3
t/m 3
1.9403.3 0.03108 0.19360 0.2325 1 1.9403 1.9403 2.5785 2.3023
0.001 0.01602 0.09976 0.11983 0.51538 1 1 1.3289 1.1865
0.001 0.01602 0.09976 0.11983 0.51538 1 1 1.3289 1.1865
U.K. ton/yd 3 U.S. ton/yd 3 7.5248.4 1.2054.2 7.5080.2 9.0167.2 0.43435 0.75250 0.75250 1 0.89286
8.4278.4 1.3500.2 8.4090.2 1.0099.1 0.43435 0.84280 0.84280 1.120 1
The notation 8.3454.3 signifies 8.3454 x 103. TABLE 2.2
Conversion Factors for Units of Energy j o u l e (J)
joule (J) ftlbf calth Callx liter.atm kJ Btu hp.h kWh thermie
1 1.35582 4.184 4.1868 101.328 1000 1055.05 2.6845.+6 3.600.+6 4.184.+6
ft.lbf
calth
0.73756 1 3.08596 3.08798 74.735 737.56 778.16 1.98.+6 2.6557.+6 3.087.+6
0.23901 0.32405 1 1.00066 24.218 239.01 252.16 641,617 860,564 10 6
calyx 0.23885 0.32384 0.99934 1 24.202 238.85 252.00 641,197 8.6.+5 9.9934.+5
liter.atm 9.8690.3 1.33205.2 0.04129 0.04132 1 9.86896 10.4122 26,494 35,534 4.129.+3
kJ
Btu
hph
10 3 9.4783.4 3.7251.7 1.3558.3 1.2851.3 5.0505.7 4 . 1 8 4 .   3 3.9657.3 1.5586.6 4.1868.3 3.9683.3 1.5596.6 0.10325 9.6041.2 3.7745.5 1 0.94783 3.7251.4 1.05505 1 3.9301.4 2684.52 2544.5 1 3600 3412.8 1.34125 4.184.+3 3.9657.+3 1.5586
kWh 2.7773.7 3.7655.7 1.1620.6 1.1628.6 2.8142.5 2.7773.4 2.9302.4 0.74558 1 1.1620
The notation 9.8690.3, 4.184.+6 signifies 9.8690 x 103, 4.184 x 106. 2.1
2.2
CHAPTER TWO
TABLE 2.3
Conversion Factors for Units of Mass
g Ibm kg slug U.S. ton (short ton) t (metric ton) U.K. ton (long ton)
g
Ibm
kg
slug
1 453.592 1000 14,593.9 907,185 106 1,016,047
2.2046.3 1 2.20462 32.1740 2000 2204.62 2240
0.001 0.45359 1 14.5939 907.185 1000 1016.05
6.8522.5 0.031081 0.06852 1 62.162 68.5218 69.621
U.S. ton (short ton)
t (metric ton)
U.K. ton (long ton)
1.1023.6 0.0005 1.1023.3 0.01609 1 1.10231 1.12
106 4.5359.4 0.001 0.01459 0.90719 1 1.01604
9.8421.7 4.4643.4 9.8421.4 0.01436 0.89286 0.98421 1
The notation 2.2046.3 signifies 2.2046 x 103. National Bureau of Standards Letter Circular 1071, 7 pp., 1976.
Sourc~
TABLE 2.4
Conversion Factors for Units of Pressure
dyn/cm a* N/m 2= Pa dyn/cm 2 N/m 2 lbf/ft2 mmHg in (H20) in (Hg) lbf/in2 kg/cm 2 bar atm
1 10 478.79 1333.22 2490.8 33864 68,947 980,665 106 1,013,250
0.1 1 47.879 133.32 249.08 3386.4 6894.7 98,067 105 101,325
lbf/fta
mmHg
in (H20)
in (Hg)
lbf/in2
kg/cm 2
bar
atm
2.0886.3 2.0886.2 1 2.7845 5.2023 70.727 144 2048.2 2088.5 2116.2
7.5006.4 7.5006.3 0.35913 1 1.8683 25.400 51.715 735.57 750.06 760
4.0148.4 4.0148.3 0.19221 0.53526 1 13.596 27.680 393.71 401.47 406.79
2.9530.5 2.9530.4 1.4138.2 0.03937 0.07355 1 2.03601 28.959 29.530 29.921
1.4504.5 1.4504.4 6.9444.3 0.01934 0.03613 0.49116 1 14.223 14.504 14.696
1.0197.6 1.0197.5 4.8824.4 1.3595.3 2.5399.3 0.03453 0.07031 1 1.01972 1.03323
106 105 4.7880.4 1.3332.3 2.4908.3 0.03386 0.06895 0.98067 1 1.01325
9.8692.7 9.8692.6 4.7254.4 1.3158.3 2.4585.3 0.03342 0.06805 0.96784 0.98692 1
* 1 dyn/cm2= 1 microbar. The notation 2.0886.3 signifies 2.0886 x 103.
TABLE 2.5
ft'lbf/lbm J/g Btu/lbm cal/g
TABLE 2.6
Conversion Factors for Units of Specific Energy ft'lbf/lbm
J/g
Btu/lbm
cal/g
1 334.54 778.16 1400
2.989.3 1 2.326 4.184
1.285.3 0.4299 1 1.8
7.143.4 0.2388 0.5556 1
Conversion Factors for Units of Specific Energy per Degree
,,
J/(g. K) BtUth/(lbm"°F) caltn/(g" °C) B tuxT/(lbm"°F) calIT/(g" °C)
J/(g. K)
BtUth/(lb" °F)
calth/(g" °C)
BtU~T/(Ibm" °F)
callv/(g" °C)
1 4.184 4.184 4.1868 4.1868
0.23901 1 1 1.00067 1.00067
0.23901 1 1 1.00067 1.00067
0.23885 0.99933 0.99933 1 1
0.23885 0.99933 0.99933 1 1
THERMOPHYSICAL PROPERTIES TABLE 2.7
2.3
Conversion Factors for Units of Thermal Conductivity Btu.in/(hft2°F) W/(m.K)
Btu.in/(h.ft2.°F) W/(mK) kcal/(hm. °C) Btu/(hft.°F) W/(cm.K) cal/(scm°C) Btu.in/(s.ft 2.°F)
1 6.938 8.064 12 694 2903 3600
kcal/(hm.°C)
Btu/(hft.°F)
0.1240 0.8604 1 1.488 86.04 360 446.7
0.08333 0.5782 0.6720 1 57.82 241.9 300
0.1441 1 1.162 1.730 100 418.4 519.2
W/(cm.K) cal/(scm.°C) Btu.in/(s.ft2.°F) 1.441.3 0.01 0.01162 0.01730 1 4.184 5.192
3.445.4 2.390.3 2.778.3 4.134.3 0.2390 1 1.2402
2.777.4 1.926.3 2.240.3 3.333.3 0.1926 0.8063 1
The notation 1.441.3 signifies 1.441 x 103.
TABLE 2.8
micropoise lbm/(ft'h) centipoise slug/(ft.h) poise (P) N.s/m 2 Pa.s lbm/(ft's) lbcs/ft 2
Conversion Factors for Units of Dynamic Viscosity micropoise
lbm/(ft'h)
centipoise
slug/(ft.h)
poise (P)
N.s/m 2
Pa.s
lbm/(S'ft)
lbcs/ft 2
1 4134 104 1.3300.+5 106 107 107 1.4882.+7 4.7880.+8
2.4191.4 1 2.4191 32.174 241.91 2419.1 2419.1 3600 1.1583.+5
104 0.4134 1 13.300 100 1000 1000 1488.2 4.7880.+4
7.5188.6 3.1081.2 7.5188.2 1 7.5188 75.188 75.188 111.89 3600
106 4.1338.3 0.01 0.1330 1 10 10 14.882 478.80
107 4.1338.4 0.001 1.3300.2 0.1 1 1 1.4882 47.880
107 4.1338.4 0.001 1.3300.2 0.1 1 1 1.4882 47.880
6.7197.8 2.7778.4 6.7197.4 8.9372.3 6.7197.2 0.6720 0.6720 1 32.174
2.0885.9 8.6336.6 2.0885.5 2.7778.4 2.0835.3 2.0885.2 2.0885.2 0.03108 1
1 lbm/(ft'h)= 1 poundalh/ft2; 1 P = 1 g/(cms). The notation 2.4191.4,1.4882.+7 signifies 2.4191 x 104, 1.4882 x 107.
TABLE 2.9
ft2/h stokes (St) m2/h ft2/s m2/s
Conversion Factors for Units of Kinematic Viscosity ft2/h
stokes (St)
m2/h
ft2/s
m2/s
1 3.8750 10.7639 3.600 38,750
0.2581 1 2.7778 929.03 10,000
0.0929 0.36 1 334.45 3600
2.778.4 1.076.3 2.990.3 1 10.7639
2.581.5 104 2.778.4 0.09290 1
The notation 2.581.5 signifies 2.581 x 105. 1 stoke = 1 cm2/s.
THERMOPHYSICAL PROPERTIES OF GASES Table 2.10 treats the specific heats, d y n a m i c viscosities, and t h e r m a l conductivities as functions of t e m p e r a t u r e only. To obtain the density of a gas, the perfect gas law m a y be used, i.e.,
P=pRT F r o m the specific heat and density and using o t h e r given properties, the t h e r m a l diffusivity and P r a n d t l n u m b e r m a y be calculated. For each gas, s k e l e t o n tables of the p r o p e r t i e s are given at several t e m p e r a t u r e s so that c o m p u t e r p r o g r a m checks can be made.
TABLE 2.10
Thermophysical Properties of Thirteen Common Gases Using Computer Equations
Air At/mol wt (kg/mol): 28.966 Gas constant (kJ/kg K): .287040 At/mol formula: (mixture)
Critical temperature (K): 132.6 Critical pressure (MPa): 3.77
cp = 2 [A(N)T N] A(0) A(1) A(2) A(3) A(4)
k
= 0.103409E+1 = 0.2848870E3 = 0.7816818E6 = 0.4970786E9 = 0.1077024E12
= S'.
[C(N)T N]
Temperature range: 250 < T < 1050 K Coefficients: C(0) = 2.276501E3 C(1) = 1.2598485E4 C(2) = 1.4815235E7 C(3) = 1.73550646E10
C(4) = 1.066657E13 C(5) = 2.47663035E17 C(6) = 0.0
~ = ~ [B(N)T N] Temperature range: 250 < T < 600 K
Temperature range: 600 < T < 1050 K
Coefficients: B(0) = 9.8601E1 B(1) = 9.080125E2 B(2) = 1.17635575E4 B(3) = 1.2349703E7
Coefficients: B(0) = 4.8856745 B(1) = 5.43232E2 B(2) = 2.4261775E5 B(3) = 7.9306E9
B(4) = 5.7971299E11 B(5) = 0.0 B(6) = 0.0
B(4) = 1.10398E12 B(5) = 0.0 B(6) = 0.0
Skeleton table T (K)
Cp (kJ/kg K)
~t (Ns/m 2) E6
k (W/m K) E3
300 500 1000
1.0064 1.0317 1.1415
18.53 26.82 41.77
26.07 39.48 67.21
Argon At/mol wt (kg/mol): 39.948 Gas constant (kJ/kg K): .208129 At/mol formula: Ar
Critical temperature (K): 150.8 Critical pressure (MPa): 4.87 Sat temp at one atmosphere (K): 87.5
c. = E [A(N)T ~]
k = y_ [C(N)T ~]
Temperature range: 200 < T < 1600 K
Temperature range: 200 < T < 1000 K
Coefficients: A(0) =0.52034 A(1) =0.0 A(2) =0.0 A(3) = 0.0
Coefficients: C( 0 ) = 5.2839462 E4 C(1) = 7.60706705E5 C(2) = 6.4749393E8 C(3) = 5.41874502E11
A(4) =0.0 A(5) =0.0 A(6) =0.0
C( 4 ) = 3.22024235 E 14 C(5) = 1.17962552E17 C(6) =1.86231745E21
~t= y [B(N)T ~] Temperature range: 200 < T < 540 K
Temperature range: 540 _< T < 1000 K
Coefficients: B(0) = 1.22573 B(1) = 5.9456964E2 B(2) = 1.897011E4 B(3) = 8.171242E7
Coefficients: B(0) = 4.03764 B(1) = 7.3665688E2 B(2) = 3.3867E5 B(3) = 1.127158E8
B(4) = 1.2939183E9 B(5) = 7.5027442E13 B(6) = 0.0
B(4) = 1.585569E12 B(5) =0.0 B(6) = 0.0
Skeleton table T (K)
Cp (kJ/kg K)
~t (Ns/m 2) E6
k (W/m K) E3
300 500 1000
0.5203 0.5203 0.5203
22.73 33.66 53.52
17.69 26.42 42.71
Extracted from Ref. 4 with permission. E2 signifies xl0 2, etc.
2.4
TABLE 2.10
Thermophysical Properties of Thirteen Common Gases Using Computer Equations
(Continued) nButane
At/mol wt (kg/mol): 58.124 Gas constant (kJ/kg K): .143044 At/mol formula: C4H10
Critical temperature (K): 408.1 Critical pressure (MPa): 3.65 Sat temp at one atmosphere (K): 261.5
cp= Y~[A(N)T N] Temperature range: 280 < T < 755 K
Temperature range: 755 < T < 1080 K
Coefficients: A(0) = 2.3665134E1 A(1) = 5.10573E3 A(2) = 4.16089E7 A(3) = 1.1450804E9
Coefficients: A(0) = 4.40126486 A(1) = 1.390866545E2 A(2) = 3.471109E5 A(3) = 3.45278E8
A(4) = 0.0 A(5) =0.0 A(6) = 0.0
A(4) = 1.619382E11 A(5) = 2.966666E15 A(6) = 0.0
k= ~ [C(N)T N]
B = ~ [B(N)T N] Temperature range: 270 < T < 520 K
Temperature range: 280 < T < 500 K
Coefficients: B(0) = 1.099487E2 B(1) = 2.634504E2 B(2) = 3.54700854E6 B(3) =0.0
Coefficients: C(0) = 3.79912E3 C(1) = 3.38011396E5 C(2) = 3.15886537E7 C(3) = 2.25600514E10
B(4) = 0.0 B(5) = 0.0 B(6) = 0.0
C(4) = 0.0 C(5) = 0.0 C(6) = 0.0
Skeleton table T (K)
Cp(kJ/kg K)
~t (Ns/m 2) E6
k (W/m K) E3
300 500 1000
1.700 2.542 3.903
7.573 12.27 
16.00 37.67
Carbon dioxide
At/mol wt (kg/mol): 44.01 Gas constant (kJ/kg K): .188919 At/mol formula: CO2
Critical temperature (K): 304.1 Critical pressure (MPa): 7.38 Sat temp at one atmosphere (K): 194.7
B= S'. [B(N)T N]
cp = E [A(N)T N] Temperature range: 200 ___T < 1000 K
Temperature range: 200 < T < 1000 K
Coefficients: A(0) = 4.5386462E1 A(1) = 1.5334795E3 A(2) = 4.195556E7 A(3) = 1.871946E9
Coefficients: B(0) = 8.095191E1 B(1) = 6.0395329E2 B(2) = 2.824853E5 B(3) = 9.843776E9
A(4) = 2.862388E12 A(5) =1.6962E15 A(6) = 3.717285E19
B(4) = 1.47315277E12 B(5) =0.0 B(6) = 0.0
k= ~'. [C(N)T N] Temperature range: 200 < T < 600 K
Temperature range: 600 < T < 1000 K
Coefficients: C(0) = 2.971488E3 C(1) =1.33471677E5 C(2) = 3.14443715E7 C(3) = 4.75106178E10
Coefficients: C(0) = 6.085375E2 C(1) = 3.63680275E4 C(2) = 1.0134366E6 C(3) = 9.7042356E10
C(4) = 2.68500151E13 C(5) =0.0 C(6) = 0.0
C(4) = 3.27864115E13 C(5) = 0.0 C(6) = 0.0
Skeleton table T (K)
Cp(kJ/kg K)
kt (Ns/m 2) E6
k (W/m K) E3
300 500 1000
0.845 1.013 1.234
15.02 23.46 39.71
16.61 32.30 68.05
Extracted from Ref. 4 with permission. E2 signifies X102, etc.
2.5
TABLE 2.10
Thermophysical Properties of Thirteen Common Gases Using Computer Equations
(Continued) Carbon monoxide At/mol wt (kg/mol): 28.011 Gas constant (kJ/kg K): .296828 At/mol formula: CO
Critical temperature (K): 132.9 Critical pressure (MPa): 3.5 Sat temp at one atmosphere (K): 81.6
c. = z [A(N)T"] Temperature range: 250 < T < 1050 K Coefficients: A(0) = 1.020802 A(1) = 3.82075E4 A(2) = 2.4945E6 A(3) = 6.81145E9
A(4) =7.93722E12 A(5) = 4.291972E15 A(6) = 8.903274E19
1~= ~ [B(N)T N]
k = ~ [C(N)T ~]
Temperature range: 250 < T _, J::l e~ 0
t,
¢q I,M ,I
,.t
7 7
2~
~
ci
p
o
6
e,i
o4
M
~
M
M
M
~
,.~
~
~
M
.1
M
~
,;
M
,.1
~
d
M
.,~
ei
•~
=~
•
~
.~
•~,~ ~
...¢, ~
~, ~ .~ ~, ~
, >
~,E
.~~ ~ ~
t'~
¢q
o O0
ii ~
I
I
I
.1
I
I
I
,1
,1
I
I
I
r.~
r.~
,.d
¢.~
~5 oo
.~=
2.19
v
'6
e,
.o
e.~ r', 0
#'1
ei Ill ..I
III
2.20
oj
¢xl
(,6
p.
t¢1
¢,q
¢xl
1..4
b
8
.1= h~
1=
.~.
,'4
r4
c5
~.
o
c5
o
o
,1
,1
,1
.1
,1
~
.~
~uu
,.1
,.1
•~
~
~.~
,.1
=
~ r.) ~.
~
o~
o
o
OJ
c4
..
o
.
e~g
~:~z~,~
N
~
I
7 7
8
c~
u~
'~~
0
c~
c~ v
f
c~
c~
c~
I
..
~
o
o
~
o
~o~7.
~
~
.~ ~ ~
~
o
t,,
o
o
o
o o
o o
t"I
o
t~ o
¢'1
o o
t'q
¢'q 0
o
~~
o
o
o
o
t"q
t"q o
o
0
0
o~
e'~
o~
~
cq
o
o
o
o
o
~o
00
o
0
0
o
o
e,i
~
,..t
o
o
o
o
c.i
r',i
o
o
o
o
~.
o
o,,
o
o
~ ,,.*
~
o4
~
" ~
. vi
~
o o
o
t'N
o
o o
~
o
~
~,~ ~.
"~
o
z
o
o
o
o
0o
~
e~
.~z
'~,
o
~
.
,..t
~
•
o
00
o

o
o o
o
r,i
0o
,...1
,,1
,,,1
1::1
¢1.
,..¢I
u~
~" I ~ ~=
:.E
__.Z
, ,,_; ~ ~ ~
2.21
2.22
0
,o
0
I.I.I ,..I
§ o~
oo
o
"T
c~
rd
p. p;
cq
t'N
t'q
,...,I
=
£
~
8 ~
cq
cd
...,I
~
,.1
,..1
,.1
~
~
~
,.I
,1
,1
;..E '>. "~
g
:E
~ EI
2.23
THERMOPHYSICAL PROPERTIES
TABLE 2.14
Fickian Diffusion Coefficients [(m2/s) × 104] at Atmospheric Pressure
Dij
T (K)
T (K)
Dij
T (K)
Dij
T (K)
Aircarbon dioxide[20]
Carbon dioxideargon [20]
Watercarbon dioxide [4]
Neonargon [15]
276.2 317.2
276.2 317.2
307.5 328.6 352.4
0.202 0.211 0.245
273.0 288.0 303.0 318.0
0.902 1.011 1.121
Neonneon [7]
0.1420 0.1772
0.1326 0.1652
Ammoniahelium[23]
Nitrogennitrogen [7]
274.2 308.2 331.1
77.5 194.5 273 298 353
0.668 0.783 0.881
Ammon~neon[23] 274.2 308.4 333.1
0.298 0.378 0.419
Ammon~xenon[23] 274.2 308.4 333.1
0.114 0.145 0.173
77.5 90 194.5 273 295 353
0.0134 0.0180 0.0830 0.156 0.178 0.249
Argonargon[12] 273 293 303 318
0.156 0.175 0.186 0.204
ArgonheHum[11] 287.9 354.0 418.0
0.697 0.979 1.398
Argonhelium[12] 273.0 288.0 303.0 318.0
0.640 0.701 0.760 0.825
Argonxenon[12] 273.0 288.0 303.0 318.0
0.0943 0.102 0.114 0.128
Argonxenon[13] 194.7 273.2 329.9 378.0
0.0508 0.0962 0.1366 0.1759
Carbon dioxideargon[25] 293
242.2 274.6 303.45 334.2
0.139
0.0854 0.1070 0.1301 0.1549
293.1 322.7 365.6 365.6 372.5
0.200
Hydrogen (trace)oxygen [2]
0.135 0.168 0.202 0.239
300 400 500 600 700 800 900
Oxygenargon [24] Oxygenargon [16] 243.2 274.7 304.5 334.0
Oxygenhelium [16] 244.2 274.0 304.4 334.0
0.536 0.640 0.761 0.912
Oxygenoxygen [7] 77.5 194.5 273 298 353
0.0153 0.104 0.187 0.232 0.301
Oxygenwater [4] 307.9 328.8 352.2
0.282 0.318 0.352
Oxygenxenon [16] 242.2 274.75 303.55 333.6
0.084 0.100 0.126 0.149
0.820 1.40 2.10 2.89 3.81 4.74 5.74
Hydrogenneon [10] 242.2 274.2 303.2 341.2
0.792 0.974 1.150 1.405
Hydrogenxenon [10] 242.2 274.2 303.9 341.2
0.410 0.508 0.612 0.751
Methanemethane[7] 90 194.5 273 298 353
0.244 0.357 0.377
298.2 353.6 382.6
Watercarbon dioxide [3]
Methanewater [4]
296.1 365.6 372.6
307.5 328.6 352.1
Not¢" See page 2.25 for footnotes and references.
0.850 1.012 1.24 1.26 1.28
0.0266 0.0992 0.206 0.240 0.318
Methanemethane [21]
Waterair [3] 289.9 365.6 372.5
Waterhelium [4] 307.2 328.5 352.5
Wate~hydrogen[3]
Nitrogenxenon [17]
293.2
Argonargon[7]
0.0168 0.104 0.185 0.212 0.287
0.164 0.249 0.259
0.235 0.315 0.360 0.292 0.331 0.356
77.5 194.5 273 298 353
Dij 0.276 0.300 0.327 0.357
0.0492 0.255 0.452 0.516 0.703
Neonxenon [14] 273.0 288.0 303.0 318.0
0.186 0.202 0.221 0.244
Nitrogenargon [17] 244.2 274.6 303.55 334.7
0.1348 0.1689 0.1999 0.2433
Nitrogenhelium [17] 243.2 275.0 303.55 332.5
0.477 0.596 0.719 0.811
He~umnitrogen (20% N2) [271 190 298 300 305 310 320 330 340 350 360 370 380 390 400
0.305 0.712 0.738 0.747 0.740 0.812 0.857 0.881 0.946 0.967 1.035 1.051 1.107 1.157
Heliumnitrogen (50% Ne) [271 190 298 300 305
0.310 0.725 0.751 0.758
2.24
CHAPTER TWO
TABLE 2.14
Fickian Diffusion Coefficients [(mZ/s) × 104] at Atmospheric Pressure (Continued)
T (K) Dij Heliumnitrogen (50% N2) [27] (Continued) 310 320 330 340 350 360 370 380 390 400
0.759 0.827 0.879 0.899 0.966 0.985 1.058 1.068 1.144 1.180
Heliumnitrogen (100%N2 extrapo~d) [27] 190 298 300 305 310 320 330 340 350 360 370 380 390 400
0.317 0.740 0.766 0.774 0.775 0.845 0.902 0.921 0.989 1.013 1.086 1.094 1.168 1.210
Heliumoxygen (trace) [18] 298 323 353 383 413 443 473 498
0.729 0.809 0.987 1.120 1.245 1.420 1.595 1.683 0.501 0.550 0.604 0.655
Hydrogenargon[lO] 242.2 274.2 303.9 341.2
0.562 0.698 0.830 1.010
Hydrogenargon[11] 287.9 354.2 418.0
295 448 628 806 958 1069
0.83 1.76 3.21 4.86 6.81 8.10
Heliumargon (trace) [18] 413 443 473 498
1.237 1.401 1.612 1.728
Helium (trace)argon [8] 300 400 500 600 700 800 900 1000 1100
0.76 1.26 1.86 2.56 3.35 4.23 5.20 6.25 7.38
Heliumcarbon dioxide [20] 276.2 317.2 346.2
0.5312 0.6607 0.7646
Heliumcarbon dioxide ~race) [18]
Heliumxenon[12] 273.0 288.0 303.0 318.0
T (K) Dij Hydrogen (trace)argon[9]
0.828 1.111 1.714
298 323 353 583 413 443 473 498
0.612 0.678 0.800 0.884 1.040 1.133 1.279 1.414
Heliummethyl alcohol (trace) [18] 423 443 463 483 503 523
1.032 1.135 1.218 1.335 1.389 1.475
Heliumneon [14] 273.0 288.0 303.0 318.0
0.906 0.986 1.065 1.158
T (K)
Dij
Heliumnitrogen ~race) [18] 298 323 353 383 413 443 473 498
0.687 0.766 0.893 1.077 1.200 1.289 1.569 1.650
0.743 1.21 1.76 2.40 3.11 3.90 4.76 5.69 7.74
Carbon dioxidenitrogen (trace) [1] 300 400 500 600 700 800 900 1000 1100
0.177 0.300 0.445 0.610 0.798 0.998 1.22 1.47 1.70
Carbon dioxidenitrogen[26] 295 1156 1158 1286 1333 1426 1430 1469 1490 1653
0.159 1.78 1.92 2.34 2.26 2.55 2.72 2.85 2.92 3.32
Carbon dioxidenitrous oxide [19] 194.8 273.2 312.8 362.6
300 400 500 600 700 800 900 1000
0.160 0.270 0.400 0.565 0.740 0.928 1.14 1.39
Carbon monoxidecarbon monoxide [22]
Helium (trace)nitrogen [1] 300 400 500 600 700 800 900 1000 1200
T (K) Dij Carbon dioxideoxygen (trace) [2]
194.7 273.2 319.6 373.0
0.109 0.190 0.247 0.323
Carbon monoxidenitrogen [22] 194.7 273.2 319.6 373.0
0.105 0.186 0.242 0.318
Carbon monoxide (trace)oxygen [2] 300 400 500 600 700 800
0.212 0.376 0.552 0.746 0.961 1.22
Heliumair [20] 276.2 317.2 346.2
0.6242 0.7652 0.9019
Heliumargon [20] 276.2 317.2 346.2
0.6460 0.7968 0.9244
Heliumargon (trace) [18] 298 323 353 383
0.729 0.809 0.978 1.122
Carbon dioxideargon [26] 0.0531 0.0996 0.1280 0.1683
295 1181 1207 1315
0.139 1.88 1.88 2.38
THERMOPHYSICAL PROPERTIES
TABLE 2 . 1 4
F i c k i a n D i f f u s i o n Coefficients [(m2/s) × 10 4] at A t m o s p h e r i c P r e s s u r e (Continued)
Dq
T(K)
T(K)
Carbon dioxideargon [26] (Continued)
Carbon dioxidecarbon dioxide[19]
1368 1383 1427 1445 1495 1503 1538 1676
194.8 273.2 312.8 362.6
2.59 2.13 2.53 2.66 2.65 2.84 3.08 3.21
Carbon dioxidecarbon dioxide [7] 194.7 273 298 353
2.25
0.0500 0.0907 0.113 0.153
Dij
T(K)
Dij
0.0516 0.0970 0.1248 0.1644
Carbon dioxidecarbon dioxide [5] 233 253
0.0662 0.0794
274 293 313 333 363 393
0.0925 0.1087 0.1239 0.1395 0.1613 0.1876
423 453 483
0.2164 0.2477 0.2892
All the Dij values are in (m2/s) x 10~. For example, at 276.2 K the interdiffusion coefficient for the aircarbon dioxide mixture is 1.420 x 105 m2/s. For an extensive review with formula fits but no data tables, see Marrero and Mason, J. Phys. Chem. Ref. Data, 1:3118 (1972). Interpolation from a graph of log D;j versus log T is often simple. References for Fickian interdiffusion coefficients 1. R. E. Walker and A. A. Westenberg, "Molecular Diffusion Studies in Gases at High Temperatures. II. Interpretation of Results on the HeN2 and CO2N2 Systems," J. Chem. Phys., 29:1147,1958. 2. R. E. Walker and A. A. Westenberg, "Molecular Diffusion Studies in Gases at High Temperatures. IV. Results and Interpretation of the CO2O2, CH4O2, H2O2, COO 2 and H 2 0  O 2 Systems," J. Chem. Phys, 32:436,1960. 3. M. Trautz and W. MUller, "Die Reibung, W~irmeleitung und Diffusion in Gasmischungen. XXXIII. Die Korrektion der bisher mit der Verdampfungsmethode gemessenen Diffusionskonstanten," Ann. Physik, 22:333,1935. 4. E A. Schwertz and J. E. Brow, "Diffusivity of Water Vapor in Some Common Gases," J. Chem. Phys., 19:.640, 1951. 5. K. Sch~ifer and P. Reinhard, "Zwischenmolekulare Kr~ifte und die Temperaturabh~ingigkeit der Selbstdiffusion von CO2," Z. Naturforsch, 18:187,1963. 6. G. Ember, J. R. Ferron, and K. Wohl, "SelfDiffusion Coefficients of Carbon Dioxide at 1180°1680°K," J. Chem. Phys., 37:891,1962. 7. E. B. Winn, "The Temperature Dependence of the SelfDiffusion Coefficients of Argon, Neon, Nitrogen, Oxygen, Carbon Dioxide, and Methane," Phys. Rev., 80:.1024, 1950. 8. R. E. Walker and A. A. Westenberg, "Molecular Diffusion Studies in Gases at High Temperature. III. Results and Interpretation of the HeA System," J. Chem. Phys., 31:319, 1959. 9. A. A. Westenberg and G. Frazier, "Molecular Diffusion Studies in Gases at High Temperature. V. Results for the H2Ar System," J. Chem. Phys., 36:3499,1962. 10. R. Paul and I. B. Srivastava, "Mutual Diffusion of the Gas Pairs HENe, HEAr, and HEXe at Different Temperatures," J. Chem. Phys., 35:1621,1961. 11. R. A. Strehlow, "The Temperature Dependence of the Mutual Diffusion Coefficient for Four Gaseous Systems," J. Chem. Phys., 21:2101,1953. 12. K. E Srivastava, "Mutual Diffusion of Binary Mixtures of
T(K)
Dij
Carbon dioxidenitrogen [24]
Carbon dioxidecarbon dioxide[6] 296 298 1180
0.109 0.109 1.73 1.84
1218 1330 1445
2.04 2.38 2.80
1450 1487 1490 1520 1576
2.86 2.56 2.88 2.98 2.78 3.12
1580 1665 1680
3.33 3.29 3.50
289
0.158
Waterhydrogen [4] 307.3 328.6 352.7
1.020 1.121 1.200
Waternitrogen [4] 307.6 328.6 352.2
0.256 0.303 0.359
X e n o n  x e n o n [13] 194.7 273.2 293.0 300.5 329.9 378.0
0.0257 0.0480 0.0443 0.0576 0.0684 0.0900
Helium, Argon, and Xenon at Different Temperatures," Physica, 25:571, 1959. 13. I. Amdur and T. E Schatzki, "Diffusion Coefficients of the Systems XeXe and AXe," J. Chem. Phys., 27:1049,1957. 14. K. P. Srivastava and A. K. Barua, "The Temperature Dependence of Interdiffusion Coefficient for Some Pairs of Rare Gases," Indian J. Phys., 33:229,1959. 15. B. N. Srivastava and K. P. Srivastava, "Mutual Diffusion of Pairs of Rare Gases at Different Temperatures," J. Chem. Phys., 30:.984,1959. 16. R. Paul and I. B. Srivastava, "Studies on Binary Diffusion of the Gas Pairs O2A, O2Xe, and O:He," Indian J. Phys., 35:465,1961. 17. R. Paul and I. B. Srivastava, "Studies on the Binary Diffusion of the Gas Pairs N2A, N2Xe, and N2He," Indian J. Phys., 35:523,1961. 18. S. L. Seager, L. R. Geertson, and J. C. Giddings, "Temperature Dependence of Gas and Vapor Diffusion Coefficients," J. Chem. Eng. Data, 8:168, 1963. 19. I. Amdur, J. W. Irvine, Jr., E. A. Mason, and J. Ross, "Diffusion Coefficients of the Systems CO2CO2 and CO2N20," J. Chem. Phy&, 20:436, 1952. 20. J. N. Holsen and M. R. Strunk, "Binary Diffusion Coefficients in Nonpolar Gases," Ind. Eng. Chem. Fund., 3:143,1964. 21. C. R. Mueller and R. W. Cahill, "Mass Spectrometric Measurement of Diffusion Coefficients," J. Chem. Phys, 40:651,1964. 22. I. Amdur and L. M. Shuler, "Diffusion Coefficients of the Systems COCO and CON2," J. Chem. Phys., 38:188,1963. 23. I. B. Srivastava, "Mutual Diffusion of Binary Mixtures of Ammonia with He, Ne and Xe," Indian J. Phys., 36:193, 1962. 24. L. E. Boardman and N. E. Wild, "The Diffusion of Pairs of Gases with Molecules of Equal Mass," Proc. Royal Soc. A162:511,1937. 25. L. Waldmann, "Die Temperaturerscheinungen bei der Diffusion in ruhenden Gasen und ihre messtechnische Anwendung," Z. Phys., 124:2, 1947. 26. T. A. Pakurar and J. R. Ferron, "Measurement and Prediction of Diffusivities to 1700°K in Binary Systems Containing Carbon Dioxide," Univ. of Delaware Tech. Rept. DEL14P, 1964. 27. J.W. Yang, "A New Method of Measuring the Mass Diffusion Coefficient and Thermal Diffusion Factor in a Binary Gas System," doctoral dissertation, Univ. of Minnesota, 1966. 28. R. E. Walker, L. Monchick, A. A. Westenberg, and S. Favin, "High Temperature Gaseous Diffusion Experiments and Intermolecular Potential Energy Functions," Planet. Space Sci., 3:221,1961.
2.26
CHAPTER TWO
THERMOPHYSICAL PROPERTIES OF LIQUIDS
TABLE 2.15
Thermophysical Properties of Saturated Water and Steam Liquid
T (°C)
11" 107
~103
0 10 20 30 40 50 60 70 80 90
17 525 12 992 10 015 7 970 6 513 5 440 4 630 4 005 3 510 3 113
569 586 602 617 630 643 653 662 669 675
100 110 120 130 140 150 160 170 180 190
2 790 2 522 2 300 2 110 1 950 1 810 1 690 1 585 1 493 1 412
200 210 220 230 240 250 260 270 280 290 300 310 320 330 340 350 360 370
Steam c,
Cp
Pr
17.6 18.2 18.8 19.4 20.1 20.9 21.6 22.3 23.1 23.9
1.864 1.868 1.874 1.883 1.894 1.907 1.924 1.944 1.969 1.999
0.85 0.87 0.88 0.90 0.91 0.92 0.94 0.95 0.96 0.98
121 124 128 132 135 139 142 146 149 153
24.8 25.8 26.7 27.8 28.8 30.0 31.3 32.6 34.1 35.7
2.034 2.075 2.124 2.180 2.245 2.320 2.406 2.504 2.615 2.741
0.99 1.00 1.02 1.04 1.05 1.08 1.09 1.12 1.14 1.17
0.91 0.88 0.86 0.85 0.85 0.84 0.85 0.86 0.89 0.92
156 160 163 167 171 174 178 182 187 193
37.5 39.4 41.5 43.9 46.5 49.5 52.8 56.6 60.9 66.0
2.883 3.043 3.223 3.426 3.656 3.918 4.221 4.574 4.996 5.51
1.20 1.24 1.27 1.30 1.34 1.38 1.42 1.47 1.53 1.61
0.96 1.01 1.09 1.19 1.34 1.62 2.41 8.99
198 205 214 225 238 256 282 335
71.9 79.1 87.8 98.9 113 130 150 183
Pr
TI • 107
4.217 4.193 4.182 4.179 4.179 4.181 4.185 4.190 4.197 4.205
12.99 9.30 6.96 5.40 4.32 3.54 2.97 2.54 2.20 1.94
80.4 84.5 88.5 92.6 96.6 100 105 109 113 117
680 683 685 687 687 686 684 681 676 671
4.216 4.229 4.245 4.263 4.285 4.310 4.339 4.371 4.408 4.449
1.73 1.56 1.43 1.31 1.22 1.14 1.07 1.02 0.97 0.94
1 338 1 273 1 215 1 162 1 114 1 070 1 030 994 961 930
664 657 648 639 629 617 604 589 573 557
4.497 4.551 4.614 4.686 4.770 4.869 4.985 5.13 5.30 5.51
901 865 830 790 748 700 644 564
540 522 503 482 460 435 401 338
5.77 6.12 6.59 7.25 8.27 10.08 14.99 53.9
~103
6.14 6.96 8.05 9.59 11.92 15.95 26.79 112.9
Viscosity rl (Ns/m2), thermal conductivity ~, (W/m.deg), heat capacity Cp(kJ/kg.deg), Prandtl number Pr. Ref. 2 with permission.
Source"
1.69 1.80 1.96 2.18 2.51 3.14 5.04 20.66
o "0
0
0J
0
I,M ..I
~D
C~
C~ O0 C~
O0 0'~ C~ O0
~
C4 C~ ~
V.~ C~I ,~. (~r~
T
I
c~
"T
T
c~
"T
oo
T
×
"T
o
2.27
2.28
CHAPTER TWO
TABLE 2.17
Isobaric Specific Heat for Water and Steam at Various Temperatures and Pressures Pressure, bar
T(°C)
0.1
1
10
20
40
60
80
100
0 50 100 120 140 160 180 200 220 240 260 280 300 320 340 350 360 365 370 375 380 385 390 395 400 405 410 415 420 425 430 440 450 460 480 500 520 540 560 580 600 620 640 660 680 700 800
4.218 1.929 1.910 1.913 1.918 1.926 1.933 1.944 1.954 1.964 1.976 1.987 1.999 2.011 2.024 2.030 2.037 2.040 2.043 2.046 2.049 2.052 2.056 2.059 2.062 2.066 2.069 2.072 2.076 2.079 2.082 2.089 2.095 2.102 2.116 2.129 2.142 2.156 2.170 2.184 2.198 2.212 2.226 2.240 2.254 2.268 2.339
4.217 4.181 2.038 2.007 1.984 1.977 1.974 1.975 1.979 1.985 1.993 2.001 2.010 2.021 2.032 2.038 2.044 2.048 2.050 2.053 2.056 2.059 2.061 2.065 2.068 2.071 2.074 2.077 2.080 2.083 2.086 2.093 2.099 2.106 2.119 2.132 2.146 2.159 2.173 2.187 2.200 2.213 2.227 2.241 2.255 2.270 2.341
4.212 4.179 4.214 4.243 4.283 4.337 2.613 2.433 2.316 2.242 2.194 2.163 2.141 2.126 2.122 2.125 2.127 2.128 2.128 2.127 2.127 2.126 2.125 2.125 2.126 2.127 2.128 2.129 2.131 2.132 2.134 2.138 2.141 2.146 2.154 2.164 2.175 2.185 2.197 2.208 2.219 2.230 2.243 2.256 2.270 2.283 2.352
4.207 4.176 4.211 4.240 4.280 4.334 4.403 4.494 2.939 2.674 2.505 2.395 2.321 2.268 2.239 2.235 2.231 2.227 2.222 2.218 2.212 2.207 2.202 2.200 2.197 2.195 2.193 2.192 2.192 2.190 2.190 2.190 2.191 2.192 2.196 2.201 2.208 2.216 2.226 2.233 2.240 2.250 2.260 2.272 2.286 2.299 2.364
4.196 4.172 4.207 4.235 4.275 4.327 4.395 4.483 4.601 4.763 3.582 3.116 2.834 2.649 2.536 2.504 2.478 2.462 2.446 2.428 2.412 2.396 2.381 2.369 2.358 2.349 2.340 2.334 2.327 2.321 2.316 2.307 2.300 2.294 2.286 2.281 2.280 2.280 2.285 2.285 2.287 2.291 2.298 2.307 2.317 2.330 2.389
4.186 4.167 4.202 4.230 4.269 4.320 4.386 4.472 4.586 4.741 4.964 4.514 3.679 3.217 2.943 2.861 2.793 2.759 2.725 2.690 2.657 2.627 2.600 2.575 2.553 2.534 2.517 2.501 2.487 2.474 2.462 2.441 2.424 2.409 2.385 2.368 2.357 2.349 2.349 2.342 2.336 2.334 2.337 2.343 2.352 2.362 2.414
4.176 4.163 4.198 4.226 4.263 4.313 4.378 4.461 4.571 4.720 4.932 5.25 5.31 4.118 3.526 3.350 3.216 3.134 3.072 3.018 2.964 2.913 2.867 2.826 2.789 2.756 2.727 2.700 2.675 2.653 2.632 2.596 2.565 2.538 2.496 2.464 3.441 2.423 2.416 2.401 2.389 2.381 2.379 2.381 2.388 2.398 2.440
4.165 4.158 4.194 4.221 4.258 4.307 4.370 4.450 4.557 4.700 4.902 5.20 5.70 5.79 4.412 4.043 3.769 3.655 3.546 3.446 3.356 3.274 3.201 3.137 3.078 3.025 2.979 2.936 2.898 2.863 2.830 2.773 2.726 2.684 2.618 2.569 2.531 2.502 2.487 2.465 2.445 2.431 2.423 2.421 2.424 2.429 2.465
THERMOPHYSICAL PROPERTIES TABLE 2.17
2.29
Isobaric Specific Heat for Water and Steam at Various Temperatures and Pressures
(Continued) Pressure, bar T (°C)
150
175
200
210
220
225
230
240
0 50 100 120 140 160 180 200 220 240 260 280 300 320 340 350 360 365 370 375 380 385 390 395 400 405 410 415 420 425 430 440 450 460 480 500 520 540 560 580 600 620 640 660 680 700 800
4.141 4.148 4.183 4.209 4.245 4.291 4.350 4.425 4.523 4.653 4.832 5.09 5.50 6.23 8.14 8.68 6.86 6.15 5.69 5.33 5.02 4.750 4.520 4.325 4.155 4.007 3.879 3.764 3.664 3.573 3.491 3.350 3.235 3.138 2.986 2.875 2.791 2.726 2.683 2.638 2.598 2.566 2.542 2.528 2.520 2.518 2.531
4.129 4.142 4.178 4.204 4.238 4.283 4.340 4.413 4.508 4.632 4.801 5.04 5.41 6.05 7.45 9.27 12.57 9.84 8.36 7.40 6.68 6.13 5.68 5.32 5.02 4.770 4.556 4.371 4.211 4.069 4.945 3.734 3.564 3.424 3.210 3.056 2.940 2.852 2.791 2.733 2.682 2.640 2.607 2.585 2.572 2.565 2.564
4.117 4.137 4.173 4.198 4.232 4.276 4.331 4.402 4.492 4.611 4.772 4.997 5.33 5.89 7.01 9.10 11.37 19.72 18.38 12.71 10.19 8.68 7.65 6.90 6.33 5.87 5.50 5.19 4.933 4.711 4.520 4.205 3.959 3.761 3.465 3.257 3.104 2.989 2.906 2.833 2.770 2.717 2.675 2.644 2.625 2.613 2.598
4.113 4.135 4.171 4.196 4.229 4.273 4.328 4.397 4.486 4.603 4.760 4.979 5.31 5.84 6.87 7.81 10.18 13.77 75.67 19.03 13.14 10.49 8.90 7.83 7.06 6.46 5.99 5.61 5.29 5.02 4.795 4.424 4.139 3.912 3.576 3.343 3.174 3.046 2.954 2.875 2.807 2.709 2.703 2.669 2.646 2.632 2.611
4.108 4.133 4.169 4.194 4.227 4.270 4.324 4.393 4.481 4.595 4.749 4.963 5.28 5.79 6.74 7.56 9.40 11.62 18.38 52.7 19.19 13.38 10.68 9.06 7.97 7.18 6.57 6.09 5.70 5.37 5.10 4.664 4.333 4.074 3.695 3.434 3.247 3.106 3.003 2.918 2.844 2.781 2.731 2.694 2.669 2.652 2.625
4.106 4.132 4.168 4.193 4.226 4.268 4.322 4.390 4.478 4.591 4.744 4.955 5.26 5.76 6.68 7.45 9.10 10.94 15.56 81.49 25.71 15.62 11.88 9.84 8.53 7.60 6.90 6.36 5.92 5.56 5.26 4.791 4.435 4.159 3.756 3.481 3.284 3.136 3.028 2.939 2.863 2.798 2.746 2.707 2.680 2.662 2.632
4.103 4.131 4.167 4.192 4.224 4.267 4.320 4.388 4.475 4.588 4.738 4.947 5.25 5.74 6.63 7.35 8.84 10.40 13.84 29.52 40.95 18.88 13.42 10.77 9.16 8.06 7.26 6.65 6.16 5.77 5.44 4.927 4.544 4.247 3.819 3.529 3.322 3.167 3.054 2.961 2.882 2.814 2.760 2.719 2.691 2.672 2.639
4.099 4.129 4.165 4.189 4.222 4.264 4.317 4.384 4.469 4.580 4.728 4.931 5.23 5.69 6.53 7.17 8.41 9.58 11.79 17.44 68.4 33.4 18.21 13.29 10.76 9.20 8.12 7.32 6.71 6.22 5.83 5.22 4.77 4.43 3.95 3.63 3.40 3.23 3.10 3.01 2.92 2.85 2.79 2.75 2.71 2.69 2.65
2.30
CHAPTER TWO
TABLE 2.17
Isobaric Specific Heat for Water and Steam at Various Temperatures and Pressures
(Continued) Pressure, bar T (°C)
250
270
300
400
500
600
800
1000
0 50 100 120 140 160 180 200 220 240 260 280 300 320 340 350 360 365 370 375 380 385 390 395 400 405 410 415 420 425 430 440 450 460 480 500 520 540 560 580 600 620 640 660 680 700 800
4.095 4.127 4.163 4.187 4.220 4.261 4.313 4.379 4.464 4.572 4.717 4.916 5.20 5.65 6.43 7.02 8.07 8.99 10.56 13.76 23.37 73.1 28.04 17.31 13.02 10.67 9.17 8.12 7.35 6.74 6.26 5.54 5.02 4.631 4.089 3.731 3.481 3.295 3.158 3.051 2.960 2.882 2.819 2.771 2.736 2.713 2.666
4.086 4.123 4.159 4.183 4.215 4.255 4.306 4.371 4.452 4.558 4.697 4.886 5.16 5.57 6.27 6.76 7.56 8.18 9.12 10.67 13.51 20.07 38.02 33.71 21.11 15.32 12.22 10.30 8.99 8.04 7.32 6.28 5.58 5.08 4.389 3.951 3.650 3.431 3.268 3.144 3.040 2.952 2.880 2.824 2.783 2.755 2.694
4.073 4.117 1.153 4.177 4.208 4.247 4.296 4.358 4.437 4.537 4.669 4.845 5.09 5.46 6.07 6.45 7.03 7.43 7.98 8.76 9.90 11.68 14.60 19.68 25.71 24.85 19.59 15.45 12.70 10.83 9.49 7.73 6.62 5.87 4.902 4.316 3.926 3.650 3.442 3.290 3.165 3.060 2.974 2.906 2.855 2.819 2.736
4.032 4.098 4.135 4.156 4.185 4.220 4.265 4.319 4.388 4.474 4.584 4.728 4.920 5.19 5.60 5.81 6.10 6.27 6.48 6.70 6.97 7.30 7.71 8.19 8.78 9.47 10.25 11.12 12.00 12.73 13.13 12.54 10.89 9.28 7.08 5.81 5.02 4.487 4.095 3.823 3.614 3.446 3.308 3.197 3.110 3.044 2.879
3.993 4.080 4.117 4.137 4.163 4.196 4.235 4.284 4.344 4.419 4.514 4.633 4.788 4.996 5.30 5.45 5.64 5.73 5.84 5.96 6.10 6.26 6.43 6.61 6.81 7.04 7.29 7.57 7.87 8.18 8.50 9.08 9.48 9.52 8.55 7.20 6.13 5.37 4.796 4.387 4.082 3.845 3.654 3.500 3.376 3.279 3.024
3.956 4.064 4.100 4.119 4.143 4.172 4.208 4.252 4.305 4.371 4.453 4.555 4.683 4.848 5.08 5.20 5.34 5.40 5.47 5.56 5.65 5.75 5.84 5.94 6.05 6.16 6.27 6.40 6.54 6.69 6.84 7.17 7.47 7.71 7.87 7.48 6.76 6.03 5.38 5.890 4.510 4.216 3.981 3.791 3.637 3.513 3.168
3.882 4.035 4.068 4.085 4.105 4.130 4.159 4.195 4.237 4.290 4.354 4.432 4.524 4.633 4.766 4.871 4.954 4.987 5.03 5.08 5.14 5.20 5.25 5.30 5.34 5.38 5.42 5.46 5.51 5.56 5.61 5.72 5.84 5.97 6.19 6.31 6.28 6.10 5.75 5.39 5.03 4.724 4.465 4.249 4.068 3.916 3.441
3.800 4.010 4.039 4.054 4.071 4.092 4.116 4.145 4.180 4.223 4.276 4.340 4.411 4.485 4.552 4.663 4.719 4.737 4.764 4.802 4.843 4.884 4.919 4.949 4.974 4.996 5.02 5.04 5.06 5.08 5.10 5.15 5.20 5.26 5.40 5.51 5.58 5.56 5.43 5.28 5.08 4.871 4.669 4.485 4.322 4.178 3.669
Source:
Ref. 2 with permission.
THERMOPHYSICAL PROPERTIES
2.31
TABLE 2.18 Dynamic Viscosity [11 • 107 (N's/m2)] of Water and Steam at Various Temperatures and Pressures
Pressure, bar r (°C)
1
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250 260 270 280 290 300 310 320 330 340 350 360 370 380 390 400 410 420 430 440 450 460 470 480 490
17,525 12,992 10,015 7,971 6,513 5,441 4,630 4,004 3,509 3,113 121 125 129 133 137 141 146 150 154 158 162 166 170 174 178 182 186 190 194 198 202 207 211 215 219 223 227 231 235 239 243 247 251 255 260 264 268 272 276 280
20 17,514 12,986 10,013 7,970 6,514 5,443 4,633 4,007 3,513 3,116 2,793 2,526 2,303 2,114 1,953 1,814 1,693 1,588 1,495 1,413 1,339 1,275 164 169 174 179 183 188 193 197 202 206 211 216 220 225 229 233 238 242 246 250 254 258 262 266 270 274 278 282
40 17,502 12,980 10,010 7,970 6,515 5,445 4,636 4,010 3,516 3,120 2,797 2,530 2,307 2,118 1,957 1,818 1,698 1,592 1,500 1,417 1,343 1,278 1,218 1,164 1,115 1,070 180 185 191 196 201 206 211 216 222 227 231 236 240 244 248 252 256 260 264 268 272 276 280 284
60 17,491 12,975 10,008 7,970 6,516 5,447 6,638 4,013 3,520 3,124 2,801 2,534 2,311 2,123 1,962 1,823 1,702 1,597 1,504 1,422 1,348 1,282 1,223 1,169 1,120 1,075 1,033 995 189 194 200 206 212 218 224 229 234 239 243 247 251 255 259 263 267 270 274 278 282 286
80 17,480 12,969 10,005 7,970 6,517 5,449 4,641 4,016 3,523 3,128 2,805 2,538 2,315 2,127 1,966 1,827 1,707 1,601 1,509 1,426 1,353 1,287 1,228 1,174 1,125 1,080 1,039 1,000 964 931 199 206 212 219 226 232 237 243 246 250 254 258 262 266 269 273 277 281 285 289
100 17,468 12,963 10,003 7,969 6,519 5,451 4,644 4,019 3,527 3,131 2,809 2,542 2,319 2,131 1,970 1,832 1,711 1,606 1,513 1,431 1,358 1,292 1,232 1,179 1,129 1,084 1,043 1,005 969 936 904 866 213 221 229 236 241 246 250 254 258 261 265 269 272 276 280 284 288 291
150 17,439 12,948 9,997 7,968 6,521 5,456 4,650 4,027 3,535 3,141 2,819 2,552 2,330 2,142 1,981 1,843 1,722 1,617 1,525 1,442 1,369 1,303 1,244 1,190 1,141 1,096 1,055 1,017 981 948 917 881 843 800 749 248 255 259 263 266 268 272 275 278 281 285 288 292 295 299
200 17,411 12,934 9,991 7,968 6,524 5,461 4,657 4,036 3,544 3,150 2,828 2,563 2,340 2,152 1,992 1,854 1,734 1,628 1,536 1,454 1,381 1,315 1,256 1,202 1,153 1,108 1,067 1,029 993 960 929 895 859 820 777 727 661 298 288 286 286 287 288 290 293 296 298 301 304 308
210 17,405 12,931 9,990 7,968 6,525 5,462 4,658 4,038 3,546 3,152 2,830 2,565 2,342 2,154 1,994 1,856 1,736 1,631 1,538 1,456 1,383 1,317 1,258 1,204 1,156 1,111 1,069 1,031 996 963 932 898 862 824 782 734 673 335 297 292 290 291 292 294 296 298 301 304 307 310
2.32
CHAPTERTWO TABLE 2.18 Dynamic Viscosity [11 • 107 (N's/m2)] of Water and Steam at Various Temperatures and Pressures (Continued) Pressure, bar T (°C) 500 520 540 560 580 600 620 640 660 680 700
1 284 292 300 308 316 325 333 341 349 357 365
T (°C)
220
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250 260 270 280 290 300 310 320 330 340 350 360 370
17,399 12,928 9,988 7,967 6,225 5,463 4,660 4,038 3,548 3,154 2,832 2,567 2,344 2,157 1,996 1,858 1,738 1,633 1,540 1,458 1,385 1,320 1,261 1,207 1,158 1,113 1,072 1,034 998 965 934 901 865 827 786 740 683 596
20 286 294 302 310 318 326 334 342 351 359 367 230 17,394 12,925 9,987 7,967 6,526 5,464 4,661 4,040 3,549 3,155 2,834 2,569 2,347 2,159 1,998 1,860 1,740 1,635 1,543 1,461 1,388 1,322 1,263 1,209 1,160 1,116 1,074 1,036 1,001 968 937 904 868 831 790 745 692 617
40 288 296 304 312 320 328 336 344 352 360 368 240 17,388 12,922 9,986 7,967 6,526 5,465 4,662 4,041 3,551 3,157 2,836 2,571 2,349 2,161 2,000 1,862 1,742 1,637 1,545 1,463 1,390 1,324 1,265 1,212 1,163 1,118 1,077 1,038 1,003 970 939 906 871 834 794 751 700 633
60 290 298 306 314 322 330 338 346 354 362 370 250 17,382 12,919 9,985 7,967 6,527 5,466 4,664 4,043 3,553 3,159 2,838 2,573 2,351 2,163 2,003 1,865 1,745 1,640 1,547 1,465 1,392 1,327 1,268 1,214 1,165 1,120 1,079 1,041 1,006 972 941 909 874 837 798 756 707 646
80 293 301 308 316 324 332 340 348 356 364 372 300 17,353 12,905 9,979 7,966 6,529 5,471 4,670 4,051 3,561 3,168 2,848 2,583 2,361 2,174 2,013 1,876 1,756 1,651 1,559 1,477 1,404 1,338 1,279 1,226 1,177 1,132 1,091 1,053 1,018 985 954 922 888 853 817 779 738 692
100 295 303 311 319 326 334 342 350 358 366 374 400 17,296 12,875 9,967 7,965 6,535 5,481 4,684 4,066 3,579 3,187 2,867 2,603 2,382 2,195 2,035 1,898 1,778 1,674 1,581 1,500 1,427 1,362 1,303 1,249 1,201 1,156 1,115 1,077 1,042 1,009 978 948 915 881 848 815 781 746
150 302 310 317 325 332 340 348 355 363 371 378 500 17,239 12,846 9,954 7,963 6,540 5,491 4,697 4,082 3,596 3,206 2,887 2,623 2,403 2,216 2,057 1,920 1,800 1,696 1,604 1,523 1,450 1,385 1,326 1,273 1,225 1,180 1,140 1,102 1,067 1,034 1,004 972 940 908 876 845 814 784
200 311 318 324 332 339 346 353 361 368 376 384 600 17,182 12,817 9,942 7,962 6,546 5,502 4,711 4,098 3,614 3,224 2,906 2,644 2,424 2,237 2,078 1,941 1,822 1,718 1,627 1,546 1,473 1,408 1,350 1,297 1,248 1,204 1,164 1,126 1,091 1,059 1,028 997 964 932 901 871 842 813
210 313 320 326 333 340 347 355 362 370 377 385 800 17,067 12,759 9,918 7,959 6,557 5,522 4,737 4,129 3,648 3,261 2,945 2,684 2,465 2,280 2,122 1,985 1,867 1,763 1,672 1,591 1,519 1,455 1,397 1,344 1,296 1,252 1,212 1,175 1,140 1,108 1,078 1,045 1,012 980 949 920 891 864
THERMOPHYSICAL PROPERTIES
2.33
TABLE 2.18 Dynamic Viscosity [rl • 107 (N's/m2)] of Water and Steam at Various Temperatures and Pressures (Continued) Pressure, bar T (°C) 380 390 400 410 420 430 440 450 460 470 480 490 500 520 540 560 580 600 620 640 660 680 700
220 311 300 296 295 296 297 299 301 303 306 309 312 315 321 328 335 342 349 356 363 371 378 386
230
240
340 310 303 300 300 300 302 304 306 308 311 314 317 323 330 336 343 350 357 365 372 379 387
250
468 324 311 306 304 304 305 307 309 311 313 316 319 325 331 338 345 352 359 366 373 380 388
300
537 348 321 313 310 309 309 310 312 314 316 318 321 327 333 340 346 353 360 367 374 382 389
400
630 561 458 380 352 340 334 331 330 330 331 332 334 338 343 348 354 361 367 374 381 388 395
500
703 667 627 580 529 479 438 411 394 383 376 371 369 367 368 370 374 379 384 389 395 401 408
600
748 721 692 660 626 591 555 521 495 466 446 432 421 408 402 399 399 401 404 408 412 418 422
783 759 735 710 683 656 628 599 572 546 522 502 485 460 444 435 430 428 428 429 432 435 439
800 840 817 797 777 758 737 716 695 674 654 633 614 596 563 537 516 502 491 484 480 477 477 478
Source: Ref. 2 with permission.
TABLE 2.19 Thermal Conductivity [~.. 103 (W/m.deg)] of Water and Steam at Various Temperatures and Pressures Pressure, bar T (°C)
1
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170
569 588 603 617 630 643 653 662 669 675 24.5 25.2 26.0 26.9 27.7 28.6 29.5 30.4
20
40
60
80
570 589 605 620 633 645 655 664 671 677 682 686 688 689 689 688 685 682
572 590 607 622 635 647 657 665 673 679 684 687 689 690 690 689 687 683
574 592 608 623 637 648 658 667 674 680 685 688 691 692 692 690 688 685
575 594 610 625 638 650 660 668 676 682 686 690 692 693 693 692 690 686
100 577 595 612 627 640 651 661 670 677 683 688 691 693 694 694 693 691 688
150 581 599 616 631 644 655 665 674 681 687 691 694 697 698 698 696 694 691
200 585 603 620 634 648 659 669 677 684 690 694 698 700 701 701 700 698 695
2.34
CHAPTER TWO
TABLE 2.19 Thermal Conductivity [k. 103 (W/m.deg)] of Water and Steam at Various Temperatures and Pressures (Continued) Pressure, bar T (°C)
20
40
60
80
100
150
200
180 190 200 210 220 230 240 250 260 270 280 290 300 310 320 330 340 350 360 370 380 390 400 410 420 430 440 450 460 470 480 490 500 520 540 560 580 600 620 640 660 680 700
31.3 32.2 33.1 34.1 35.1 36.1 37.1 38.1 39.1 40.1 41.2 42.3 43.3 44.4 45.5 46.7 47.8 49.0 50.1 51.3 52.5 53.6 54.8 56.0 57.3 58.5 59.7 61.0 62.2 63.5 64.8 66.0 67.3 69.9 72.5 75.2 77.8 80.5 83.2 85.9 88.7 91.4 94.2
677 672 665 657 40.0 40.3 40.8 41.4 42.1 42.9 43.8 44.7 45.7 46.7 47.7 48.8 49.9 51.0 52.1 53.2 54.4 55.5 56.7 57.9 59.1 60.3 61.5 62.8 64.0 65.3 66.5 67.8 69.1 71.7 74.3 76.9 79.6 82.3 85.0 87.7 90.4 93.1 95.9
679 673 667 659 650 640 629 616 48.9 48.7 48.8 49.1 49.6 50.3 51.0 51.8 52.7 53.7 54.7 55.7 56.7 57.8 58.9 60.1 61.2 62.4 63.6 64.8 66.0 67.2 68.5 69.7 71.0 73.5 76.1 78.7 81.4 84.1 86.7 89.5 92.2 94.9 97.7
680 675 668 661 652 643 632 619 606 590 58.1 56.8 56.1 55.8 55.9 56.2 56.7 57.3 58.0 58.8 59.7 60.6 61.6 62.6 63.7 64.8 65.9 67.0 68.2 69.4 70.6 71.8 73.0 75.5 78.1 80.6 83.3 85.9 88.6 91.3 94.0 96.7 99.5
682 677 670 663 654 645 634 622 609 594 578 560 66.9 64.7 63.3 62.5 62.1 62.1 62.3 62.7 63.3 64.0 64.7 65.6 66.5 67.5 68.5 69.5 70.6 71.7 72.9 74.0 75.2 77.6 80.1 82.7 85.2 87.8 90.5 93.2 95.8 98.5 101
683 678 672 665 656 647 637 625 612 598 582 565 545 523 75.2 72.0 69.9 68.8 68.1 67.8 67.8 68.1 68.6 69.1 69.8 70.6 71.4 72.4 73.3 74.3 75.4 76.5 77.6 79.9 82.3 84.7 87.3 89.8 92.4 95.1 97.7 100 103
687 682 676 670 662 653 643 632 620 607 593 577 559 539 516 491 462 104 94.8 89.3 85.9 83.6 82.2 81.2 80.8 80.6 80.6 81.0 81.5 82.0 82.7 83.5 84.3 86.2 88.2 90.4 92.7 95.1 97.6 100 103 105 108
691 686 681 674 667 658 649 639 628 616 602 587 571 553 532 509 483 454 420 163 129 115 107 102 98.3 95.7 94.1 93.3 92.4 92.1 92.1 92.2 92.6 93.7 95.2 96.9 98.8 101 103 105 108 110 113
T (°C)
210
220
230
240
250
300
400
500
0 10 20 30 40 50 60 70
1
586 604 620 635 648 660 670 678
586 605 621 636 649 660 670 679
587 606 622 637 650 661 671 679
588 606 623 637 650 662 672 680
589 607 623 638 651 662 672 681
592 611 627 642 654 666 676 684
599 617 634 648 661 672 682 690
606 624 640 654 666 678 687 695
THERMOPHYSICAL PROPERTIES
2.35
TABLE 2.19 Thermal Conductivity [~.. 10 3 (W/m.deg)] of Water and Steam at Various Temperatures and Pressures (Continued) .....
Pressure, bar T (°C)
210
220
230
240
250
300
400
500
80 90 100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250 260 270 280 290 300 310 320 330 340 350 360 370 380 390 400 410 420 430 440 450 460 470 480 490 500 520 540 560 580 600 620 640 660 680 700
685 691 695 698 700 702 701 700 698 696 692 687 681 675 668 660 650 640 630 617 604 590 573 555 535 513 488 458 425 206 147 126 115 108 103 99.8 97.6 96.0 95.0 94.5 94.2 94.2 94.4 95.3 96.6 98.3 100 102 104 106 109 111 114
686 691 696 699 701 702 702 701 699 696 692 688 682 676 669 661 652 642 631 619 606 592 576 558 538 516 491 463 430 392 170 140 124 114 108 104 101 99.2 97.9 97.0 96.5 96.4 96.4 97.1 98.2 99.7 101 103 105 108 110 112 115
686 692 696 700 702 703 703 702 700 697 693 688 683 677 670 662 653 643 632 621 608 594 578 561 541 520 495 467 435 385 185 150 134 124 116 109 105 103 101 99.7 99.0 98.7 98.5 98.9 99.8 101 103 105 107 109 111 113 116
687 693 697 700 702 703 703 702 700 698 694 689 684 678 671 663 654 644 634 622 609 595 580 563 544 523 499 472 440 396 269 165 144 132 123 116 110 106 104 103 102 101 101 101 102 103 104 106 108 110 112 115 117
688 693 698 701 703 704 704 703 701 698 695 690 685 678 672 664 655 646 635 624 611 597 582 566 547 526 503 476 445 406 322 188 156 141 130 122 116 111 108 106 104 103 103 103 103 104 106 107 109 111 113 116 118
691 696 701 704 706 707 707 706 705 702 698 694 689 683 676 669 660 651 642 631 619 606 592 577 560 541 520 496 468 437 398 338 262 206 177 160 148 139 131 125 120 118 116 113 112 112 113 114 116 117 119 121 124
697 702 707 710 712 714 714 713 711 709 706 702 697 691 685 678 670 662 653 643 633 622 609 596 582 566 548 529 504 479 453 423 388 348 307 271 241 217 198 184 172 163 155 142 136 133 131 130 130 131 132 133 135
702 708 713 716 718 720 720 720 718 716 713 709 704 699 693 686 679 671 663 653 643 633 622 610 597 583 568 552 537 514 490 465 439 411 382 352 323 297 274 253 236 220 207 186 170 159 153 149 147 147 146 147 148
Sourc~
Ref. 2 with permission.
2.36
CHAPTER TWO
TABLE 2.20 T(°C)
o
T(°C)
o
T(°C)
o
T(°C)
o
0 10 20 30 40 50 60 70 80 90 100 110 120
75.50 74.40 72.88 71.20 69.48 67.77 66.07 64.36 62.69 60.79 58.91 56.97 54.96
130 140 150 160 170 180 190 200 210 220 230 240 250
52.90 50.79 48.68 46.51 44.38 42.19 40.00 37.77 35.51 33.21 30.88 28.52 26.13
260 270 280 290 300 310 320 330 340 350 355 360 361
23.73 21.33 18.94 16.60 14.29 12.04 9.84 7.69 5.61 3.64 2.71 1.85 1.68
362 363 364 365 366 367 368 369 370 371 372 373 374.15
1.53 1.37 1.22 1.07 0.93 0.79 0.66 0.54 0.42 0.31 0.20 0.10 0
Source:
TABLE 2.21
Surface Tension [o (dynes/cm)] of Water in Air
Ref. 2 with permission.
Surface Tension (N/m) of Various Liquids T(K)
Substance
250
260
270
280
290
300
320
340
360
380
400
Acetone Benzene Bromine Butane Chlorine Decane Diphenyl Ethane Ethanol Ethylene Heptane Hexane Methanol Nonane Octane Pentane Propane Propanol Propylene R 12 Toluene Water
0.0292 0.047 0.0176 0.0243 0.0278 . 0.0061 0.0033 0.0242 0.0230 0.0266 0.0270 0.0256 0.0210 0.0128 0.0274 0.0132 0.0147 0.0345 ~
0.0280 0.046 0.0164 0.0227 0.0269 . . 0.0049 0.0020 0.0233 0.0219 0.0257 0.0261 0.0247 0.0198 0.0114 0.0266 0.0119 0.0134 0.0330 ~
0.0267 0.0321 0.045 0.0152 0.0212 0.0260 . 0.0037 0.0247 0.0009 0.0224 0.0207 0.0248 0.0251 0.0237 0.0186 0.0101 0.0258 0.0105 0.0121 0.0315
0.0254 0.0307 0.044 0.0140 0.0197 0.0251
0.0241 0.0293 0.0425 0.0128 0.0182 0.0241
0.0178 0.0228 0.035 0.0069 0.0107 0.0196 0.0362 . . 0.0186
0.016 0.0204 0.032 0.0049 0.0079 0.0178 0.0338 . 0.0167 . 0.0137 0.0116 0.0169 0.0167 0.0155 0.0088 0.0007 0.0182 0.0005 ~ 0.0205 0.0615
0.012 0.0156 0.027 0.0016 0.0037 0.0145 0.0295
0.0015 0.0231 . 0.0204 0.0187 0.0229 0.0232 0.0219 0.0164 0.0076 0.0241 0.0077 0.0095 0.0288 0.0733
0.0203 0.0253 0.038 0.0092 0.0137 0.0215 0.0388 . 0.0204
0.014 0.0180 0.030 0.0031 0.0051 0.0161 0.0316
0.0026 0.0239 0.0002 0.0214 0.0198 0.0238 0.0242 0.0228 0.0175 0.0088 0.0249 0.0090 0.0108 0.0301 0.0747
0.0229 0.0279 0.041 0.0116 0.0167 0.0233 0.0416 0.0007 0.0222 . 0.0194 0.0176 0.0221 0.0223 0.0210 0.0153 0.0064 0.0232 0.0064 0.0082 0.0275 0.0717
0.0148
0.0126
0.0118 0.0096 0.0150 0.0148 0.0138 0.0069 ~ 0.0168
0.0100 0.0077 0.0129 0.0129 0.0123 0.0053 m 0.0155
~ 0.0185 0.0576
0.0165 0.0536
TABLE 2.22
.
.
. 0.0175 0.0154 0.0204 0.0204 0.0191 0.0131 0.0043 0.0214 0.0041 0.0057 0.0251 0.0685
. 0.0156 0.0134 0.0187 0.0186 0.0173 0.0108 0.0025 0.0198 0.0022 0.0034 0.0227 0.0651
.
Isobaric Expansion Coefficient of Water ([3) at one bar
T (°C)
~ x 104 (l/K)
T (°C)
~ X 10 4 (l/K)
T (°C)
~ × 104 (l/K)
r (°C)
~ × 104 (l/K)
10 15 20 25 30
0.883 1.51 2.08 2.59 3.05
35 40 45 50 55
3.47 3.86 4.23 4.57 4.90
60 65 70 75 80
5.22 5.53 5.82 6.12 6.40
85 90 95 99.63
6.69 6.96 7.22 7.46
Calculated from data in Ref. 7.
"0
# 0
I.I.J .I
C'.I
.
.
. ~ . .
~ . ~ .
~.~.
•
.
~.~.~.~.~.
~ . ~
.~.~.~.
~.
~.~.~.~
~
. . . .
~.~.~.~.~.
~ . ~ . ~ . ~ .
.~.~. ~ . ~ . ~ . ~ . .
2.37
r~
e~
o
Z
°~
.,J
2.38
r~
~.~.~.
•~
.
~
~
~
.~.
.
~
~.~.~.~.~.
~
~ . ~ .
~ o
=t
THERMOPHYSICAL PROPERTIES
TABLE 2.25
2.39
Thermal Conductivity of Seawater (mW/m K) at Various Temperatures and Salinities Salinity, g/kg
T (°C)
0
10
20
30
35*
40
50
60
70
80
90
100
110
120
130
140
150
0 10 20 30 40
572 589 604 618 630
570 587 603 617 629
569 586 602 616 629
567 584 600 615 628
566 584 600 614 628
565 583 599 614 627
563 581 598 613 626
562 580 597 612 626
560 578 595 611 625
558 577 594 609 624
556 575 592 608 623
554 573 591 607 622
552 571 589 606 621
550 570 588 604 620
548 568 586 603 618
546 566 585 602 617
544 564 583 600 616
50 60 70 80 90
641 651 659 666 672
641 651 659 666 672
640 650 659 667 673
640 650 659 667 673
639 650 659 667 673
639 650 659 667 674
639 649 659 667 674
638 649 659 667 674
637 649 658 667 674
637 648 658 667 675
636 648 658 667 675
635 647 658 667 675
634 646 657 667 675
633 646 657 666 675
632 645 656 666 675
631 644 656 666 675
630 644 655 666 675
100 110 120 130 140
676 680 682 683 684
677 681 683 685 685
678 682 684 686 687
678 683 685 687 688
679 683 686 688 689
679 683 686 688 689
680 684 687 690 691
680 685 688 691 692
681 685 689 692 693
681 686 690 693 694
681 687 691 694 696
682 687 691 695 697
682 688 692 695 698
682 688 693 696 699
682 688 693 697 700
682 689 694 698 701
683 689 694 699 702
150 160 170 180
683 681 678 674
684 683 680 676
686 684 682 678
688 686 684 680
688 687 685 681
689 688 686 682
691 690 687 684
692 691 689 686
694 693 691 686
695 694 693 690
696 696 694 692
698 697 696 694
699 699 698 695
700 700 699 697
701 701 701 699
702 703 702 700
703 704 704 702
110
120
130
140
150
* "Normal" seawater. Re~3 with permission.
Source.
TABLE 2.26
Prandtl Number of Seawater at Various Temperatures and Salinities Salinity, g/kg
T(°C)
0
10
20
30
0 10 20 30 40
13.1 9.29 6.95 5.40 4.33
13.1 9.35 6.99 5.45 4.38
13.1 9.39 7.04 5.49 4.41
13.2 9.46 7.11 5.54 4.46
50 60 70 80 90
3.56 2.99 2.57 2.23 1.97
3.60 3.03 2.60 2.26 2.00
3.64 3.06 2.63 2.29 2.02
3.68 3.10 2.66 2.32 2.05
3.71 3.12 2.68 2.34 2.06
3.73 3.14 2.70 2.35 2.08
3.77 3.19 2.74 2.39 2.11
3.83 3.24 2.78 2.42 2.14
3.89 3.28 2.82 2.46 2.18
3.95 3.34 2.87 2.50 2.21
4.02 3.40 2.92 2.55 2.25
4.10 3.47 2.98 2.60 2.29
4.18 3.54 3.04 2.65 2.34
4.28 3.61 3.11 2.71 2.39
4.38 3.69 3.18 2.77 2.44
4.48 3.78 3.25 2.83 2.50
4.60 3.88 3.33 2.90 2.56
100 110 120 130 140
1.75 1.59 1.44 1.33 1.23
1.78 1.61 1.47 1.35 1.24
1.80 1.63 1.49 1.37 1.26
1.83 1.65 1.51 1.38 1.28
1.84 1.66 1.51 1.39 1.29
1.86 1.68 1.53 1.40 1.30
1.88 1.70 1.55 1.42 1.31
1.92 1.73 1.57 1.44 1.33
1.94 1.75 1.60 1.46 1.35
1.98 1.78 1.62 1.49 1.37
2.01 1.81 1.65 1.51 1.39
2.05 1.84 1.68 1.54 1.42
2.09 1.88 1.71 1.57 1.44
2.13 1.92 1.75 1.60 1.47
2.18 1.96 1.78 1.63 1.50
2.23 2.00 1.82 1.66 1.53
2.28 2.05 1.86 1.70 1.56
150 160 170 180
1.14 1.08 1.01 0.959
1.16 1.08 1.03 0.975
1.18 1.10 1.04 0.983
1.19 1.11 1.05 0.997
1.20 1.12 1.06 1.00
1.21 1.13 1.06 1.00
1.22 1.14 1.07 1.02
1.24 1.16 1.09 1.03
1.26 1.17 1.10 1.04
1.27 1.19 1.12 1.06
1.30 1.21 1.13 1.07
1.32 1.23 1.16 1.09
1.34 1.25 1.17 1.10
1.36 1.28 1.20 1.13
1.39 1.30 1.22 1.14
1.42 1.32 1.24 1.17
1.45 1.35 1.26 1.19
* "Normal" seawater. Ref. 3 with permission.
Source
35*
40
50
60
70
80
90
100
13.2 13.3 13.4 13.5 13.6 13.8 14.0 14.3 14.5 14.8 15.2 15.5 16.0 9.49 9.53 9.62 9.72 9.84 9.97 10.1 10.3 10.5 10.7 11.0 11.2 11.6 7.13 7.17 7.24 7.33 7.43 7.53 7.67 7.80 7.96 8.13 8.32 8.52 8.76 5.58 5.60 5.67 5.74 5.82 5.92 6.01 6.12 6.24 6.39 6.54 6.69 6.88 4.48 4.51 4.57 4.63 4.70 4.78 4.86 4.95 5.05 5.16 5.28 5.42 5.56
c~
c~
t~
0
c~ © 0
©
f~
LU ..I ,,n
2.40
i
r~
o
0
0
'13
~j
"r~ .z= ~L 0
I.LI ,.I
~B
{'4
I
I
"T
w
;>
:>
>
:>
;>
:>
:>
;>
;>
;>
;>
:>
;>
:>
:>
:>
:>
:>
;>
;>
>.
~,.
>
;>
>.
>.
>
>
;>
;>
;>
;>
;>
:>
;>
;>
;>
:>
>.
:>
>.
>.
>
>
:>
>.
>
>
>
:>
>.
;
:>
>.
:>
;>
;~
;>
>
:>
:>
:>
;>
;>
;>
:>
:>
~>
~>
;>
;>
:>
>.
I ~.o~.,.~;~o
~'~ o~,~
>.
;>
o
c:
"~u
~°~'="
.
>
>
~:.~
uo~ ~
;>
.~
:>
u
;~
('4
('q
04.
:>
.~
>
{'~
"~
~'~==
°
..~
>
~"
I
o
;>
:>
;>
;>
:>
;>
;>
;>
;>
>
>
~
('~
;>
~
CD
.
("4
o~
~0
O0
4"~
~ = ~ . ~
I~~.~
='
"~
11 b"
2.41
0
0
'13
~j
"r~ .z= ~L 0
I.LI ,.I
~B
{'4
I
I
"T
w
;>
:>
>
:>
;>
:>
:>
;>
;>
;>
;>
:>
;>
:>
:>
:>
:>
:>
;>
;>
>.
~,.
>
;>
>.
>.
>
>
;>
;>
;>
;>
;>
:>
;>
;>
;>
:>
>.
:>
>.
>.
>
>
:>
>.
>
>
>
:>
>.
;
:>
>.
:>
;>
;~
;>
>
:>
:>
:>
;>
;>
;>
:>
:>
~>
~>
;>
;>
:>
>.
I ~.o~.,.~;~o
~'~ o~,~
>.
;>
o
c:
"~u
~°~'="
.
>
>
~:.~
uo~ ~
;>
.~
:>
u
;~
('4
('q
04.
:>
.~
>
{'~
"~
~'~==
°
..~
>
~"
I
o
;>
:>
;>
;>
:>
;>
;>
;>
;>
>
>
~
('~
;>
~
CD
.
("4
o~
~0
O0
4"~
~ = ~ . ~
I~~.~
='
"~
11 b"
2.41
>
;>
;>
>
>
;>
;>
>
;>
;>
>
>
>
;>
>
>
;>
;>
>
;>
;>
>
;>
;>
>
;>
;>
;>
>
>
;>
>
;>
>
>
;>
;>
;>
;>
>
¢5
>
v¢,I
;>
;>
g,
;>
g
F,
.~
[.= u= ~ ~
~
g
;>
.~
>
~'
>
",~
.~
,4
•.
~
>
~
>
~'~'~
~ •. ~
;>
~
>
];>
>
~.
;>
>
>
>
~.
>
>
113
113
~
>
>
•,~
>
>
"~
~ID
~
~
;>
>
>
o
>
eq
~ ~
0
cq
.=
¢,1
~
~
~
>
>
;>
~>
>.
;>
;>
>
>.
>
>.
>
>
>
~.
d
~
=
o
>
.~
¢,i
..~
;~
>.
"~
~
.~
~
~
,",,"n. ~
~
~
~
~
"
.
~., ~ ~ ' a
~
"'
~
r'~,,i ~.. ~
N
~ ~.~Z
~o
~
~,~.~
~..
"
~
G ~~o
~ " ; ~
~ o o
~*
= . .
~c)~,c)~'~'=~"
~
~'~
~
~
~
c ) ~ ~ c )
~.R
~
~d
~. O o
¢q
~
>
!
~
' ~~ . o~ ~1 Z
oo
~2 .~ ~ . . . ~ ] . ~c ) : ~ 0: ~
_
~
. . .,..,,,n. ~  ¢ ~ ,% .~
'¢'~ o ~
~ " q
. ~.
~.
~
.
~ ,.
~
0,.~o mg,,. , ~~
~'~'~
.~.u
~"
=
"
~
..
~
.0 .. ¢; ~ c )~~ 
~
°
• ...
~ r 3 
~
"~'~
. ~
..o
~=~)~
~
g
o
2.43
2.44
e~
R
0
0~
,,.I
;>
>.
;>
;>.
;>
>
;>
>
>
>"
~>
>
>"
>.
>.
D,"
>.
>.
;>" >
>.
;>
>.
;>
;>
;>,
;>
>
~,.
>
>
>.
;>
;>
;>
>.
;>
;>
>,
;>
;>
;>
>
>
2
;>
N
~,
d
;>
~
;>
~
I
>.
K"
"o6~
>
t~
,?
1
T T T 'T
?
"
>
le&e
r~
. u ~.
~/~
05
b5
~5
F. o
;>
;>
;>
;>
;>
;>
;>
;>
;>
;>
;>
;>
;>
;>
;>
;>
;>
;>
;>
;>
;>
;>
;>
;>
;~
;>
;>
;>
;>
~
;>
;>
>
;> ;>
;> ;>
;> ;~
;>
°
N,~
,.,.>
:'~ 0 ~ 1
C o
ff
~u
~
~
~
~=
~
~.~ ~
u ~
~
0
2.45
2.46
CHAPTER TWO
TABLE 2.29
Thermophysical Properties of Liquid Metals
Composition
Melting point (K)
Bismuth
544
Lead
600
Potassium
337
Sodium
371
NaK (45 %/55 %)
292
NaK (22%/78%)
262
PbBi (44.5%/55.5%)
398
T (K)
9 (kg/m 3)
cp (kJ/kg.K)
589 811 1033 644 755 977 422 700 977 366 644 977 366 644 977 366 672 1033 422 644 922
10,011 9,739 9,467 10,540 10,412 10,140 807.3 741.7 674.4 929.1 860.2 778.5 887.4 821.7 740.1 849.0 775.3 690.4 10,524 10,236 9,835
0.1444 0.1545 0.1645 0.159 0.155 m 0.80 0.75 0.75 1.38 1.30 1.26 1.130 1.055 1.043 0.946 0.879 0.883 0.147 0.147 ~
(m2/s)
k (W/m.K)
(x. 105 (m2/s)
1.617 1.133 0.8343 2.276 1.849 1.347 4.608 2.397 1.905 7.516 3.270 2.285 6.522 2.871 2.174 5.797 2.666 2.118 m 1.496 1.171
16.4 15.6 15.6 16.1 15.6 14.9 45.0 39.5 33.1 86.2 72.3 59.7 25.6 27.5 28.9 24.4 26.7 9.05 11.86 ~
0.138 1.035 1.001 1.084 1.223 6.99 7.07 6.55 6.71 6.48 6.12 2.552 3.17 3.74 3.05 3.92 ~ 0.586 0.790 ~
V" 10 7
Pr 0.0142 0.0110 0.0083 0.024 0.017 0.0066 0.0034 0.0029 0.011 0.0051 0.0037 0.026 0.0091 0.0058 0.019 0.0068
0.189
Adapted from Liquid Materials Handbook, 23rd ed., the Atomic Energy Commission, Department of the Navy, Washington, DC, 1952.
THERMOPHYSICAL PROPERTIES OF SOLIDS TABLE 2.30
Density of Selected Elements (kg/m 3) Symbol
T (K)
A1
Sb*
Ba
Be*
Bi*
Cd*
Ca
50 100 150 200 250
2736 2732 2726 2719 2710
6734 6726 6716 6706 6695
3650 3640 3630 3620 3610
1863 1862 1861 1860 1858
9880 9870 9850 9830 9810
8830 8800 8760 8720 8680
1572 1568 1563 1559 1554
300 400 500 600 800
2701 2681 2661 2639 2591
6685 6662 6638 6615 6569
3600 3580 3555 3530
1855 1848 1840 1831 1812
9790 9750 9710
8640 8560 8470 8010 7805
1550 1539 1528 1517
1000 1200 1400 1600 1800
2365 2305 2255
6431 6307 6170
1790 1768 1744
7590
Ce
Cs 1962 1944 1926 1907 1887
6860 6850 6840 6820 6790
!866 1781 1723 1666 1552
6760
1438 1311 1182
2000 * Polycrystalline form tabulated. Above the horizontal line the condensed phase is solid; below, it is liquid. *Hysteresis effect present.
THERMOPHYSICAL PROPERTIES
TABLE 2.30
2.47
Density of Selected Elements (kg/m 3) (Continued) Symbol
T (K)
Cr
Cu
Co
Dy*
Er
Eu*
50 100 150 200 250
7160 7155 7150 7145 7140
9019 9009 8992 8973 8951
8925 8919 8905 8892 8876
8578 8579 8581 8580 8567
9120 9105 9090 9080 9070
300 400 500 600 800
7135 7120 7110 7080 7040
8930 8884 8837 8787 8686
8860 8823 8784 8744 8642
8554 8530 8507 8484 8431
9060 9030 9000 8970 8910
1000 1200 1400 1600 1800
7000 6945 6890 6830 6760
8568 8458 7920 7750 7600
8561 8475
8377
8840 8740
7630
2000
6700
7460
7410
Gd*
Ga
Ge
7966 7960 7954 7949 t 5240 5190 5155 5127
Au
Hf
5363 5358 5353 5348 5344
19,490 19,460 19,420 19,380 19,340
13,350 13,340 13,330 13,320 13,310
t t 7926 7907 7866
5910 6010 5946 5880 5770
5340 5330 5320 5310 5290
19,300 19,210 19,130 19,040 18,860
13,300 13,275 13,250 13,220 13,170
7818 7754
5650 5540 5420
5265 5240
18,660 18,440 17,230 16,950
13,110 13,050
Ho
In*
Ir
Fe
La*
Pb
Li
Lu*
Mg
Mo
50 100 150 200 250
8820 8815 8810 8800 8790
7460 7430 7400 7370 7340
22,600 22,580 22,560 22,540 22,520
7910 7900 7890 7880 7870
6203 6200 6196 6193 6190
11,570 11,520 11,470 11,430 11,380
547 546 543 541 537
9830 9840 9840 9850 9840
1765 1762 1757 1752 1746
10,260 10,260 10,250 10,250 10,250
300 400 500 600 800
8780 8755 8730 8700 8650
7310 7230 6980 6810
22,500 22,450 22,410 22,360 22,250
7860 7830 7800 7760 7690
6187 6180 6160 6170 6140
11,330 11,230 11,130 11,010 10,430
533 526 492 482 462
9830 9800 9770 9740 9660
1740 1736 1731 1726 1715
10,240 10,220 10,210 10,190 10,160
1000 1200 1400 1600 1800
8600
22,140 22,030 21,920 21,790 21,660
7650 7620 7520 7420 7320
6160
10,190 9,940
442 442 402 381 361
9580 9500
1517 1409
10,120 10,080 10,040 10,000 9,950
21,510
7030
2000
341
9,900
Ni
Nb
Os
Pd
Pt
Pu
K
50 100 150 200 250
8960 8960 8940 8930 8910
8610 8600 8590 8580 8570
22,550 22,540 22,520 22,510 22,490
12,110 12,100 12,090 12,070 12,050
21,570 21,550 21,530 21,500 21,470
20,270 20,170 20,080 19,990 19,860
905 898 890 882 873
300 400 500 600 800
8900 8860 8820 8780 8690
8570 8550 8530 8510 8470
22,480 22,450 22,420 22,390 22,320
12,030 11,980 11,940 11,890 11,790
21,450 21,380 21,330 21,270 21,140
19,730 17,720 17,920 15,300 16,370
863 814 790 767 720
* Polycrystallineform tabulated. Above the horizontal line the condensed phase is solid; below, it is liquid. t Hysteresis effect present.
Pa*
15,370 15,320 15,280 15,230 15,150
2.48
CHAPTER TWO
TABLE 2.30
Density of Selected Elements (kg/m 3) (Continued) Symbol
T (K)
Ni
Nb
Os
Pd
Pt
1000 1200 1400 1600 1800
8610 8510 8410 8320 7690
8430 8380 8340 8290 8250
22,250
11,680 11,570
21,010 20,870 20,720 20,570 20,400
2000
7450
82.00
Re*
Rh
50 100 150 200 250
21,100 21,070 21,040 21,020 21,010
12,490 12,480 12,470 12,460 12,445
300 400 500 600 800
21,000 20,960 20,920 20,880 20,800
12,430 12,400 12,360 12,330 12,250
1000 1200 1400 1600 1800
20,710 20,630 20,540 20,450 20,350
12,170 12,080 11,980 11,880
2000
20,250
Rb
Sc*
14,320 13,860 13,400 12,340 11,560 10,640 9,720
Pa*
672 623 574 527
15,050 14,910
Ag
Na
Sr
Ta
10,620 10,600 10,575 10,550 10,520
1014 1007 999 990 980
2655 2638 2632 2621 2618
16,500 16,490 16,480 16,460 16,450
3000 2990 2980 2970 2950
10,490 10,430 10,360 10,300 10,160
970 921 897 874 826
2615
16,440 16,410 16,370 16,340 16,270
2930 2910
10,010 9,850 9,170 8,980
779 731 683 638
16,200 16,130 16,060 15,980 15,910 15.820
Th
Tm*
50 100 150 200 250
12,080 12,040 12,000 11,950 11,900
11,745 11,740 11,745 11,750 11,735
9370 9360 9350 9340 9330
300 400 500 600 800
11,850 11,730 11,500 11,250 10,960
11,720 11,680 11,630 11,590 11,500
9320 9280 9250 9210 9150
7280
11,400 11,300
9080
6620 6480 6340
2000
K
20,220
TI
1000 1200 1400 1600 1800
Pu
Sn
6900 6900 6760
Ti
W
U*
V
4530 4510 4515 4520 4515
19,320 19,310 19,300 19,290 19,280
19,240 19,210 19,170 19,140 19,100
6080 6074 6068 6062 6056
4510 4490 4480 4470 4440
19,270 19,240 19,220 19,190 19,130
19,070 18,980 18,890 18,790 18,550
6050 6030 6010 6000 5960
4410 4380 4350 4320
19,080 19,020 18,950 18,890 18,830
18,110 5920 17,760 5880 17,530 5830 5780 5730
4110
18 760
Yb
Y*
Zn*
Zr*
4500 4490 4485 4480 4475
7280 7260 7230 7200 7170
6540 6535 6530 6525 6520
7020 6960 6900 6850 6720
4470 4450 4440 4420 4390
7135 7070 7000 6935 6430
6515 6510 6490 6480 6450
6590
4360 4320
6260
6420 6410 6380 6340 6300
* Polycrystalline form tabulated. Above the horizontal line the condensed phase is solid; below, it is liquid. , Hysteresis effect present.
6260
w
UJ I
~
~
dd
dd
oo
gg d~
I
e
i~
dd
do
~ d~
2 ~E g~ 8~
~.~.~•
I~
.
I~
.~.~. I~.
~I
~I
.~.II I~
I ~ I
~
o ~ . o o
g
I
llllf
d
d d d ~ d
d ~ d ~ d
IZlll
d d d d d
d d d ~ d
ooe
d d d d d
~
d d d d d
d d d d d
o
e o e o
d M d d d
~
~
~ I I~.~.
d ~ d d d
d d d d d
o.
I ~ I ~ 
.
I ~ I~
o o.
g
O
O
II (N
(N
O
O
~g ~g
~g
2.49
e'~
U
,I
2.50
g
r~
•
t¢3 t",l
~
~
p.
p.
n.
t"l
t~
~g
~g
Z
P~g g
ZZ
~
~
t"l
t~
~
o
IIIII
o o o o
.
o o o
ooo
~
0 0 0 0
d
d
I ~ II1~1
m~
I I I~1
N
o o o
o o d
d
d
q
o
~
.
d d d ~ d
I
d
~
I
d d d
~
~
I~
~11
cc ',~ c~ C~
0
ii~ ~
g~gg
THERMOPHYSICAL PROPERTIES TABLE 2.32
Thermal Conductivity and Density of Selected Elements
Substance Aluminum, 99.75 %
Chemical formula AI
99%
Antimony, very pure
Beryllium, 99.5 %
Bismuth
Cadmium, pure
Sb
Be
Bi
Cd
T (°C)
T (°K)
190 0 200 300 800
83.15 273.15 473.15 573.15 1073.15
100 0 100 300
173.15 273.15 373.15 573.15
190 100 0 100 300 500
83.15 173.15 273.15 373.15 573.15 773.15
250 100 0 100 200
23.15 173.15 273.15 373.15 473.15
190 100 0 100 200
83.15 173.15 273.15 373.15 473.15
190 100 0 100 200 300
83.15 173.15 273.15 373.15 473.15 573.15
Cobalt, 97.1%
Co
20
293.15
Copper, pure 99.998 %
Cu
180 100 0 100 200 400 600
93.15 173.15 273.15 373.15 473.15 673.15 873.15
20
293.15
Commercial Electrolytic, pure
Gold 99.999%
99.98%
2.51
Au
180 0 100 300 800
93.15 273.15 373.15 573.15 1073.15
190 0 100 300
83.15 273.15 373.15 573.15
0 100
273.15 373.15
Density p (kg/m 3) 2,700
m
6,690
Thermal conductivity ~, (W/m K) 255.860 229.111 229.111 222.133 125.604 209.340 209.340 207.014 222.133 20.934 19.190 17.678 16.282 15.817 18.608
1,850
94.203 125.604 160.494 190.732 215.155
9,800
25.586 12.095 8.374 7.211 7.211
8,620
=8,900
8,930
8,300 8,900
19,290
104.670 96.529 93.040 91.877 91.296 87.807 69.780 464.037 407.050 386.116 379.138 373.323 364.019 353.552 372.160 488.460 395.420 391.931 381.464 367.508 327.966 310.521 310.521 304.706 294.239 294.239
2.52
CHAPTER TWO
TABLE 2.32
Thermal Conductivity and Density of Selected Elements (Continued)
Substance
Chemical formula
T (°C)
T (°K)
Density p (kg/m 3)
Thermal conductivity ~, (W/m K)
Iridium, pure
Ir
0 100
273.15 373.15
22,420
59.313 56.987
Iron (Armc) 99.92%
Fe
20 100 200 400 600 800
293.15 373.15 473.15 673.15 873.15 1073.15
7,850
73.169 67.454 61.639 48.846 38.379 29.075
Cast, 1% Ni
20 100 300 500
293.15 373.15 573.15 773.15
7,280
50.009 49.428 46.520 37.216
Cast, 3% C
20
293.15
7,280
Steel, 99.2% Fe, 0.2% C
0 100 300 500 800
273.15 373.15 573.15 773.15 1073.15
7,800
45.357 45.357 43.031 37.216 30.238
Wrought, pure
0 100 200 400 600 800
273.15 373.15 473.15 673.15 873.15 1073.15
7,800
59.313 56.987 52.335 44.194 37.216 29.075
250 200 100 0 20 100 300 500
23.15 73.15 173.15 273.15 293.15 373.15 573.15 773.15
Lead, pure
Pb
11,340
55.824... 63.965
48.846 40.705 36.867 35.123 34.774 33.378 29.773 16.747
Lithium, pure
Li
0 100
273.15 373.15
Magnesium, pure
Mg
190 0 200
83.15 273.15 473.15
0 100 300 500
273.15 373.15 573.15 773.15
=1,740
7,300
50.242
13,595
48.846 36.053 27.912 8.141
99.6%
Manganese
Mn
0
273.15
Mercury
Hg
190 100 50 0
83.15 173.15 223.15 273.15
(Liquid)
530
1,740
70.943 70.943 186.080 172.124 162.820 144.212 139.560 131.419 131.419
THERMOPHYSICAL PROPERTIES
TABLE 2.32
Thermal Conductivity and Density of Selected Elements (Continued)
Substance
Chemical formula
Molybdenum 99.84%
Mo
Nickel 99.94%
Ni
99.2%
97 to 99%
10,200
174.450 138.397 137.234 137.234 98.855
180 100 0 100 1000
93.15 173.15 273.15 373.15 1273.15
180 0 100 200 300 400 500
93.15 273.15 373.15 473.15 573.15 673.15 773.15
0 100 200 400 600 800
273.15 373.15 473.15 673.15 873.15 1073.15
67.454 62.802 58.150 52.335 56.987 62.802
100 0 100 200 400 600 800
173.15 273.15 373.15 473.15 673.15 873.15 1073.15
55.824 58.150 56.987 54.661 48.846 53.498 58.150 76.758 68.617 73.269
190 0 100
83.15 273.15 373.15
Platinum, pure
Pt
190 0 100 300 500 800 1000
83.15 273.15 373.15 573.15 773.15 1073.15 1273.15
Potassium, pure
K
0 100
273.15 373.15
Rhodium, pure
Rh
190 0 100
83.15 273.15 373.15
190 0 100 300
83.15 273.15 373.15 573.15
100 0 100 300 500
173.15 273.15 373.15 573.15 773.15
Ag
Thermal conductivity ~, (W/m K)
T (°K)
Pd
Silver > 99.98%
Density p (kg/m 3)
T (°C)
Palladium, pure
99.9%
2.53
8,800
21,400
860
12,500
10,500
10,500
110.485 93.040 82.573 73.269 63.965 59.313 61.639
77.921 70.013 71.408 75.595 79.084 86.062 89.551 136.071 118.626 212.829 88.388 80.247 425.658 418.680 416.354 407.050 419.843 410.539 391.931 361.693 362.856
2.54
CHAPTER TWO
TABLE 2.32
Thermal Conductivity and Density of Selected Elements (Continued)
Substance Sodium, pure
Chemical formula Na
T (°C)
T (°K)
100 0 50 100
173.15 273.15 323.15 373.15
Tantalum
Ta
0 100 1000 1400 1800
273.15 373.15 1273.15 1673.15 2073.15
Thallium, pure
T1
190 0 100
83.15 273.15 373.15
150 100 0 100 200
123.15 173.15 273.15 373.15 473.15
190 0 100 500 1000 1500 2000 2400
83.15 273.15 373.15 773.15 1273.15 1773.15 2273.15 2673.15
100 0 100 200 300
173.15 273.15 373.15 473.15 573.15
Tin, pure
Sn
Wolfram
W
Zinc, pure
Source:
Zn
Ref.2 with permission.
Density p (kg/m3) 970
16,650
Thermal conductivity ~, (W/m K) 154.679 100.018 93.040 83.736 54.661 54.080 63.965 72.106 82.573
11,840
62.802 51.172 41.868
7,300
79.084 74.432 66.058 59.313 56.987
19,300
7,130
217.481 166.309 151.190 119.789 98.855 113.974 136.071 146.538 115.137 112.811 109.904 105.833 101.181
E
[]
r~
03
¢~1 ,,=1
~
.
. ~
.
.
.
~
~ .
~
.
.
.
.
~
~
.
~ .
l
~
~
~
~
~
~ M M M ~
,~,~1
~
~ , ~
,,
.
,,,,,
,,,,,[
.
~,,,
~ , ,
~
~ ~
~,,~,
~'1¢3
ooo~
t'~lt'xl
I
~
~
~
~
I.I.I.I.I.I.
. . . .
~g~d
~
d~dd~
IIII
.~ ~ .E
IIli
N
~
Mddg
g g d ~' ~
~
d d ~ d~d ~
5
N
~
.
2.55
2.56
i
N N
g~
m
J
F:
IIIII
~, , , ~I t , ,
I
I
I
I I I 1 [
I I I I I I
I
I I I I I ] l
I
. ~ ~
~ , , , ,
I
,
I
i
I
, ~
L
I
~1 ~ ~
III
II
I
~
,
I
~
Z
¢: :::1
Z
oo. ~ . . . m
~.~m~."
~ m ~
I.I,
LI.
~ .
I. I. I.
I. I. I. I. I.
I. I. I. I.
~
L'.'.'.'. '.'.'.
~
~!'.'.'.
'.'.'.J.'. '.'.'.
~ ~ ~
g ~ @
L~!~'.
. I.L
~ m ~
i. I. I.
LL
~
I.L L L L
~
I
I
I.l.I.
I
I
I I I I I
I c4~
I1.1.
l l l l l
I.
I I I I I
,,¢
§ cq
_ _
~
i
~
~
J
~
I
~
~
, I I,,
~
~
~
Ill
I
~. I
~. . . . .
i
I.
I
i
I
i
~ 1 1 1 1
I
I
~ , ,
I
, , ,~
I
~1
I
i
~
7 7 ~ 1
I l l ~
~
I
i
~
I.
i
~
~
, ~
~
I. I
~
~
,
I. I
~
I.
~
o
'"
X
o
;..= I
o
2.57
2.58
CHAPTERTWO TABLE 2.34
Density and Thermal Conductivity of Alloys
Alloy
Composition (%)
Aluminum alloys
96 AI, 1.8 Cu, 0.9 Fe, 0.9 Cr, 0.4 Si
Aluminum bronze
95 Cu, 5AI
Aluminum magnesium
92 AI, 8 Mg
Alusil
Bismuthantimony
Brass
T (K)
Density p (kg/m 3)
Thermal conductivity ~, (W/m K)
20
293.15

104.670
7800
T (°C)
20
293.15
180 100 0 20 100 200
93.15 173.15 273.15 293.15 373.15 473.15
180 100 0 20 100 200
93.15 173.15 273.15 293.15 373.15 473.15
80 Bi, 20 Sb
0 100
273.15 373.15

6.606 8.618
50 Bi, 50 Sb
0 100
273.15 373.15

8.327 9.374
30 Bi, 70 Sb
0 100
273.15 373.15

9.653 11.660
100 0 100 200 300 400 500 600
173.15 273.15 373.15 473.15 573.15 673.15 773.15 873.15
70 Cu, 30 Zn
0 100 200 300 400 500 600
273.15 373.15 473.15 573.15 673.15 773.15 873.15
=8600
105.833 109.322 110.485 113.974 116.300 119.789 120.952
66 Cu, 33 Zn
0 100 200 300 400 500 600
273.15 373.15 473.15 573.15 673.15 773.15 873.15
=8600
100.018 106.996 112.811 120.952 127.930 134.908 151.190
60 Cu, 40 Zn
0 100 200 300 400 500 600
273.15 373.15 473.15 573.15 673.15 773.15 873.15
=8600
105.833 119.789 137.234 152.353 168.635 186.080 200.036
80 A1, 20 Si
90 Cu, 10 Zn
=2600
=2650
=8600
82.573 75.595 84.899 102.344 105.833 123.278 147.701 122.115 141.886 158.168 160.494 168.635 174.450
88.388 102.344 117.463 133.745 148.864 166.309 180.265 195.384
THERMOPHYSICAL PROPERTIES TABLE 2.34
Density and Thermal Conductivity of Alloys (Continued)
Alloy
Composition (%)
Brass
61.5 Cu, 38.5 Zn
Bronze
Chromenickel steel
T (°C)
T (K)
Density p (kg/m 3)
Thermal conductivity ~. (W/m K)
20 100
293.15 373.15
90 Cu, 10 Sn
20
293.15
8766
75 Cu, 25 Sn
20
293.15
=8900
25.586
88 Cu, 10 Sn, 2 Zn
20
293.15
8800
47.683
84 Cu, 6 Sn, 9 Zn, 1 Pb
20
293.15

58.150
86 Cu, 7 Zn, 6.4 Sn
20 100
293.15 373.15
0.8 Cr, 3.5 Ni, 0.4 C
20 100 200 400 600
293.15 373.15 473.15 673.15 873.15
8100... 8700
34.890 36.053 37.216 37.216 31.401
C r . . . Ni
20 200 500
293.15 473.15 773.15
7900
13.956 17.445 20.934
17 . . . 19 Cr, 8 Ni, 0.1 . . . 0.2 C
20 100 200 300 500
293.15 373.15 473.15 573.15 773.15
8100... 9000
14.538 15.701 16.864 18.608 20.934
10 Cr, 34 Ni
20 100 200 300 500
293.15 373.15 473.15 573.15 773.15
20 100 200 300 500
293.15 373.15 473.15 573.15 773.15
20 100 200 300 500 800
293.15 373.15 473.15 573.15 773.15 1073.15
0.8 Cr, 0.2 C
100 200 400 600
373.15 473.15 673.15 873.15
=7850
39.542 37.216 31.401 26.749
5 Cr, 0.5 Mn, 0.1 C
20 100 200 500
293.15 373.15 473.15 773.15
8100... 9000
37.216 31.635 31.053 33.727
15 Cr, 0.1 C
20 500
293.15 773.15
8100... 9000
25.586 25.586
15 Cr, 27 Ni, 3 W, 0.5 C
15 Cr, 13 Ni, 2 W, 0.5 C
Chrome steel
2.59
79.084 88.388
8600
41.868
60.476 70.943
m
12.212 13.375 15.119 16.282 19.190

11.281 12.793 13.956 15.119 18.608
m
11.630 11.630 11.630 12.212 12.793 16.282
2.60
CHAPTER TWO
TABLE 2.34
Density and Thermal Conductivity of Alloys (Continued)
Alloy Chrome steel
Composition (%)
Thermal conductivity 9~(W/m K)
293.15 373.15 473.15 573.15 773.15
8100... 9000
24.423 25.005 25.586 25.586 25.586
100 200 300 500 800
373.15 473.15 573.15 773.15 1073.15
8100... 9000
23.842 23.260 23.260 23.260 23.260
20 100 200 300 500
293.15 373.15 473.15 573.15 773.15
8100... 9000
19.771 20.934 22.097 22.911 24.423
T (K)
14 Cr, 0.3 C
20 100 200 300 500
16 Cr, 0.9 C
26 Cr, 0.1 C
Cobalt steel
5 . . . 10 Co
Constantin
60 Cu, 40 Ni
Copper alloys
Density p (kg/m 3)
T (°C)
92 AI, 8 Cu
20
293.15
100 0 20 100
173.15 273.15 293.15 373.15
180 100 0 20 100 200
93.15 173.15 273.15 293.15 373.15 473.15
7800
8800
=2800
40.705 20.934 22.213 22.679 25.586 89.551 109.322 127.930 131.419 143.049 152.353
Coppermanganese
70 Cu, 30 Mn
20
293.15
=7800
12.793
Coppernickel
90 Cu, 10 Ni
20 100
293.15 373.15
8800
58.150 75.595
80 Cu, 20 Ni
20 100
293.15 373.15
8500
33.727 40.705
40 Cu, 60 Ni
20 100
293.15 373.15
8400
22.097 25.586
18 Cu, 82 Ni
20 100
293.15 393.15
25.586 25.586
9 4 . . . 96 A1, 3 . . . 5 Cu, 180 0.5 Mg 100 0 20 100 200
93.15 173.15 273.15 293.15 373.15 473.15
90.714 125.604 159.331 165.146 181.428 194.221
Electron alloy
93 Mg, 4 Zn, 0.5 Cu
20
293.15
1800
116300
German alloy
88 AI, 10 Zn, 2 Cu
0 20 100
273.15 293.15 373.15
2900
143.049 145.375 154.679
Goldcopper alloy
88 Au, 12 Cu
0 100
273.15 373.15
m
55.824 67.454
27 Au, 73 Cu
0 100
273.15 373.15
m
90.714 113.974
Duralumin
2800
THERMOPHYSICAL PROPERTIES TABLE 2.34
Density and Thermal Conductivity of Alloys (Continued)
Alloy
Composition (%)
T (°C)
T (K)
Density p (kg/m 3)
Thermal conductivity ~. (W/m K) 11.049
Invar
35 Ni, 65 Fe
20
293.15
8130
Lautal
95 AI, 4.5 . . . 5.5 Cu, 0.3 Si
20
293.15
~
Magnesiumaluminum
92 Mg, 8 AI
93.15 173.15 273.15 293.15 373.15 473.15
2.5 A1
20
293.15

85.597
4.2 AI
20
293.15
~
69.082
6.2 AI
20
293.15
~
55.591
10.3 A1
20
293.15
~
43.496
88 Mg, 10 AI, 2 Si
Magnesiumcopper
92 Mg, 8 Cu
93.7 Mg, 6.3 Cu Manganesenickel steel
Manganese steel
Monel
139.560
180 100 0 20 100 200
Magnesiumaluminumsilicone
Manganine
2.61
180 100 0 20 100 200
93.15 173.15 273.15 293.15 373.15 473.15
180 100 0 20 100 200
93.15 173.15 273.15 293.15 373.15 473.15
=1800
=1850
=2400
41.868 50.009 60.476 61.639 69.780 79.084
30.238 40.705 55.824 58.150 68.617 75.595 88.388 106.996 124.441 125.604 130.256 132.582
20
293.15
131.419
20 100 200 300 500
293.15 373.15 473.15 573.15 773.15
13.956 14.770 16.282 17.445 19.771
20 100 300 500
293.15 373.15 573.15 773.15
2 Mn
20
5 Mn
20
12 Mn, 3 Ni, 0.75 C
1.6 Mn, 0.5 C
84 Cu, 4 Ni, 12 Mn
29 Cu, 67 Ni, 2 Fe

=7850
40.705 40.705 37.216 34.890
293.15
=7 8 5 0
32.564
293.15
=7850
18.608
100 0 20 100
173.15 273.15 293.15 373.15
20 100 200 300 400
293.15 373.15 473.15 573.15 673.15 ,,,
8400
8710
16.282 20.934 21.864 26.400 22.097 24.423 27.563 30.238 33.727
2.62
CHAPTER TWO
TABLE 2.34
Density and Thermal Conductivity of Alloys (Continued)
Alloy
Density p (kg/m 3)
Thermal conductivity ~L(W/m K)
123.15 173.15 293.15 373.15 473.15 573.15 673.15
8433
17.678 19.170 25.005 31.401 39.542 45.357 48.846
20
293.15
8200
34.890
0 20 100 200 300 400
273.15 293.15 373.15 473.15 573.15 673.15
8220
17.096 17.445 18.957 20.934 22.795 24.656
80 Ni, 20 Cr
0 20 100 200 300 400 600
273.15 293.15 373.15 473.15 573.15 673.15 873.15
8200
12.212 12.560 13.840 15.584 17.212 18.957 22.562
61 Ni, 15 Cr, 20 Fe, 4 Mn
20 100 200 300 400 600 800
293.15 373.15 473.15 573.15 673.15 873.15 1073.15
8190
11.630 11.863 12.212 12.444 12.677 13.142 13.956
61 Ni, 16 Cr, 23 Fe
0 20 100 200 300 400
273.15 293.15 373.15 473.15 573.15 673.15
8190
11.863 12.095 13.258 14.654 16.049 17.445
70 Ni, 18 Cr, 12 Fe
20
293.15
62 Ni, 12 Cr, 26 Fe
20
293.15
0 100
273.15 373.15

29.308 37.216
20 20 20 20 20 20 20 20 20 20 20 20
293.15 293.15 293.15 293.15 293.15 293.15 293.15 293.15 293.15 293.15 293.15 293.15
8130
34.890 27.912 22.097 18.608 15.119 12.212 11.049 11.049 14.538 19.190 25.586 32.564
Composition (%)
T (°C)
T (K)
New silver
62 Cu, 15 Ni, 22 Zn
150 100 +20 100 200 300 400
Nickel alloy
70 Ni, 28 Cu, 2 Fe
Nickelchrome
90 Ni, 10 Cr
Nickelchrome steel
Nickelsilver

Nickel steel
5 Ni 10 Ni 15 Ni 20 Ni 25 Ni 30 Ni 35 Ni 40 Ni 50 Ni 60 Ni 70 Ni 80 Ni
~8100
11.514 13.491
TABLE 2.34
Density and Thermal Conductivity of Alloys (Continued)
Alloy Nickel steel
Composition (%)
T (°C)
30 Ni, 1 Mn, 0.25 C 36 Ni, 0.8 Mn
Phosphor bronze
20 100
T (K) 293.15 373.15
Density p (kg/m 3)
Thermal conductivity ~. (W/m K)
8190
12.095 13.607
20
293.15
~
12.095
1.4 Ni, 0.5 Cr, 0.3 C
20 100 300 500
293.15 373.15 573.15 773.15
=7850
45.357 44.194 40.705 37.216
92.8 Cu, 5 Sn, 2 Zn, 0.15P
20
293.15
=8766
79.084
91.7 Cu, 8 Sn, 0.3 P
20 100 200
293.15 373.15 473.15
8800
45.357 52.335 61.639
20
293.15
87.8 Cu, 10 Sn, 0.2 P
2 Zn,
87.2 Cu, 12.4 Sn, 0.4 P
~
41.868
20
293.15
8700
36.053
91.5 AI, 4.6 Cu, 1.8 Ni, 1.5 Mg
0 20 100 200
273.15 293.15 373.15 473.15
2800
143.049 144.212 151.190 158.168
84 A1, 12 Si, 1.2 Cu, 1 Ni
0 20 100 200
273.15 293.15 373.15 473.15
2800
134.908 134.908 137.234 139.560
Platinumiridium
90 Pt, 10 Ir
0 100
273.15 373.15
~
30.936 31.401
Platinumrhodium
90 Pt, 10 Rh
0 100
273.15 373.15

30.238 30.587
Rose's metal
50 Bi, 25 Pb, 25 Sn
20
293.15
~
Silumin
8 6 . . . 89 A1, 11 . . . 14 Si
0 20 100
273.15 293.15 373.15
2600
159.331 161.657 170.961
Steel
0.1 C
0 100 200 300 400 600 900
273.15 373.15 473.15 573.15 673.15 873.15 1173.15
7850
59.313 52.335 52.335 46.520 44.194 37.216 33.727
0.2 C
20
293.15
7850
50.009
Piston alloy, cast
16.282
0.6 C
20
293.15
7850
46.520
Bessemer
0.52 C, 0.34 Si
20
293.15
7850
40.240
Tungsten steel
1 W, 0.6 Cr, 0.3 C
20 100 300 500
293.15 373.15 573.15 773.15
7900
39.542 38.379 36.053 33.727
V 1 A steel

20
293.15
m
20.934
V 2 A steel

20
293.15
7860
15.119
Wood's metal
48 Bi, 26 Pb, 13 Sn, 13 Cd
20
293.15
m
12.793
Source"
Ref. 1 with permission. 2.63
2.64
CHAPTER TWO TABLE 2.35
Thermophysical Properties of Miscellaneous Materials Typical properties at 300 K
Description/composition
Structural building materials Building boards Asbestoscement board Gypsum or plaster board Plywood Sheathing, regular density Acoustic tile Hardboard, siding Hardboard, high density Particle board, low density Particle board, high density Woods Hardwoods (oak, maple) Softwoods (fir, pine) Masonry materials Cement mortar Brick, common Brick, face Clay tile, hollow I cell deep, 10 cm thick 3 cells deep, 30 cm thick Concrete block, 3 oval cores Sand/gravel, 20 cm thick Cinder aggregate, 20 cm thick Concrete block, rectangular core 2 core, 20 cm thick, 16 kg Same with filled cores Plastering materials Cement plaster, sand aggregate Gypsum plaster, sand aggregate Gypsum plaster, vermiculite aggregate
Density p (kg/m 3)
Thermal conductivity k (W/m.K)
Specific heat ce (J/kg.K)
1920 800 545 290 290 640 1010 590 1000
0.58 0.17 0.12 0.055 0.058 0.094 0.15 0.078 0.170
1215 1300 1340 1170 1380 1300 1300
720 510
0.16 0.12
1255 1380
1860 1920 2083
0.72 0.72 1.3
780 835
0.52 0.69
m
1.0 0.67 1.1 0.60
1860 1680 720
0.72 0.22 0.25
m
m
m
m
n
m
D
1085
THERMOPHYSICAL
TABLE 2.35
2.65
PROPERTIES
Thermophysical Properties of Miscellaneous Materials (Continued) Typical properties at 300 K
Description/composition Insulating materials and systems Blanket and batt Glass fiber, paper faced
Glass fiber, coated; duct liner Board and slab Cellular glass Glass fiber, organic bonded Polystyrene, expanded Extruded (R12) Molded beads Mineral fiberboard; roofing material Wood, shredded/cemented Cork Loose fill Cork, granulated Diatomaceous silica, coarse powder Diatomaceous silica, fine powder Glass fiber, poured or blown Vermiculite, flakes Formed/foamed in place Mineral wool granules with asbestos/inorganic binders, sprayed Polyvinyl acetate cork mastic, sprayed or troweled Urethane, twopart mixture; rigid foam Reflective Aluminum foil separating fluffy glass mats; 1012 layers; evacuated; for cryogenic application (150 K) Aluminum foil and glass paper laminate; 75150 layers; evacuated; for cryogenic application (150 K) Typical silica powder, evacuated
Density p (kg/m 3)
Thermal conductivity k (W/m.K)
Specific heat cp (J/kg.K)
16 28 40 32
0.046 0.038 0.035 0.038
835
145 105
0.058 0.036
1000 795
55 16 265 350 120
0.027 0.040 0.049 0.087 0.039
1210 1210
160 35O 400 200 275 16 80 160
0.045 0.069 0.091 0.052 0.061 0.043 0.068 0.063
190
0.046
m
1590 1800
m
m
m
m
835 835 1000
0.100
70
0.026
40
0.00016
120 160
0.000017 0.0017
1045
m
m
e', 0
,,,,I
2.68
v
t"I
t~
o
.0
~
oo
O
o~lrlltl
O
"~'~
O
O
.oo.o.
O
IlllLIL
Illllll
Illllll
O
o o.o.o
~
~~ ~.~ ~ ~
O O O O
~ l l ~ oooo
O
o
"O • ~ e"
p' t¢~
c5c~
cSc~ CD~
c~c~
c~c~
c~
~D
c~
c5c~
O
~ L.
illl
.o.
I lll~
o. OOCD
I
I
11
l o
o
t",l ~ o o
I~
o. o
c5
c~
c~
I I
~
II II I I I I o.I
oo.o.
o o o
c~ c~ c~
c5 c5 c5 ~CD~
c~ c5 c~ c~ oh
o.o.o
' ~ ,~ ,"~ "~ ' ' "O
I
II Illl
Illl
d
~l
d
.oqq
d
.qo
ddd
~
og.~
~
O O O O
d
I
0~.~ ~ . ~
~ . ~ N ~
THERMOPHYSICAL PROPERTIES
TABLE 2.35
Thermophysical Properties of Miscellaneous Materials (Continued) Specific heat Cp (J/kg.K)
T (K)
Density p (kg/m 3)
Thermal conductivity k (W/m.K)
Asphalt
300
2115
0.062
Bakelite
300
1300
1.4
1465
872 1672 473 823 1173 478 1145 773 1073 1373 773 1073 1373 478 922 1478 478 922 1478
~ ~ 3010 ~ ~ ~ ~ 2050 ~ 2325 ~ 2645 ~ ~ ~ ~
18.5 11.0 2.3 2.5 2.0 0.25 0.30 1.0 1.1 1.1 1.3 1.4 1.4 1.0 1.5 1.8 3.8 2.8 1.9
m
Clay
300
1460
1.3
880
Coal, anthracite
300
1350
0.26
1260
Concrete (stone mix)
300
2300
Cotton
300
80
300 300 300 300 198 233 253 263 273 283 293
980 840 720 280 ~ ~ ~ ~
300 300
2500 2225
Ice
273 253
920 
0.188 0.203
Leather (sole)
300
998
0.013
Paper
300
930
0.011
1340
Paraffin
300
900
0.020
2890
Description/composition
Other materials
Brick, refractory Carborundum Chrome brick
Diatomaceous silica, fired Fire clay, burnt 1600 K
Fire clay, burnt 1725 K
Fire clay brick
Magnesite
Foodstuffs Banana (75.7% water content) Apple, red (75% water content) Cake batter Cake, fully done Chicken meat, white (74.4% water content)
Glass Plate (soda lime) Pyrex
920
835
960
960
960
1130
1.4
880
0.06
1300
0.481 0.513 0.223 0.121 1.60 1.49 1.35 1.20 0.476 0.480 0.489
3350 3600
1.4 1.4
750 835 2040 1945
2.67
2.68
CHAPTER TWO TABLE 2.35
Thermophysical Properties of Miscellaneous Materials (Continued)
T (K)
Density p (kg/m 3)
Thermal conductivity k (W/m.K)
Specific heat cp (J/kg.K)
Other materials (continued) Rock Granite, Barre Limestone, Salem Marble, Halston Quartzite, Sioux Sandstone, Berea
300 300 300 300 300
2630 2320 2680 2640 2150
2.79 2.15 2.80 5.38 2.90
775 810 830 1105 745
Rubber, vulcanized Soft Hard
300 300
1100 1190
0.012 0.013
2010 m
Description/composition
Sand
300
1515
0.027
800
Soil
300
2050
0.52
1840
Snow
273
110 500
0.049 0.190
Teflon
300 400
2200 
0.35 0.45
Tissue, human Skin Fat layer (adipose) Muscle
300 300 300
Wood, cross grain Balsa Cypress Fir Oak Yellow pine White pine
300 300 300 300 300 300
140 465 415 545 640 435
0.055 0.097 0.11 0.17 0.15 0.11
Wood, radial Oak Fir
300 300
545 420
0.19 0.14
Source: Ref. 6 with permission.
0.37 0.2 0.41
m
n
m
m
2720 2385 2805
2385 2720
THERMOPHYSICAL PROPERTIES
2.69
THERMOPHYSICAL PROPERTIES OF SATURATED REFRIGERANTS TABLE 2.36 Ts (°C)
Saturation Properties for Refrigerant 22 Ps (MPa)
140 120 100 80
0.00023 0.00200 0.01035
60
0.03747
40*
0.10132
20
0.24529
0.00
0.49811
20
0.91041
40
1.5341
60
2.4274
80
3.6627
96.14*
4.9900
p (kg/m 3) 1675.3 L mV 1624.0 L 0.01571 V 1571.7 L 0.12051 V 1518.3 L 0.56129 V 1463.6 L 1.86102 V 1409.1 L 4.7046 V 1346.8 L 10.797 V 1281.8 L 21.263 V 1210.0 L 38.565 V 1128.4 L 66.357 V 1030.5 L 111.73 V 894.8 L 195.69 V 523.8 L 523.8 V
* Boiling point. * Critical point. L, liquid; V, vapor. Extracted from Ref. 8 with permission.
Cp (kJ/kg K)
~t (Pas) x 106
0.445 0.470 ~ 0.497 1.070 0.527 1.076 0.563 1.092 0.606 1.125 0.667 1.171 0.744 1.238 0.849 1.338 1.009 1.528 1.307 2.176 2.268 ~ ~
m
~: (mW/m K)
o (mN/m) 35.70 32.00
m
28.37 m
24.83 m
260.1 210.1 11.80 169.1 136.3
123.1 5.61 114.1 6.93 104.8 8.27 96.2 9.5O 87.8 10.71 79.8 11.90
21.39 18.18
0.00
2.70
CHAPTER TWO
TABLE 2.37
Saturation Properties for Refrigerant 123
L (°c)
Ps (MPa)
p (kg/m 3)
107.15"
0.0000
100
0.00001
80
0.00013
60
0.00081
40
0.00358
1770.9 L 0.00047V 1754.5 L 0.00123V 1709.5 L 0.01195 V 1665.0 L 0.06977V 1619.9 L 0.28314V 1573.7 L 0.87999V 1526.0 L 2.2417V 1476.5 L 4.9169V 1457.5 L 6.3917V 1424.7 L 9.6296V 1369.9 L 17.310V 1311.2 L 29.188V 1246.9 L 46.996V 1174.3 L 73.471V 1088.2 L 113.71V 975.66 L 180.24V 765.88 L 341.95V 550.00L 550.00V
20
0.01200
0
0.03265
20
0.07561
27.46 t
0.10000
40
0.15447
60
0.28589
80
0.48909
100
0.78554
120
1.1989
140
1.7562
160
2.4901
180
3.4505
183.68'
3.6618
* Triple point. *Normal boiling point. *Critical point. L, liquid; V, vapor. Extracted from Ref. 9 with permission.
Cp (kJ/kg K) 0.9287 0.4737 0.9259 0.4863 0.9325 0.5202 0.9319 0.5529 0.9480 0.5850 0.9681 0.6174 0.9902 0.6508 1.0135 0.6861 1.0226 0.6999 1.0384 0.7242 1.0662 0.7667 1.0996 0.8162 1.1432 0.8779 1.2072 0.9643 1.3177 1.1106 1.5835 1.4728 4.5494 5.6622 
(Pas) × 106
(mW/m K)
(~ (mNim)
m
b
b
m
k
m
23.19
b
735.33 9.085 564.55 9.838 442.57 10.562 405.86 10.825 352.37 11.259 233.84 11.939 230.53 12.625 188.08 13.370 153.35 14.289 123.81 15.646
89.320 8.051 83.816 9.089 78.512 10.163 76.581 10.576 73.388 11.291 68.417 12.496 63.563 13.807 58.769 15.260
20.65 18.18 15.77 14.88 13.42 11.15 8.97 6.88 4.91 3.08 1.44 0.14 0.00
THERMOPHYSICAL PROPERTIES TABLE 2.38
Saturation Properties for Refrigerant 134a
T~ (°C)
P~ (MPa)
p (kg/m 3)
Cp (kJ/kg K)
103.30"
0.00039
100
0.0056
80
0.00367
60
0.01591
40
0.05121
26.08*
0.10133
20
0.13273
1591.1 L 0.02817 V 1582.3 L 0.03969 V 1529.0 L 0.23429 V 1474.3 L 0.92676 V 1417.7 L 2.7695 V 1376.6 L 5.2566 V 1358.2 L 6.7845 V 1294.7 L 14.428 V 1225.3 L 27.780 V 1146.7 L 50.085 V 1052.8 L 87.379 V 928.24 L 115.07 V 511.94 L 511.94 V
1.1838 0.5853 1.1842 0.5932 1.1981 0.6416 1.2230 0.6923 1.2546 0.7490 1.2805 0.7941 1.2930 0.8158 1.3410 0.8972 1.4048 1.0006 1.4984 1.1445 1.6601 1.3868 2.0648 2.0122 ~ ~
0
0.2928
20
0.5717
40
1.0165
60
1.6817
80
2.6332
101.06'
2.71
4.0592
* Triple point. * Boiling point. * Critical point. Extracted from Ref. 10 with permission.
~t (Pas) x 10 6 2186.6 6.63 1958.2 6.76 1109.9 7.57 715.4 8.38 502.2 9.20 363.1 9.90 337.2 10.16 265.3 11.02 208.7 11.91 162.7 12.89 124.1 14.15 89.69 16.31 ~ m
K: (mW/m K)
g (mN/m)
m
28.15

27.56
m
24.11
121.1
20.81
111.9 8.19 105.1 9.55 102.4 10.11 93.67 11.96 84.78 13.93 75.69 16.19 66.36 19.14 57.15 24.0 ~
17.66 15.54 14.51 11.56 8.76 6.13 3.72 1.60 0.0
2.72
CHAPTER TWO
TABLE 2.39 T~ (°C)
Saturation Properties for Refrigerant 502 (Azeotrope of R22 and Rl15) Ps (MPa)
p (kg/m 3)
Cp (kJ/kg K)
70
0.02757
60
0.04872
45.42*
0.10132
40
0.12964
20
0.29101
0
0.57313
20
1.0197
40
1.6770
60
2.6014
82.2*
4.075
1557.6 L 1.8501 V 1527.2 L 3.1417 V 1481.5 L 6.2181 V 1464.0 L 7.8315 V 1396.4 L 16.818 V 1322.5 L 32.425 V 1239.4 L 58.038 V 1140.7 L 99.502 V 1010.5 L 171.23 V 561 L 561 V
1.024 1.042 0.574 1.071 0.600 1.082 0.609 1.128 0.649 1.178 0.709 1.234 0.804 1.295 0.949 . . 
la (Pas) x 106
K (mW/m K)
~ (mN/m)


97.9 92.1 90.0 7.11 82.4 8.47 74.8 9.80 67.1 11.21 12.81 
17.41
543.6 469.7 383.9 358.1 282.6 229.2 11.69 12.84 13.99 . .
. .
15.16 14.3.5 11.42 8.64 
. .

* Boiling point. * Critical point. Extracted from Ref. 8 with permission.
TABLE 2.40
Saturation Properties for Ammonia
Tsa t (K) Psat (kPa)
239.75 101.3
250 165.4
270 381.9
290 775.3
310 1424.9
330 2422
350 3870
370 5891
390 8606
400 10,280
Pe, kg/m 3 pg, kg/m 3 he, kJ/kg hg, kJ/kg z~khg,e, k J / k g Cp,e, kJ/(kg K) Cp,g, kJ/(kg K) rie, l,tNs/m2 rig, ktNs/m 2 ~,e (mW/mZ)/(K/m) ~,g (mW/m2)/(K/m) Pre Prg cy, mN/m ~e,e, kK 1
682 0.86 808.0 2176 1368 4.472 2.12 285 9.25 614 18.8 2.06 1.04 33.9 1.90
669 1.41 854.0 2192 1338 4.513 2.32 246 9.59 592 19.8 1.88 1.11 31.5 1.98
643 3.09 945.7 2219 1273 4.585 2.69 190 10.30 569 22.7 1.58 1.17 26.9 2.22
615 6.08 1039.6 2240 1200 4.649 3.04 152 11.05 501 25.2 1.39 1.25 22.4 2.63
584 11.0 1135.7 2251 1115 4.857 3.44 125 11.86 456 28.9 1.36 1.31 18.0 3.18
551 18.9 1235.7 2255 1019 5.066 3.90 105 12.74 411 34.3 1.32 1.34 13.7 4.01
512 31.5 1341.9 2251 899 5.401 4.62 88.5 13.75 365 39.5 1.34 1.49 9.60 5.50
466 52.6 1457.5 2202 744 5.861 6.21 70.2 15.06 320 50.4 1.41 1.70 5.74 8.75
400 93.3 1591.4 2099 508 7.74 8.07 50.7 17.15 275 69.2 1.43 1.86 2.21 19.7
344 137 1675.3 1982 307
Source:
Ref. 3 with permission.
39.5 19.5 252 79.4
0.68 29.2
THERMOPHYSICALPROPERTIES
2.73
ACKNOWLEDGMENT The author gratefully acknowledges the use of a number of thermophysical property tables from previous editions of Handbook of Heat Transfer These include Table 12 from the first edition, prepared by Professor Warren Ibele of the University of Minnesota, and Tables 11, 12, 14, 16, 21, 29, 30, and 33, prepared by Professor Peter Liley of Purdue University.
NOMENCLATURE
Symbol, Definition, SI Units, English Units cp cp:
Cv Dij g k
k: M P Pr R T v
Z
specific heat at constant pressure: kJ/(kg.K), Btu/(lbm'°F) specific heat at constant pressure of saturated liquid: kJ/(kg.K), Btu/(lbm'°F) specific heat at constant volume: kJ/(kgK), Btu/(lbm" °F) diffusion coefficient: m2/s, ft2/s gravitational acceleration: m/s 2, ft/s 2 thermal conductivity: W/(m.K), Btu/(h.ft. °F) thermal conductivity of saturated liquid: W/(m.K), Btu/(hft.°F) molecular weight: kg/(kilogrammole), lbm/(poundmole) pressure: bar, lbf/in 2 (psi) Prandtl number, ktcp/k, dimensionless gas constant: kJ/(kg.K), Btu/(lbm'°R) temperature: K, °R, °C specific volume: m3/kg, ft3/lbm compressibility factor, Pv/RT, dimensionless
Greek Symbols
~,orK rl or ~t v P
thermal diffusivity: m2/s, ft2/s coefficient of volumetric thermal expansion: K 1, °R1 thermal conductivity: W/mK, Btu/h.ft. °F dynamic viscosity: Pa.s, lbm/(h'ft) kinematic viscosity: m2/s, ft2/s density: kg/m 3, lbm/ft3 surface tension: N/m, lbf/ft
REFERENCES 1. K. Raznjavi6, Handbook of Thermodynamic Tables, 2d ed., 392 pp., Begell House, New York, ISBN 1567000460, 1996. 2. N. B. Vargaftik, Y. K. Vinogradov, and V. S. Yargan, Handbook of Physical Properties of Liquids and Gases, 1370 pp., Begell House, New York, ISBN 1567000630, 1996. 3. C. E Beaton and G. E Hewitt, Physical Property Data for the Design Engineer, 394 pp., Hemisphere Publishing, New York, ISBN 0891167390, 1989.
2.74
CHAPTER TWO 4. T. E Irvine Jr. and E Liley, Steam and Gas Tables with Computer Equations, Academic Press, San Diego, ISBN 0123740800, 1984. 5. G. E Hewitt, ed., Handbook of Heat Exchanger Design, Begell House, New York, ISBN 1567000002, 1992. 6. E E Incropera and D. P. De Witt, Fundamentals of Heat and Mass Transfer, 3d ed., Wiley, New York, ISBN 0471042711X, 1990. 7. L. Hoar, J. S. Gallagher, and G. S. Kell, NBS/NRC Steam Tables, Hemisphere Publishing, New York, ISBN 0891163549, 1984. 8. American Society of Heating and Air Conditioning Engineers, 1993 ASHRAE Handbook, Fundamentals, SI Edition, ISBN 0910110972, 1993. 9. International Institute of Refrigeration, Thermodynamic and Physical Properties, R123, Paris, ISBN 2903633703, 1995. 10. International Institute of Refrigeration, Extended Thermophysical Properties, R134a, Paris, ISBN 2903633738, 1995.
SELECTED ADDITIONAL SOURCES OF THERMOPHYSICAL PROPERTIES 1. D. S. Viswanath and G. Natarajan, Data Book on the Viscosity of Liquids, 990 pp., Hemisphere Publishing, New York, ISBN 0891167781, 1989. 2. Y. S. Touloukian, R. W. Powell, C. Y. Ho, and E G. Klemens, Thermophysical Properties ofMatter, vol. 1, Thermal Conductivity, Metallic Elements and Alloys, 1469 pp., IFI/Plenum, New York, SBN 306670216, 1970. 3. Y. S. Touloukian, R. W. Powell, C. Y. Ho, and E G. Klemens, Thermophysical Properties ofMatter, vol. 2, Thermal Conductivity, Nonmetallic Solids, 1172 pp., IFI/Plenum, New York, SBN 306670224, 1970. 4. Y. S. Touloukian, E E. Liley, and S. C. Saxena, Thermophysical Properties of Matter, vol. 3, Thermal Conductivity of Nonmetallic Liquids and Gases, 531 pp., IFI/Plenum, New York, SBN 306670232, 1970. 5. Y. S. Touloukian, S. C. Saxena, and E Hestermans, Thermophysical Properties of Matter, vol. 11, Viscosity, 643 pp., IFI/Plenum, New York, ISBN 0306670313, 1975. 6. B. Platzer, A. Polt, and G. Maurer, Thermophysical Properties of Refrigerants, SpringerVerlag, Berlin, 1990. 7. J. T. R. Watson, Viscosity of Gases in Metric Units, National Engineering Laboratory, HSMO, Edinburgh, 1972. 8. R. E Danner and T. E. Daubert, Physical and Thermodynamic Properties of Pure Chemicals, DIPPR, Hemisphere Publishing, New York, 1989. 9. Warmeatlas, VDIVerlag GMB H, Dusseldorf, 1984. 10. A. L. Harvath, Physical Properties of Inorganic Compounds SI Units, Crane, Russak & Co., New York, 1975. 11. C. L. Yaws, Physical Properties, McGrawHill, New York, 1972.
CHAPTER 3
CONDUCTION AND THERMAL CONTACT RESISTANCES (CONDUCTANCES) M. M. Yovanovich University of Waterloo
INTRODUCTION When steadystate conduction occurs within and outside solids, or between two contacting solids, it is frequently handled by means of conduction shape factors and thermal contact conductances (or contact resistances), respectively. This chapter covers the basic equations, definitions, and relationships that define shape factors and the thermal contact, gap, and joint conductances for conforming, rough surfaces, and nonconforming, smooth surfaces. Shape factors for two and threedimensional systems are presented. General expressions formulated in orthogonal curvilinear coordinates are developed. The general expression is used to develop numerous general expressions in several important coordinate systems such as (1) circular, elliptical, and bicylinder coordinates and (2) spheroidal coordinates (spherical, oblate spheroidal, and prolate spheroidal). The integral form of the shape factor for an ellipsoid is presented and then used to obtain analytical expressions and numerical values for the shape factors of several isothermal geometries (spheres, oblate and prolate spheroids, circular and elliptical disks). It is demonstrated that the dimensionless shape factor is a weak function of the geometry (shape and aspect ratio) provided that the square root of the total active surface area is selected as the characteristic body length. A general dimensionless expression is proposed for accurate estimations of shape factors of threedimensional bodies such as cuboids. Shape factor expressions are presented for twodimensional systems bounded by isothermal coaxial (1) regular polygons, (2) internal circles and outer regular polygons, and (3) internal regular polygons and outer circles. A method is given for estimating the shape factors of systems bounded by two isothermal cubes and other combinations of internal and external geometries. The shape factor results of this chapter are used in the chapter on natural convection to model heat transfer from isothermal bodies of arbitrary shape. Transient conduction within solids and into full and halfspaces is presented for a wide range of two and threedimensional geometries. Steadystate and transient constriction (spreading) resistances for a range of geometries for isothermal and isoflux boundary conditions are given. Analytical solutions for halfspaces and heat flux tubes and channels are reported. Elastoconstriction resistance and gap and joint resistances for line and point contacts are presented. Contact conductances of conforming rough surfaces that undergo (1) elastic, (2) 3.1
3.2
CHAPTER THREE
plastic, and (3) elastoplastic deformation are reported. The gap conductance integral is presented. The overall joint conductance is considered. Analytical solutions and correlation equations are presented rather than graphic results. The availability of many computer algebra systems such as Macsyma, MathCad, Maple, MATLAB, and Mathematica, as well as spreadsheets such as Excel and Quattro Pro that provide symbolic, numerical, and plotting capabilities, makes the analytical solutions amenable to quick, accurate computations. All equations and correlations reported in this chapter have been verified in Maple worksheets and Mathematica notebooks. These worksheets and notebooks will be available on my home page on the Internet. Some spreadsheet solutions will also be developed and made available on the Internet.*
BASIC EQUATIONS, DEFINITIONS, AND RELATIONSHIPS Shape factors of isothermal, threedimensional convex bodies having complex shapes and small to large aspect ratios are of considerable interest for applications in the nuclear, aerospace, microelectronic, and telecommunication industries. The shape factor S also has applications in such diverse areas as antenna design, electron optics, electrostatics, fluid mechanics, and plasma dynamics [27]. In electrostatics, for example, the capacitance C is the total charge Oe required to raise the potential ~e of an isolated body to the electrical potential Ve, and the relationship between them is (e.g., Greenspan [27], Morse and Feshbach [68, 69], Smythe [98], and Stratton [111])
Qe ffme. ~n O~e dA
C=~e =
where e is the permittivity of the surrounding space, (~eis the nondimensional electric potential, n is the outwarddirected normal on the surface, and A is the total surface area of the body. Mathematicians prefer to deal with the capacity C* of a body, which they [81,113] define as
1~ ffa ~n a~e dA C*=T~
ShapeFactor, Thermal Resistance, and Diffusion Length. The shape factor S, the thermal resistance R, and the thermal diffusion length A are three useful and related thermal parameters They are defined by the following relationships: 1
A
Q
S  kR A  k(T0 T~)
_ ff  JJA
~)~ 
~n dZ
(3.1)
where k is the thermal conductivity, To is the temperature of the isothermal body, T. is the temperature of points remote from the body, and ~ is the dimensionless temperature
( T(7.)  T~)/(To  T~). The relationships between the shape factor S, the capacitance C, and the capacity C* are S = __C= 4nC*
(3.2)
E
The three parameters have units of length. Analytical solutions are available for a small number of geometries such as the family of geometries related to the ellipsoid (e.g., sphere, oblate and prolate spheroids, elliptical and circular disks). Precise numerical values of S for other axisymmetric convex bodies have been obtained by various numerical methods such as that proposed by Greenspan [27] and that proposed by Wang and Yovanovich [123]. * The Internet address is mmyovemhtl.uwaterloo.ca.
CONDUCTION AND THERMAL CONTACT RESISTANCES (CONDUCTANCES)
3.3
Chow and Yovanovich [15] showed, by analytical and numerical methods, that the capacitance is a slowly changing function of the conductor shape and aspect ratio provided the total area of the conductor is held constant. Wang and Yovanovich [123] showed that the dimensionless shape factor
S~A S,:~ ,:~ ffA ~n 3(~)dZ A  Z
(3.3)
where the characteristic scale length ~ was chosen to be ~ as recommended by Yovanovich [133], Yovanovich and Burde [134], and Yovanovich [144146], when applied to a range of axisymmetric, convex bodies is a weak function of the body shape and its aspect ratios. This chapter reports and demonstrates, through inclusion of additional accurate numerical results of Greenspan [27] for complex body shapes such as cubes, ellipsoids, and circular and elliptical toroids, a lens that is formed by the intersection of two spheres such that S~A is a relatively weak function of the body shape and its aspect ratios. This chapter also introduces the geometric length A, called the diffusion length, and shows that this physical length scale is closely related to the square root of the total body surface area when the body is convex. The dimensionless geometric parameter X/~/A is proposed as an alternate parameter for determination of shape factors of complex convex bodies.
Shape Factors
Formulation of the Problem in General Coordinates. Consider the steady flow of heat Q from an isothermal surface A1 at temperature T1 through a homogeneous medium of thermal conductivity k to a second isothermal surface A2 at temperature T2(T1 > T2). The stationary temperature field depends on the geometry of the isothermal boundary surfaces. When these isothermal surfaces can be made coincident with a coordinate surface by a judicious choice of coordinates, then the temperature field will be onedimensional in that coordinate system. In other words, heat conduction occurs across two surfaces of an orthogonal curvilinear parallelepiped (Fig. 3.1a), and the remaining four coordinate surfaces are adiabatic. Let the general coordinates ul, U2, /'/3 be so chosen that T = T(Ul) and, therefore, t)T[~u2 = OT/Ou3 = 0. Under these conditions, the heat flux vector will have one component in the Ul direction: q, =k(dT/ds) =k(dT/V~g~ du,) where ~ is the metric coefficient in the Ul direction. The metric coefficients are defined by the general line element ds expressed in terms of the differentials of arc lengths on the coordinate lines [67] (ds) 2= gl(dUl) 2 + gz(du2) 2 + g3(du3) 2 ua
The product terms such as dui duj (i :~ j) do not appear because of the orthogonality property of the chosen coordinate system. These metric coefficients can also be generated by means of the following formula [67]:
ds a
u2
ds2 U1
FIGURE 3.1a
Orthogonal curvilinear parallelepiped.
g~  (axlaui) 2 + (ay/aui) 2 + (azlaui) 2
i=1,2,3
provided that the Cartesian coordinates x, y, z can be expressed in terms of the new coordinates ul, u2, u3 by the equations X = X(Ul, U:, U3),
y  y(Ul, U2, U3),
Z = Z(Ul, U2, U3)
3.4
CHAPTERTHREE The elemental coordinate surface located at u~, which is orthogonal to the therefore
U 1 direction,
is
dA1 = ds2 ds3 = V~zg3 du2 du3 and the heat flow per unit time through this surface into the volume element is, according to Fourier's law of conduction,
Q1 =  k da,(dT/dsl) =k(V/g/g~)(dT/dul) du2 du3 where g  glgzg3 [67]. The heat flow rate out of the volume element is
Q~ + (dQ~/ds~) ds~ = Q~ + (dQ~/du~) dul neglecting the higherorder terms of the Taylor series expansion of Q1 about Ul. The net rate of heat conduction out of the volume element in the Ul direction is
(d/du~)[k(X/g/gl)(dr/dUl)] dUl duz du3 For steadystate conditions with no heat sources within the volume element, the Laplace equation in general coordinates is obtained by dividing by the elemental volume V~ dUl du2 du3 and equating to zero. Therefore,
(1/X/g)(d/dul )[k(X/g/g~ )(dT/du~ ) ] = 0 The above equation is the governing differential equation and it is nonlinear when k is a function of temperature. The isothermal temperature boundary conditions are
ul =a, Ul =b,
T= T1 T= T2
The above equation can be reduced to a linear differential equation by the introduction of a new temperature 0 related to the temperature Tby the Kirchhoff transformation [4, 11]: 0 = (l/k0) I, T° k d r where k0 denotes the value of the thermal conductivity at some convenient reference temperature, say T = O. It follows that
dO/dUl  (k/ko)(dT/du,) and, therefore, we have
(d/dul)[(N/ g/gl)(dO/dUl) ] = 0 after multiplying through by ~/g/ko. The boundary conditions become U 1 = a,
0 : 01 : ( 1 / k o ) fo T1 k d T
Ul  b ~
0 = 02 = (l/k0) foT2k d r
The solution of the linear equation is
0 = C1
(gl/V~) dul + C2
CONDUCTION AND THERMAL CONTACT RESISTANCES (CONDUCTANCES)
3.5
where the constants C 1 and C2 are obtained from
and
C 2 = 01  C 1
fOa (gl/X/g) dUl
Temperature Distribution in Orthogonal Curvilinear Coordinates.
The temperature dis
tribution in orthogonal curvilinear coordinates is
f;l (gl/'~/g)
01  0
d/'il

o, o~
,
a < U1 < b
(3.4)
f~ (g, lXl~) du,
The local heat flux in the Ul direction is k0 dO
k0(01  02)
_
ql X/~l dUl
The heat flow rate through the elemental surface dA1
qldZl
(3.5)
gv~2g3f~ (gl/X/g) dua
is
f k0(01(gl/X/g)02)du1duzdu3
(3.6)
The total heat flow rate through the thermal system can be obtained by integration between the appropriate limits. Therefore,
Q= k°(OaOz) f
2 fa~
(ga/X/g)dU2dU3dua
(3.7)
An examination of the above equation shows that k0(01 02) =
k dT= ka(Zl  Z2)
(3.8)
where ka, the average value of the thermal conductivity, is given by k~ = ko[1 + (1~/2)(T1 + 7"2)]
(3.9)
if k = k0(1 + 13T).
Shape Factor and Thermal Resistance in Orthogonal Curvilinear Coordinates.
The definition of thermal resistance of a system (total temperature drop across the system divided by the total heat flow rate) yields the following general expression for the thermal resistance R and the conduction shape factor S:
S_(Rka)_X__ff f~ du~du~ u~ .~
(3.10)
(gllW~) du,
The right side of the previous equation has units of length and depends on the geometry of the body only. There are several very important and useful coordinate systems that can be used to solve many seemingly complex conduction problems. Since each coordinate system
3.6
CHAPTER
THREE
has three principal directions, there are three sets of shape factors corresponding to each of these directions. The conduction shape factors for several coordinate systems are given in the following section. This section by no means represents the total n u m b e r of coordinate systems that are amenable to this type of analysis. It does, however, contain the most frequently used coordinate systems.
rd2 I I
,~',z)
x
¥=0
General Expressions f o r Conduction Shape Factors i i
Circular Cylinder Coordinates (r, ~, z): Fig. 3.1b r direction. Let Ul = r; therefore, u2 = ~, u3 = z, and gl/X/g = 1/r. If
FIGURE 3.1b Circular cylinder coordinates.
a b ___c was given in the integral form [113]: 1 1 c~ dv C*  2 Jo V~(a 2 + v)(b 2 + v)(c 2 + v)
(3.29)
where v is a dummy variable. This expression is used to develop the expression for the dimensionless shape factor of isothermal ellipsoids. Since S = 4rtC*, one can set the space variable v = a2t, where t is a dimensionless variable. Next we normalize the two smaller axes b, c of the ellipsoid with respect to the largest semiaxis a such that 13= b/a and 7 = c/a. This leads to the following dimensionless integral [150]: 8rta I~ dt S  1(13, 7) = M'(1 + t)(f~ 2 + 0(72+ t)'
0 _ x2 and Y = x2 if x2 > x~. The shape parameter n was found to lie in the range 1 < n < 1.2 for a very wide range of body shapes. The results obtained through this method show agreement to within about 5 percent with those obtained from numerical or existing analytical techniques.
Shape Factors for TwoDimensional Systems The shape factors for twodimensional systems are available for (1) long cylinders bounded by homologous, regular polygons having N sides (Fig. 3.5), (2) long cylinders bounded internally by circles and bounded externally by regular polygons (Fig. 3.6), and (3) long cylinders bounded internally by regular polygons and bounded externally by circles (Fig. 3.7). In all three cases, as the number of sides N of the regular polygon becomes large (N > 10), the shape factor approaches the shape factor for the system bounded by two coaxial circular cylinders.
N=3
N=4
N=5
~ilii N=6
N=3
N=4
N=5
i!
N~oo
FIGURE 3.5 Regions bounded by regular polygons.
N=6
N~oo
FIGURE 3.6 Regions bounded by inner circles and outer regular polygons.
CONDUCTION AND THERMAL CONTACT RESISTANCES (CONDUCTANCES)
N=3
N=4
3.21
N=5
N=6
N~oo
FIGURE 3.7 Regions bounded by inner regular polygons and outer circles.
Regular Polygon Inside a Regular Polygon. The shape factor of twodimensional regions bounded internally and externally by isothermal regular polygons of sides N > 3 (Fig. 3.5) is obtained by means of the general expression S
L

4n
N>3
[
(D/d)2 I ] ' In 1 + (N/n) tan (n/N)
(3.61)

where d and D are the diameters of the inscribed circles of the inner and outer polygons, respectively. For the region bounded by two squares (N = 4), the general expression reduces to S L  In [1 +
4n
(n/4)[(D/d) 2 1]]
(3.62)
The square/square problem has a complex analytical solution that requires the numerical computation of complete elliptical integrals of the first kind (see Ref. 8). For a large range of the parameter d/D, the analytical solution provides an accurate asymptotic expression: S [1 + (d/D)] 8 In 2, L  4 [1  (d/D)]  n
0.3 < 
d/O < 1 
(3.63)
Two correlation equations based on electrical measurements were reported by Smith et al. [93]: S 2n L  0.93 In [0.947(D/d)] ' and
S 2n L  0.785 In (D/d)'
D ~d > 1.4
D ~ < 1.4
(3.64)
(3.65)
Circle Inside a Regular Polygon. Several expressions have been developed for this system (Fig. 3.6). Two relationships that give results to within a fraction of 1 percent are given. The first is [52] S 2n  ~ L In [As[3]
(3.66)
3.22
CHAPTERTHREE where 13= b/a > 1 is the ratio of the radius of the inner circle to the radius of the inscribed circle. The parameter As is obtained by means of a numerical integration of the integral
As =
iSo
1
(1 + uN) 2IN
du,
N >3
(3.67)
Laura and Susemihl [52] gave several values of the parameter As for several values of N. The alternate relationship [91] is S 4N [V'A2+I rt] L  A V ' A z + 1 tan1 ~ tan ~ ,
N>3
(3.68)
where the parameter A = V'2 In 13. The second relationship does not require a numerical integration.
Regular Polygon Inside a Circle. Numerous analytical, numerical, and experimental studies have produced results for shape factors and resistances for regions bounded internally by isothermal regular polygons of N sides where N > 3 and externally by an isothermal circle (Fig. 3.7). Lewis [55] gave the following general analytical result S 2n L In [AN(D/d)]
(3.69)
where the coefficients are given by
AN = [(N/N ~ / ' N  2)]2/N(N 2) ~/N,
N >_3
(3.70)
The above relationship is limited to the range 0 < d/D < cos n/N, where d and D are the diameters of the inscribed circle and the outer circle, respectively. The relationship gives accurate values of S for small values of d/D and for all values of N. The accuracy decreases for values of d/D ~ cos n/N for small values of N. Ramachandra Murthy and Ramachandran [82] obtained two empirical correlation equations for regions bounded by squares and hexagons Their correlation equations were developed from electrical measurements and have been shown to be in good agreement with the above relationship for a limited range of values of d/D.
Polygons With Layers. Hassani et al. [34] presented a procedure for obtaining a close upper bound for shape factors for a uniform thickness twodimensional layer on cylinders having cross sections of the following forms: an equilateral triangle, a square, a rhombus, and a rectangle (Fig. 3.8). The shape factor per unit length of the inner cylinder is obtained from S 2n L  In [1 + (2nB/Pi)]
(3.71)
B
°
(a)
(b)
FIGURE 3.8 Polygonswith uniform thickness layers.
(c)
(d)
CONDUCTION AND THERMAL CONTACTRESISTANCES(CONDUCTANCES)
3.23
where B is the layer thickness and P; is the perimeter of the inner boundary. The accuracy of the above relationship was verified by comparison of the predicted values against numerical values for the layer thicknesstoside dimension range 0.10 < B/L < 3.00. The proposed relationship overpredicts the shape factor by approximately 13 percent.
TRANSIENT CONDUCTION Introduction Transient conduction internal and external to various bodies subjected to the boundary conditions of the (1) first kind (Dirichlet), (2) second kind (Neumann), and (3) third kind (Robin) are presented in this section. Analytical solutions are presented in the form of series or integrals. Since these analytical solutions can be computed quickly and accurately using computer algebra systems, the solutions are not presented in graphic form.
Internal Transient Conduction Internal onedimensional transient conduction within infinite plates, infinite circular cylinders, and spheres is the subject of this section. The dimensionless temperature ¢ = 0/0i is a function of three dimensionless parameters: (1) dimensionless position ~ = x/Y, (2) dimensionless time Fo = ott/~g2, and (3) the Biot number Bi = h~/k, which depends on the convective boundary condition. The characteristic length ~ is the halfthickness L of the plate and the radius a of the cylinder or the sphere. The thermophysical properties k, a, the thermal conductivity and the thermal diffusivity, are constant. The basic solutions for the plate and the cylinder can be used to obtain solutions within rectangular plates, cuboids, and finite circular cylinders. The equations and the initial and boundary conditions are well known [4, 11, 23, 28, 29, 38, 49, 56, 80, 87]. The solutions presented below follow the recent review of Yovanovich [151]. The Heisler [36] cooling charts for dimensionless temperature are obtained from the series solution: ~o
¢ = ~" A, exp(5 ] Fo)S,(5,~)
(3.72)
n=l
with the temperature Fourier coefficients A, for the plate, cylinder, and sphere, respectively, given in Table 3.5. The spatial functions for the three basic geometries are given in Table 3.6. The eigenvalues 5, are the positive roots of the characteristic equations found in Table 3.6 where Bi, the physical parameter, ranges between 0 and oo.
TABLE 3.5
Fourier Coefficients for Temperature and Heat Loss
A.
B. 2Bi 2
Plate
2 sin 5. 5, + sin 5, cos 5,
52(Bi 2+ Bi + 52)
Cylinder
2./,(5.) 5.[./2(5.) + J2(5.) l
4Bi 2 52(52 + Bi 2)
Geometry
Sphere
(_1) ,,+,
2Bi [52 + (Bi 1)21~a (5.2+ Bi2Bi)
6Bi2 2 2+Bi 2_ Bi) 5.(5.
3.24
CHAPTER
THREE
TABLE 3.6 Space Functions and Characteristic Equations i
Geometry
S.
Characteristic equation
Plate Cylinder
cos (~5.~) J0(~i.~)
x sin x = Bi cos x xJl(x) = Bi Jo(x)
Sphere
sin (~.~) (~.~)
(1  Bi) sin x = x cos x
The Grober charts [29] for the heat loss fraction Q/Qi, where Qi = internal energy, are obtained from the series solution:
pcpVOiis the initial total
,,o
Q  1  ~ ' B, exp(~5 ] Fo) ai
(3.73)
n=1
The Fourier coefficients B, are given in Table 3.5 for the three geometries. These coefficients depend on the Biot number.
Lumped Capacitance Model When the Biot number is sufficiently small (Bi < 0.2), the series solutions converge to the first term for all values of Fo > 0. The values of the Fourier coefficients A1 and B1 approach 1, and the dimensionless temperature and the heat loss fraction approach the general lumped capacitance solutions (~ = e(hS/pcpv) t
and
Q  1  e (hs/pcpv)t ai
where S and V are the total active surface area and the volume. The lumped capacitance solutions for the three geometries are given in Table 3.7.
TABLE 3.7
Lumped Capacitance Solutions Bi < 0.2
Q/Qi
Geometry
~
Plate Cylinder Sphere
e  B i Vo
1
e2Bi ro
1  e 2Bi Fo
e3Bi Vo
1  e 3Bi vo
e Bi Fo
Heisler and Grober ChartsSingleTerm Approximations The Heisler [36] cooling charts and the Grober [29] heat loss fraction charts for the three geometries can be calculated accurately by the singleterm approximations [28, 56] 0 
exp(512
= A1
0i and
O
Qi
 1

O 1
Fo)Sl(~l~)
exp(5 2 Fo)
CONDUCTION AND THERMAL CONTACTRESISTANCES (CONDUCTANCES)
3.25
Asymptotic Values of First Eigenvalues, Correlation Parameter n, and Critical Fourier Number
TABLE 3.8
Geometry
Bi ~ 0
Plate Cylinder Sphere
81>~ 81>~ 81>3 V ~
Bi >oo
n
Foc
81 >re/2 2.139 0.24 81>2.4048255 2.238 0.21 81>rt 2.314 0.18
for all values of the Biot number, provided Fo > Foc. The critical values of the Fourier number for the three geometries are given in Table 3.8. The first eigenvalue can be computed accurately by means of the correlation [151] 81,oo
8~ = [1 + (8~/~ ~1~/~ tvl,O) J which is valid for all values of Bi. The asymptotic values of the first eigenvalues corresponding to very small and very large values of Bi are given in Table 3.8. The correlation parameter n is also given in Table 3.8. The correlation equation predicts values of 81 that differ from the exact values by less than 0.4 percent. For Fo < Foc, additional terms in the series solutions must be included. It is therefore necessary to use numerical methods to compute the higherorder eigenvalues 8n that lie in the intervals nrc < 8n < (n + 1/2)rt for the plate and (n  1)r~< 8n < nrt for the cylinder and the sphere. Computer algebra systems are very effective in computing the eigenvalues. Chen and Kuo [12] have presented approximate solutions of O/Oiand Q/Qi for the plates and long cylinders. The accuracy of these solutions is acceptable for engineering calculations.
Multidimensional Systems The basic solutions for the infinite plates and infinitely long cylinders can be used to obtain solutions for multidimensional systems such as long rectangular plates, cuboids, and finite circular cylinders with end cooling. The texts on conduction heat transfer [4, 11, 23, 29, 38, 49, 56, 87] should be consulted for the proofs of the method and other examples. Langston [51] showed how to obtain the heat loss from multidimensional systems using the onedimensional solutions given above. Twodimensional systems such as long rectangular plates and finite circular cylinders are characterized by two Biot numbers, two Fourier numbers, and two dimensionless position parameters. Threedimensional systems such as cuboids are characterized by three sets of values of Bi, Fo, and ~ corresponding to the three cartesian coordinates. When the two or three sets of Fourier numbers are greater than the critical values given in Table 3.8, then the firstterm approximate solutions dis0 cussed above can be used to develop composite solutions. Yovanovich [151] has discussed the application of the basic solutions to long rectangular plates, cuboids, and finitelength circular cylinders. "~2X I
Dimensionless Temperature and Heat Loss Fraction for FiniteLength Cylinders. The finitelength circular cylin
FIGURE 3.9 Circularcylinder of finite length.
der of radius R and length 2X, shown in Fig. 3.9, has constant properties and is cooled through the sides and the two ends by uniform film coefficients hr and hx, respectively. The sys
3.26
CHAPTER THREE
tem is characterized by four physical parameters: Bix = hxX/k, Bic = hrR/k, Fox = ou/X 2, and (zt/R 2. The dimensionless temperature within the cylinder is obtained from the product solution:
F o r 
(~xr:(~x(~r where ~x and ~)r a r e the solutions corresponding to the x and r coordinates, respectively. According to Langston [51], the heat loss fraction can be obtained from
(Q)xr=(Q)x+(Q)r(Q)x'(Q)r where Qi = pCp2nXR20i. The subscripts x and r denote solutions corresponding to the x and r coordinates, respectively.
Transient OneDimensional Conduction in HalfSpaces The analytical solutions for transient onedimensional conduction in halfspaces x > 0 are well known and appear in most heat transfer texts. The solutions are given here for completeness and to review important characteristics of the solutions.
Equation and Initial and Boundary Conditions.
The diffusion equation and the initial and boundary conditions are presented first, followed by the solutions with some important relationships. 320
3x 2
1 30 tx 3 t '
t > 0,
x>0
(3.74)
where 0 = T(x, t )  Ti is the instantaneous temperature rise within the halfspace. The initial condition is 0=0,
t=0,
x>0
(3.75)
and the boundary condition at remote points in the halfspace is 0 >0,
t>0,
x >oo
(3.76)
There are three options for the boundary condition at x = 0.
Dirichlet Condition O= To Ti,
t>0
(3.77)
t>0
(3.78)
where To is the fixed temperature on the surface.
Neumann Condition 30 ~=~ 3x
q0 k'
where q0 is the constant heat flux imposed on the surface.
Robin Condition 30 ~x 
h k ( %  0),
t>0
(3.79)
CONDUCTION AND THERMAL CONTACT RESISTANCES (CONDUCTANCES)
3.27
where h is the constant film or contact conductance that connects the surface to the heat source and 0i = Ti  Ti is the constant temperature difference between the heat source temperature and the initial temperature. The solutions have been obtained by several analytical methods. Introducing the dimensionless parameters ~ = ( T(x, t)  Ti)/(To Ti), 11 = x/(2V'at), and Bi = (h/k)V~t, the three solutions are given below.
Dirichlet Solution = erfc (11),
11 > 0
(3.80)
which gives the instantaneous and timeaverage surface fluxes qo(t) = ~
1
k(ToTi) V~
2 g(ToTi) q0(t) = V ~ V~
and
(3.81)
(3.82)
The time average value of any function f(t) is defined as f(t) = (l/t) ~o f(t) dt. Neumann Solution
k[ T(x, t )  T,] 2qoV~
1 = V ~ exp(rl2)  rl erfc (11)
(3.83)
which gives the following relationships for the instantaneous and timeaverage surface temperatures kiT(0, t)  Ti] 1 2q0V~ : V~ and
k [ T ( 0 )  Ti] 2 2q0V~ = 3V~
(3.84)
(3.85)
Robin Solution
T(x, t )  T,. r~r/
= erfc (11) exp(2rl Bi + Bi 2) erfc (11 + Bi)
(3.86)
which yields the following two relationships for the instantaneous surface temperature and the surface heat flux:
T(0, 0  T,
TiTi and
= 1  e x p ( B i 2) erfc (Bi)
q°(t)Vr~t~t = Bi exp(Bi 2) erfc (Bi)
k(Tf Ti)
(3.87)
(3.88)
For large values of the parameter Bi > 100, the Robin solution approaches the Dirichlet solution. The three onedimensional solutions presented above give important shorttime results that appear in other solutions such as the external transient threedimensional conduction from isothermal bodies of arbitrary shape into large regions. These solutions are presented in the next section.
3.28
CHAPTER THREE
External Transient Conduction From Long Cylinders Introduction. Transient onedimensional conduction external to long circular cylinders is considered in this section. The conduction equation, the boundary and initial conditions, and the solutions for the Dirichlet and Neumann conditions are presented. The conduction equation for the instantaneous temperature rise O(r, t)  Ti in the region external to a long circular cylinder of radius a is 320 1 30 1 30 37 + r 3 r  a at'
t>O,
r>a
(3.89)
The initial condition is 0=0,
t =0
and the boundary condition at remote points in the full space is 0 + 0,
r+ oo
Two types of boundary conditions at the cylinder boundary r = a will be considered: (1) Dirichlet and (2) Neumann.
Dirichlet Condition 0 = T0 Ti,
t>0
where To is the fixed surface temperature.
Neumann Condition
20 3r
~
qo k'
t>0
where q0 is the constant heat flux on the cylinder surface. The solutions for the two boundary conditions are reported in Carslaw and Jaeger [11]. The solutions were obtained by means of the Laplace transform method. The solutions are given as infinite integrals and the integrand consists of Bessel functions of the first and second kinds of order zero.
Dirichlet Solution 0 _ 1 + 2 F eF° ~2 Jo(fJr/a)Yo(fS) Yo(fJr/a)Jo(fS) dr3 ~ Jo J2(13) + Y2(13) 13
(3.90)
0i
with Fo = (xt/a 2 > 0. The integral can be evaluated accurately and easily for all dimensionless times using computer algebra systems.
Instantaneous Surface Heat Flux. at r = a, is given by the integral
The instantaneous surface flux, defined as q(t) =kaO/3r
aq(t) 4 f : e_FO~2 dr3 kOi  71;2 [j02(~) + y2(~)]~
(3.91)
Carslaw and Jaeger [1 1] presented shorttime expressions for the instantaneous temperature rise and the surface heat flux.
ShortTime Temperature Rise 0 1 [p1 ] (p1)~/~o ierfc [ p  1 0 i  ~ erfc 2x/Foo + 4133/2
2x/Go]+
(929792 ) [P 1 ] 32pS/2 i2erfc 2N/~o
C O N D U C T I O N AND T H E R M A L CONTACT RESISTANCES (CONDUCTANCES)
3.29
with p = r/a > 1. The special functions ierfc (x) and i2erfc (x) are integrals of the complementary error function and are defined in Carslaw and Jaeger [11]. ShortTime Surface Heat Flux.
The instantaneous surface heat flux is given by
aq(t) kOi
V ~ /  ~ o + ~   ~
+~Fo
The first term corresponds to the halfspace solution when the dimensionless time is very small, i.e., Fo < 103. N e u m a n n Solution. The instantaneous temperature rise for arbitrary dimensionless time Fo > 0 at arbitrary radius r/a > 1 is given by the integral solution:
kO _  2 If (1 eF°~) Jo(fSr/a)Y,([5) Yo(~F/a)Jl(~) dr3 qo rr j2 (~) + y2 (~) ~2 Carslaw and Jaeger [11] presented an approximate shorttime solution for arbitrary radius:
plJ
k0 2 ~/~o{ierfc[ q0  V~ 2V~o with p = r/a > 1 and Fo is given by
=

(3p+1)i2erfc [ p  1 ]} 4p 2~o
off/a 2. The instantaneous surface temperature rise 00 for short times k00_ 2 ~ /  ~ o _ 1 q0 ~ 2
Transient External Conduction From Spheres Introduction. Solutions of transient conduction from a sphere of radius a into an isotropic space whose properties are constant and whose initial temperature Ti is constant are considered here. The dimensionless equation is
a2(~
ap 2 +
2 a~ p ap
a~ 
aFo'
Fo > 0,
p>1
(3.92)
where ¢~is the dimensionless temperature, p = r/a, and Fo = m/a 2 is the dimensionless time defined with respect to the sphere radius. Solutions are available in Carslaw and Jaeger [11] for three boundary conditions: (1) the Dirichlet condition where T(a, t)= To, (2) the Neumann condition where OT(a, t)/br =qo/k, and (3) the Robin condition where OT(a, t)/br =  ( h / k ) [ T r  T(a, t)]. The thermophysical parameters To, TI, q0, and h are constants. These boundary conditions in dimensionless form are ~ = 1, O¢/0p =1, and 0~/0p =Bi(1 ¢~) with Bi = ha/k for the three boundary conditions, respectively. The three definitions of dimensionless temperature are presented below. The three dimensionless solutions [11] are: Dirichlet Solution
lerfc/2 / p1
¢=p w h e r e , = ( T(r, t )  Ti)/( To  Ti).
3.30
CHAPTERTHREE
Neumann Solution
, = P erfc
 P exp(p  1 + Fo) erfc
+ x/FTo
(3.94)
w h e r e , = k(T(r, t)  Ti)/(aqo). Robin Solution Bi
1
~ Bi+l
p
erfc( p  1
p1 Bi i exp[(Bi + 1)(p  1) + (Bi + 1)2 Fo] erfc 2X/~o + (Bi + 1)%/Foo] (3.95) Bi+l p where ~ = ( T(r, t )  Ti)/( T I  Ti).
Instantaneous Surface Temperature and Heat Flux. The previous solutions give the following important results for the instantaneous surface temperature and surface heat flux. Dirichlet Condition.
The instantaneous surface heat flux is given by
aq(a, t) 1 k ( T o  Ti) = 1 + V~~~o
(3.96)
The instantaneous surface temperature is given by
Neumann Condition.
k(T(a, t )  Ti) = 1  e F° erfc (~/~o) aqo
(3.97)
Robin Condition. The Robin solution given above yields expressions for the instantaneous surface temperature and the instantaneous surface heat flux. They are as follows:
(T(a, t )  Ti)
Bi 11  e (B'÷1)2F° erfc [(Bi + 1)v/Foo]l Bi+l
(r~ r/) and
aq(a, t) _ Bi l1 + e (B~÷1)2F° erfc [(Bi + 1)X/~oll k ( T I  Ti) B i + l
(3.98)
(3.99)
Instantaneous Thermal Resistance Resistance Definition. The instantaneous thermal resistance for the three boundary conditions is defined as R = (T(a, t)  Ti)/Q where Q = q(a, t)4r~a2. The results given above yield the following expressions. Dirichlet Condition Resistance 1
4rckaRo = [1 + tvr~vro)l'l/"/'~'"
(3.100)
Neumann Condition Resistance
4~kaRN = 1  eF° erfc (X~o)
(3.101)
CONDUCTION AND THERMAL CONTACTRESISTANCES(CONDUCTANCES)
3.31
Robin Condition Resistance 1  e z2 erfc 4 r t k a R R = 1 + e z2 erfc
(z) (z)
(3.102)
where z = (Bi + 1)V~o. The three previous expressions approach the steadystate result 4 r t k a R = 1 for large dimensionless time. The previous expression for the Robin condition can be calculated by the following rational approximation with a maximum error of less than 1.2 percent: 1  a~s  a2 $2  a3 $3 4 r t k a R R = 1 + a l s + a2s 2 + a3 $3
(3.103)
where s = 1/(1 + p z ) and the coefficients are al = 0.3480242, a2 = 0.0958798, a3 = 0.7478556, p = 0.47047. The three solutions corresponding to the three boundary conditions can be used to obtain approximate solutions for other convex bodies, such as a cube, for which there are no analytical solutions available. The dimensionless parameters Bi and Fo are defined with respect to the equivalent sphere radius, which is obtained by setting the surface area of the sphere equal to the surface area of the given body, i.e., a = VA/(4r0. This will be considered in the following section, which covers transient external conduction from isothermal convex bodies.
Transient External Conduction From Isothermal Convex Bodies External transient conduction from an isothermal convex body into a surrounding space has been solved numerically (Yovanovich et al. [149]) for several axisymmetric bodies: circular disks, oblate and prolate spheroids, and cuboids such as square disks, cubes, and tall square cuboids (Fig. 3.10). The sphere has a complete analytical solution [11] that is applicable for all dimensionless times Fov~  m / A . T h e dimensionless instantaneous heat transfer rate is Q~/~a = Q V /  A / ( k A O o ) , where k is the thermal conductivity of the surrounding space, A is the total area of the convex body, and 00 To Ti is the temperature excess of the body relative to the initial temperature of the surrounding space. The analytical solution for the sphere is given by Q~/~A= 2X/~ +
1
(3.104)
which consists of the linear superposition of the steadystate solution (dimensionless shape factor) and the smalltime solution (halfspace solution). This observation was used to propose a simple approximate solution for all body shapes of the form 1
Q~AA= S~AA+
(3.105)
where S~A is the dimensionless shape factor for isothermal convex bodies in full space. This parameter is a relatively weak function of shape and aspect ratio; its values lie in the range 3.192 b
a_b
HyperEllipse a>b
(x}n+
0 4 × 102.
a3 =
Transient Spreading Resistance Within SemiInfinite Flux Tubes and Channels
Isoflux Circular Contact Area on Circular Flux Tube. Turyk and Yovanovich [118] reported the analytical solutions for transient spreading resistance within semiinfinite circular
CONDUCTION
AND THERMAL
CONTACT RESISTANCES (CONDUCTANCES)
3.51
flux tubes and twodimensional channels. The circular contact and the rectangular strip are subjected to uniform and constant heat fluxes. The dimensionless spreading resistance for the flux tube is given by the series solution 16 1 ~ J2(6.e) erf (~5.ex/Fdo) 4kaRs  ~ e .=~ ~53J~(8,)
(3.167)
where e = a/b < 1, Fo  o~t]a2, and 5n are the roots of Jl(X) = 0. The series solution approaches the steadystate solution presented in an earlier section when the dimensionless time satisfies the criterion Fo > 1/62 or when the real time satisfies the criterion t > a2/(o~2).
Isoflux Strip on TwoDimensional Channel
The dimensionless spreading resistance within a twodimensional channel of width 2b and thermal conductivity k was reported as
kRs = 1_~ ~ ~36 m = 1
sin 2 (mne) erf (mneV~o)
(3.168)
m3
where e = a/b < 1 is the relative size of the contact strip, and the dimensionless time is defined as Fo = t~t/a2. There is no halfspace solution for the twodimensional channel. The transient solution is within 1 percent of the steadystate solution when the dimensionless time satisfies the criterion Fo _>1.46/~2.
CONTACT, GAP, AND JOINT RESISTANCES AND CONTACT CONDUCTANCES Point and Line Contact Models The thermal resistance models for steadystate conduction through contact areas and gaps formed when nonconforming smooth surfaces are placed in contact are based on the Hertz elastic deformation model [37, 41,117, 121] and the thermal spreading (constriction) results presented previously. The general elastoconstriction models for point and line contacts are reviewed by Yovanovich [143]. In the general case the contact area is elliptical and its dimensions are much smaller than the dimensions of the contacting bodies. The gap that is formed is a function of the shape of the contacting bodies, and in general the local gap thickness is described by complex integrals and special functions called elliptical integrals [8, 10]. Two important special cases are I considered in the following sections: spherefiat and circular cylinderflat contacts. The review of Yovanovich [143] can Sphere be consulted for the general case.
el
Gap
£2
I
o I
Elastoconstriction Resistance of SphereFlat Contacts. The contact resistance of the spherefiat contact shown in Fig. 3.23 is discussed in this section. The thermal conductivities of the sphere and flux tube a r e k l and k2, respectively. The total contact resistance is the sum of the constriction resistance in the sphere and the spreading resistance within the flux tube. The contact radius a is much smaller than the sphere diameter D and the tube diameter. Assuming isothermal contact area, the general elastoconstriction resistance model [143] becomes:
I
FIGURE 3.23 Sphereflatcontact with gap.
Rc 
1
2ksa
(3.169)
3.52
CHAPTER THREE
where ks = 2klk2/(kl + k2) is the harmonic mean thermal conductivity of the contact, and the contact radius is obtained from the Hertz elastic model [117]: 2a [3FA] '/3 O[ Dz ]
(3.170)
where F is the mechanical load at the contact, and the physical parameter is defined as 1
E
(3.171)
(1  v 2) +
A = ~
E1
E2
where Vl and v2 are the Poisson's ratio and E1 and E2 are the elastic modulus of the sphere and flat contacts, respectively.
Elastoconstriction Resistance of CylinderFlat Contacts.
The thermal contact resistance model for the contact formed by a smooth circular cylinder of diameter D and thermal conductivity kl, and a smooth flat of thermal conductivity k2, was reported by McGee et al. [63] to be
ksRc 2n kl In
 2k~2n kTIn (4nF*)
(3.172)
where ks is the harmonic mean thermal conductivity and A is the Hertz physical parameter defined above. The dimensionless mechanical load is defined as F*  FA/(DL), where Fis the total load at the contact strip and L is the length of the cylinder and the flat.
Gap Resistance Model of SphereFlat Contacts. The general elastogap resistance model for point contacts [143] reduces for the sphereflat contact to
1(o)
R g  L kg.olg.p
(3.173)
where L = D/(2a) is the relative contact size defined previously. The gap integral for point contacts proposed by Kitscha and Yovanovich [46] is defined as
Ig,p =
I
L 2x tan 1 V'x 2  1 (28/D) + (2M/D) dx
(3.174)
The local gap thickness ~i is obtained from 28 D 1
((L)2) 1
1/2
+~
1 [
( 2  x 2) sin ~
(1)
+V'x 2  1
1
1  L2
(3.175)
where L  D/(2a), x  r/a, and 1 _<x ___L. The gap gas rarefaction parameter is defined as M = alSA
(3.176)
with ct = (2  Ctl)/txl + (2  Ctz)/Ct2, where O~l and ix2 are the accommodation coefficients at the gassolid interfaces, respectively. The gas parameter 13= (27)/[(7 + 1)/Pr], where 7= Cp/Cv and Pr is the Prandtl number. The mean free path A of the gas molecules is given in terms of Ag,0, the mean free path at some reference gas temperature To and reference gas pressure P0, as follows:
Tg Pg,o A=Ag.0 Tg,o Pg
C O N D U C T I O N A N D T H E R M A L CONTACT R E S I S T A N C E S ( C O N D U C T A N C E S )
3.53
Gap Resistance Model of CylinderFlat Contacts. The general elastogap resistance model for line contacts [143] reduces for the circular cylinderfiat contact to 1 _ 4aLcyl kgolgt Rg O ' '
(3.177)
where D and L are the diameter and length of the cylinder, and a is the halfwidth of the contact strip. The gap integral for line contacts [63] is
Igl = % i I" c°sh1 (;) d; ' rt (28/0) + (2M/D)
(3.178)
The local gap thickness for line contacts [121] is 28 (1___/1/2 ( ~2~1/2 2 ~ ~  l  L 2 ]  1~] + [~(~21)1/2cosh1(~)~2+1]
(3.179)
where L = D/(2a), ~ = xlL and 1 < x < L. The dimensionless contact strip width is obtained from the Hertz theory [117] 2a / FA D  4~/ rtDLcyl
(3.180)
where F is the total load at the contact and A is the physical parameter defined earlier.
Radiative Resistance Model of SphereFlat Contacts. The radiative resistance of a gap formed by two bodies in elastic contact, such as a sphereflat or cylinderflat contact, respectively, is complex because it depends on the geometry of the gapthe surface emissivities of the boundaries, which includes the side walls that form the enclosure. Kitscha and Yovanovich [46] and Kitscha [47] proposed the following radiative resistance model for a spheredisk contact with bounding side walls. All surfaces were assumed to be gray with constant emissivity values 61, 62, and 6 3 for the sphere, disk, and side walls, respectively. The sphere and disk were assumed to be isothermal at temperatures 7'1 and T2, respectively. The following expression was proposed: 1
Rr = A2ff ,24(yT3
(3.181)
where A2 is the surface area of the disk, ~ is the StefanBoltzmann constant,_and Tm = (T1 + T2)/2 is the mean absolute temperature of the contact. The radiative parameter F12 is defined as 1 M
F12
1 61
1 6 2
 +~ 261 62
+ 1.104
(3.182)
Joint Resistance Model of SphereFlat Contacts. The joint thermal resistance of a contact formed by elastic bodies, such as a sphereflat contact is obtained from the model proposed by Kitscha and Yovanovich [46] and Kitscha [47]" 1
1
1
1
Rj  R c ÷ l~g~ ÷ rR
(3.183)
which is a function of the constriction resistance Rc, the gap resistance R s, and the radiative resistance Rr, which are in parallel. The accuracy of the proposed model was verified by numerous experiments.
3.54
CHAPTER THREE
Experimental Verification of Elastoconstriction and Elastogap Models.
Experimental data have been obtained for the elastoconstriction resistance of point contacts [47] and line contacts [63] for a range of sphere and cylinder diameters, material properties, and mechanical loads. Data were obtained for the verification of the elastogap model for the point contact [46] and line contact [63]. The elastogap models have been tested with air, argon, helium, and nitrogen as the gap fluid at gas pressures from 106 torr to atmospheric pressure. Some representative test data for the elastoconstriction and elastogap resistances compared with the theoretical values are given in the following sections.
SphereFlat Test Results.
Kitscha [47] performed experiments on steady heat conduction through 25.4 and 50.8mm sphereflat contacts in an air and argon environment at pressures between 105 torr and atmospheric pressure. He obtained vacuum data for the 25.4mmdiameter smooth sphere in contact with a polished flat having a surface roughness of approximately 0.13 ktm RMS. The mechanical load ranged from 16 to 46 N. The mean contact temperature ranged between 321 and 316 K. The harmonic mean thermal conductivity of the sphereflat contact was found to be 51.5 W/mK. The emissivities of the sphere and flat were estimated to be ~ = 0.2 and ~2 = 0.8, respectively. The contact, gap, radiative, and joint resistances were nondimensionalized as R* = DksR. The dimensionless radiative resistance for the spherefiat contact given above becomes 3.82 x 101° Rr*=
T~m
The dimensionless constriction resistance is R* = L and the dimensionless joint resistance in a vacuum is 1

1
+~
1
(3.184)
The model predictions and the vacuum experimental results are compared in Table 3.16. TABLE 3.16
Dimensionless Load, Constriction, Radiative, and Joint Resistances [47] N, newtons
L, D/2a
Tin, K
R*r
R~
R j test
16.0 22.2 55.6 87.2 195.7 266.9 467.0
115.1 103.2 76.0 65.4 50.0 45.1 37.4
321 321 321 320 319 318 316
1155 1155 1155 1164 1177 1188 1211
104.7 94.7 71.3 61.9 48.0 43.4 36.4
107.0 99.4 70.9 61.9 48.8 42.6 35.4
The radiative resistance was approximately 10 times the constriction resistance at the lightest load and 30 times at the highest load. The largest difference between the theory and experiments is approximately 4.7 percent, within the probable experimental error. These and other vacuum tests [47] verified the accuracy of the elastoconstriction and the radiation model~ The elastogap model for a point contact was verified by Kitscha [47] and Ogniewicz [159]. For air, the gas parameter M depends on Tm, Pg as follows: M = 1.373 x 104 DPg Tm
(3.185t
where D is in cm, Tm in K and Pg in mmHg. The numerical value M is based on air properties at Tg,0= 288K and Pg,0= 760 mmHg.
CONDUCTION AND THERMAL CONTACTRESISTANCES (CONDUCTANCES)
3.55
TABLE 3.17 Elastogap Resistance Theory and Air Data [47]: D = 25.4 mm, D/(2a)  115.1
Tin, K
Pg, mm Hg
R •g
R*r
Rj*theory
Rj*test
309 310 311 316 318 321 322 325 321
400.0 100.0 40.0 4.4 1.8 0.6 0.5 0.2 Vacuum
76.9 87.4 97.1 137.2 167.2 227.9 246.6 345.4 oo
1293 1280 1268 1209 1186 1153 1143 1111 1155
44.5 47.8 50.6 59.5 64.5 71.7 73.4 80.1 104.7
46.8 49.6 52.3 59.0 65.7 73.1 74.3 80.3 107.0
The elastogap model and the experimental results are compared over a range of gas pressures in Table 3.17. Although tests were conducted at smaller values of the dimensionless parameter L over a range of gas pressures, sphere diameters, and gases, the results given in Table 3.17 are representative of the other data and they also correspond to the case that challenges the validity of the proposed elastoconstriction and elastogap models. First note that the radiative resistance is approximately 10 times the constriction resistance. Second, observe that the gap resistance is approximately % of the constriction resistance at the highest gas pressure, approximately equal to the constriction resistance at a gas pressure between 4 and 40 mmHg, and finally 3 times the constriction resistance at Pg = 0.2 mmHg. The agreement between the theory and the tests is very good to excellent. The largest difference occurs at the highest gas pressures, where the theory predicts lower joint resistances by approximately 5 percent. The agreement between theory and experiment improves with decreasing gas pressure. It can also be seen in Table 3.17 that the air within the spherefiat gap significantly decreases the joint resistance when compared with the vacuum result.
Thermal Contact, Gap, and Joint Conductance Models Thermal contact, gap, and joint conductance models developed by many researchers over the past five decades are reviewed and summarized in several articles [20, 23, 50, 58,143,147,148] and in the recent text of Madhusudana [59]. The models are, in general, based on the assumption that the contacting surfaces are conforming (or fiat) and that the surface asperities have particular height and asperity slope distributions [26, 116, 125]. The models assume either plastic or elastic deformation of the contacting asperities, and require the thermal spreading (constriction) resistance results presented above.
Plastic Contact Conductance Model of Cooper, Mikic, and Yovanovich.
The thermal contact conductance models are based on three fundamental models: (1) the metrology model (surface roughness and asperity slope), (2) the contact mechanics model (deformation of the softer contacting surface asperities), and (3) the thermal constriction (spreading) resistance model for the microcontact areas. Cooper et al. [14] presented the contact conductance model, which is based on the Gaussian distribution of the asperity heights and slopes, the plastic deformation of the contacting asperities, and the constriction resistance, which is based on the isothermal circular contact area on a circular flux tube result. The development of the dimensionless contact conductance model for conforming rough surfaces has been presented in several publications [14, 65, 139, 143, 147, 148]. The theoretical dimensionless contact conductance has the form hc 1 exp(x 2) Cc (3.186) m ks 2V~n (1e)15
3.56
CHAPTER THREE
with x = erfc 1 (2P/Hc) and e = ~ c , where P is the contact pressure and H~ is the flow pressure at the plastically deformed surface asperities. The surface parameters are 6 = V'621 + 6 2 and m = kTm 2 + m 2, where 61 and 62 represent the RMS surface roughness of the two contacting surfaces and rnl and m2 represent the absolute mean slopes of the surface asperities of the contacting surfaces (Fig. 3.24). The interface effective thermal conductivity is defined as k s  2klk2/ (kl + k2). The metrology model also gives the following geometric relationships [139141].
Y
I
....
,
!
:iii!!~i~
Relative Real Contact Area.
......:~,~i#: :i .f#i~ .'.........Ni: .J~... N .....
i;iiN:;:~i~i~i!!!~ F I G U R E 3.24 parameters.
~.2= Ar _ 1 erfc (x) Aa 2
~.,~!iii:.i:
m =,' m ] + m 2
where At is the total real contact area and Aa is the corresponding total apparent area.
Contact Spot Density.
C o n f o r m i n g r o u g h surface g e o m e t r i c
1 ( ~ ) 2 exp(2x2) n=~ erfc (x)
Mean Contact Spot Radius.
a: g ( )exp x' er,c x, where x = (1/V2)(Y/o) and Y is the mean plane separation (Fig. 3.24). The relative mean plane separation is obtained from
13
which is approximated by [139141]
[ ( :/1
Y  1.184 ln 3.132
o
He]_!
which is valid in the ranges 4.75 > Y/o > 2.0 and 10 6 < P/Hc < 2 × 102, and it has a maximum error of approximately 1 percent. The relative mean plane separation appears in the gap conductance model. Yovanovich [139] proposed the correlation equation for the contact conductance model: [ p '~0.95 Co= 1 . 2 5 ( ~ [ )
(3.187)
which is valid for the wide range 106 _> 1, the buoyancy force that drives the motion (term IV in Eq. 4.6) can be seen to be balanced almost exclusively by the viscous force (term III). With this approximation, and because, for Pr >> 1, Pr disappears from Eq. 4.7, Ra is the only remaining dimensionless group in the simplified equations. For Pr > 1
(4.9a)
Nu = f(Ra Pr)
for Pr 106 are believed to be due to turbulence, which is not included in their steadystate laminar solution. This comparison lends support to the approximations inherent in the simplified equations of motion.
ThinLayer Approximation. Laminar analyses often make the further approximation that the boundary layer is so thin that when the simplified equations of motion are rewritten in terms of local surface coordinates, i.e., in terms of the x and y of Fig. 4.3a, several terms normally associated with curvature effects can be dropped. The Nusselt number equation, based on solutions to such laminar "thinlayer" equation sets, always takes the form Nu = c Ra TM
(4.10)
where c is independent of Ra but does depend on geometry and Prandtl number. For example, c  0.393 for the horizontal circular cylinder in air (Pr = 0.71) [240]. Plotted in Fig. 4.2, Eq. 4.10 with this value of c falls consistently below the experimental data; also (since no single power law can fit the data) its functional form is incorrect. At high Ra it approaches Kuehn and Goldstein's laminar solution to the complete simplified equations of motion, but, as already discussed, the presence of turbulence has rendered laminar solution invalid in this range. The thinlayer approximation fails because natural convective boundary layers are not thin. From the interferometric fringes in Fig. 4.2b (which are essentially isotherms), the thermal boundary layer around a circular cylinder is seen to be nearly 30 percent of the cylinder diameter. For such thick boundary layers, curvature effects are important. Despite this failure, thinlayer solutions provide an important foundation for the development of correlation equations, as explained in the section on heat transfer correlation method.
Problem Classification
Classification of Flow. The flow near bodies like that in Fig. 4.1a can be classified into one of three types: twodimensional (2D), axisymmetric, and threedimensional (3D). The flow is 2D if the body is invariant in crosssectional shape along a long horizontal axis (e.g., a long horizontal circular cylinder). An axisymmetric flow takes place near a body (Fig. 4.3b) whose shape can be generated by revolving a body contour about a vertical line; for example, a sphere is generated by rotating a semicircle. If the body meets neither the 2D or axisymmetric requirements, its flow is classified as 3D; this class includes the flow around 2D and axisymmetric bodies whose axes have been tilted. In addition to this geometric classification, the flow can be classified as fully laminar (i.e., laminar over the entire body), fully turbulent (turbulent over the entire body), or laminar and
4.6
CHAPTERFOUR turbulent (i.e., laminar over one portion and turbulent over the other). Laminar flow is confined to a boundary layer, but turbulent flow may be either attached or detached, as discussed in the section on turbulent Nusselt number.
Thermal Boundary Conditions. The simplest thermal boundary conditions are Tw and Too, both specified constants. If Too varies, it is assumed to be a function only of the elevation z; also, dTJdz is always positive, since, with 13positive, a situation where dTJdz is negative is always unstable and cannot be maintained in an extensive fluid. Either T~ or q" may be specified on the body surface. The difference between the temperature at a given point on the body and Tooat the same elevation is denoted AT; AT is the areaweighted average value of AT over the surface, and AT0 is an arbitrary reference temperature difference, usually set equal to AT. If AT is positive over the lower part of the body and negative over the upper part, the surface flows over each part are in opposite directions, and these two flows meet and detach from the body near the line on the surface where AT = 0. Direction of Heat Transfer (Cooling versus Heating).
The relationship of Nu to Ra and Pr for a heated body is precisely the same as when the body is inverted and cooled, except that Tw  Toois replaced by Too Tw. This principle implies, for example, that the relationship for the heated triangular cylinder in Fig. 4.4a is identical to that for the inverted cooled triangular cylinder in Fig. 4.4b. In this chapter it is normally assumed that the surface is heated, and the results for a cooled surface can be inferred from results for a corresponding heated surface by using the above principle. Xw
0.1. Nuxr = He(Ra*) 1/5
(
He =
Pr
)1/5
(4.34a)
4 + 9~/~r + 10Pr
Nue, = 0.4/ln (1 + 0.4/Nu~r)
(4.34b)
(CV)3/4(Ra*) TM Nu,,x = 1 + (C2 Pr/Ra*) 3
10 ~2< C2 < 2 × 10 ~3 C2 = 7 × 10 ~2 (nominal)
Nux = ( ( N u e j m + (Nu,,x)m)'/m
(4.34c)
m = 3.0
(4.34d)
The value of C2 determines the transition between laminar and turbulent heat transfer. A nominal value is C2 = 7 × 10 ~2. Comparison With Data. Equation 4.34 is in excellent agreement with the data of Humphreys and Welty [147] and Chang and Akins [40] for Pr = 0.023 (mercury), but the data lie only in the laminar regime. There is also good agreement with measurements of Goldstein and Eckert [115], Vliet and Liu [275], and Qureshi and Gebhart [222] for water, although the observed transition depends on the level of heat flux. By choosing an appropriate value for C2 for each q", Fig. 4.8 shows that Eq. 4.34 can be made to fit each data set. For a nominal value
103

"
'
.
.
.
.
.
.
i
.
s.s3E2
Cz 1.06r13
1.32 E3
1.04 E13
F1
2.33E3
8.01 E l 2
0
371E3
5 97 E12
•
449E3
33s 12
q 'tW/m 2]
©
. /X
,
"
,
,
"'"
',
" '
•
I
•
 " '
'
;
"=
e.Z..~, 
o/~f..:~, ~ , ~ , ~ : : ~ /
/
/

v
,,6:a;~'
/

.
./ .."" /
x
z
f
2
10012
.
'
,
.
.
.
.
.
!
!
1014
1013
. . . . .
1015
Ra* x
FIGURE 4.8 Comparison of local heat transfer measurements of Qureshi and Gebhart [222] with Eq. 4.34 for a uniform heat flux, vertical flat plate in water.
NATURAL CONVECTION
4.15
by C2 = 7 x 1012, Eq. 4.34 gives good agreement with measurements for oils, Pr = 60 and Pr = 140, except in the transition region. Fujii and Fujii [99] have shown that strong vertical temperature gradient in_the ambient fluid also affects the value of Nux in the transition regime. Calculation of AT. In some cases involving uniform q", it is sufficient to know the average temperature difference, defined as
AT= A
(Tw T=) dA
(4.35)
This is obtained from the average Nusselt number Nu defined in Fig. 4.6. A rough estimate of Nu can be obtained for Ra > 105 by using the equations in the section on vertical flat plates with uniform Tw and T=, ~ = 90 °, with AT replaced by AT. For higher accuracy, and for convenience since Ra* is used in place of Ra, use the following equation set:

6(
Nur = He Ra .1/5
Pr
)1/5
He = ~ 4 + 9~/Prr+ 10 Pr
(4.36a)
1.0 Nue= In (1 + 1.0/Nu T)
(4.36b)
/( (c2pr)04)
Nu,= (CV) 3/4 Ra *1/4 1 + Nu = (Nu~' + NU~n)1/m
Ra*
C2 = 7 x 1012
(4.36c)
m = 6.0
(4.36d)
m
He is tabulated in Table 4.1.
Vertical Plates of Various Planforms. There is a class of vertical plates, such as shown in Fig. 4.9, for which any vertical line on the plate intersects the edge only twice. When such plates are isothermal, the thinlayer laminar Nusselt number Nu r can be calculated [225] by dividing the surface into strips of width Ax and length S(x), applying Eq. 4.33a to compute the heat transfer from each strip, adding these to obtain the total (thinlayer) heat transfer, and using this heat transfer to calculate Nur. The result can be expressed as ~1/4 f0W S 3/4 d x N u T=
GCt R a TM
G=
A
(4.37a)
where/? is any characteristic dimension of the surface. This also results from the application of Eq. 4.14. Figure 4.9 shows examples of the constants G for three body shapes.
B C
FIGURE 4.9
e=D
e=H
e=Btan0
G=1.05
G=8/7
G=
1+7cos0
I
_ ~
B < C cos 0
I w
qe Nu  AA~k
gl3,~Te3 Ra 
w
gl3q"e 4 Ra* =
v~k
Definition sketch for natural convection from plates of various planform.
4.16
CHAPTER FOUR
For laminar heat transfer, the correlations for such disks have been given by Eq. 4.18 [227]: Nut = NUcoND + Nur
(4.37b)
Yovanovich and Jafarpur [294] have shown that, for the length scales in Fig. 4.9, e NUcoNo = X/A NUVXcOND
(4.37C)
If the plate is heated on one side only, but is immersed in a full space NUVXcOND= 3.55, if the plate is heated on one side and immersed in a halfspace (i.e., the heated plate is embedded in an infinite adiabatic surface) NUvXcOND = 2.26, and if the plate is heated on both sides and immersed in a full space NUVX.COND= 3.19. In all cases A is the surface area that is active. For a general equation for Nu: substitute Eqs. 4.37a and c into Eq. 4.37b to obtain the equation for Nut, use Eq. 4.33 for Nu,, and substitute these equations for Nut and Nut into Eq. 4.33 to obtain the equation for Nu. The isothermal surface correlation jus___ttprovided can also be used to estimate Nu for a uniform flu___xboundary by replaci_~ ATby AT. For higher accuracy, use Eq. 4.36a~to find the thinlayer ATfor each strip, find ATfor the entire surface by area weighting the ATfor each strip (Eq. 4.35), and calculate Nur from the area weighted AT. This results in A e 1/5 m
Nur = C2Ht Ra *v5
C2 ~w
(4.38)
$6,5 dx
%
where e is again any characteristic dimension of the surface, and terms are defined in Fig. 4.9. To find Nu, substitute Eq. 4.38 for Nur and Eq. 4.37c for NUcoND into Eq. 4.37b to obtain an expression for Nut. Use this Nut and Eq. 4.36c for Nut in Eq. 4.36d to obtain Nu.
Horizontal Heated UpwardFacing Plates (~ = 0 °) With Uniform T. and Too Correlation. For horizontal isothermal plates of various planforms with unrestricted inflow at the edges as shown in Fig. 4.10, the heat transfer is correlated for 1 < Ra < 10~°by the equation: Nur = 0.835Ce R a TM
(4.39a)
1.4 Nut = In (1 + 1.4/Nu r)
(4.39b)
Nu, = C,v R a 1/3
(4.39c)
Nu = ((Nut) m + (Nut)m) l/m
(4.39d)
m=10
Use of the length scale L* [116], defined in Fig. 4.10c, is intended to remove explicit dependence on the planform from the correlation. ~ Perimeter p A L*= Alp qL* Nu= A A T k Re = cj/3 ~T (L*) 3 va
Edge view (a)
Plon view (b)
(c)
FIGURE 4.10 Definition sketch for natural convection on a horizontal upwardfacing plate of arbitrary planform. Only the top heated surface of area A is heated.
NATURAL CONVECTION i
•
10 2
,
i
'
o
AIArabi & EIRiedy, circular
n
AIArabi & EIRiedy, square
I
'
i
•
./ .,.~"
Clausing&Berton,square (Nr)
o
4.17
J
Clausing & Berton, square (nitrogen) •
Yousef,Tarasuk,&McKeen,square
+
Bovy&Woelkrectangul , ar
o
o
i
10 0

~
J m
101
f
1lO*/,
.
I
10 2
,
I
,
,
10 4
I
10 6
i
I
10 0
,
,
10 ~°
Ra FIGURE 4.11 Comparisonof Eq. 4.39 to data for upwardfacing heated plates of various planform in air.
Comparison With Data. Figure 4.11 shows that Eq. (39) is in good agreement with measurements for gases (Pr  0.7). The Clausing and Berton [62] data shown in the figure have been extrapolated to Tw/To. = 1 using their correlation. The scatter in the data of Yousef et al. [290] is due to temporal changes in the heat transfer. Excellent agreement was also found with the data of Goldstein et al. [116] for 1.9 < Pr < 2.5 for a variety of shapes, but the data of Sahraoui et al. [237] for a disk and flat annular ring, 1 < Ra < 103, fall below this equation. For Pr > 100, Eq. 4.39 is in excellent agreement with the measurements of Lloyd and Moran [297] for 107 < Ra < 109, and agrees, within the experimental scatter, with the measurements of Lewandowski et al. [179] for 102 < Ra < 104. Horizontal Heated UpwardFacing Plates (~ = 0 °) With Uniform Heat Flux Correlation. The nomenclature for this problem is also given in Fig. 4.10, where AT is the surface average temperature difference. Equation 4.39 should also be used for this case, where the calculated Nu value provides the average temperature difference AT. Comparison With Data. Equation 4.39 agrees to within about 10 percent with the data of Fujii and Imura [103] and Kitamura and Kimura [161], for heat transfer in water (Pr = 6), for 104 < Ra < 1011. Both these experiments were performed using effectively infinite strips of finite width.
Horizontal Isothermal Heated DownwardFacing Plates (~ = 180 °) Correlation. Definitions and a typical flow pattern for this problem are shown in Fig. 4.12. The heat transfer relations given here assume that the downwardfacing surface is substantially all heated; if the heated surface is set into a larger surface, the heat transfer will be reduced. Since the buoyancy force is mainly into the surface, laminar flow prevails up to very high Rayleigh numbers. The following equation can be used for 103 < Ra < 101°:
4.18
CHAPTER FOUR
qL* Nu Heated plate
AATk
Ra : g~ AT (L*)3 VQ
~=
A p
_ _
=
heater area heater perimeter
FIGURE 4.12 Definition sketch for natural convection on a downwardfacing plate. Only the bottom surface of area A is heated.
Nu r =
0.527 Ra 1/5 (1 + (1.9/Pr)9/~°) ~9
(4.40a)
2.5 Nue = In (l + 2.5/Nu r)
(4.40b)
The coefficient in Eq. 4.40a was obtained by fitting the results of the integral analysis of Fujii et al. [102]. Comparison With Data. Measurements for isothermal plates in air are compared to Eq. 4.40 in Fig. 4.13 for rectangular plates. Data lie within about +__20percent of the correlation. Measurements have also been done using water, but only with a uniform heat flux boundary condition. For water, the data of Fujii and Imura [103] for a simulated 2D strip lie about 30 percent below Eq. 4.40, but the data of Birkebak and Abdulkadir [20] lie about 3 percent
10
2
........
,
........
,
........
,
........
o
Faw & Dullforce, square
D
Hatfleld& Edwards, square
,
A
Halfleld& Edwards, rectangular 3:1
•
Aihara,Yamada, & Endo, 2D strip
........
,
........
,
".
J
•
....
i f
+
Bevy& Woelk, rectangular 1:1,3:2, 3:1
x
Restrepo& Glicksman, s q u a r e
J
~.~.i,I'~++.~+
..
I
+
Z 101
11~
o ~ 0 o
. . . . . . . .
10 s
i
10 +

i
i
....
11
10 s
. . . . . . . .
!
i
. . . . . . .
10 s
!
10 7
. . . . . . . .
i
10 e
. . . . . . . .
!
10 9
. . . . . . .
101°
Ra FIGURE 4.13 Comparison of Eq. 4.40 to data for downwardfacing heated plates of various planform in air.
NATURAL CONVECTION
4.19
above the correlation. The importance of the heat transfer from the outer edge of the plate is believed to be the cause of such large discrepancies.
Plates at Arbitrary Angle of Tilt. The previous sections provide equations from which to compute the total heat transfer from vertical plates (~  90°), horizontal upwardfacing plates (~ = 0°), and horizontal downwardfacing surfaces (~ = 180°). These equations are the basis for obtaining the heat transfer from tilted plates. For a wide isothermal plate at any angle of tilt, first compute the heat transfer from Eq. 4.33 (vertical plate) but with g replaced by g sin ~, then compute the heat transfer from Eq. 4.40 (downwardfacing plate) with g replaced by g(0,cos ~)max,then compute the heat transfer from Eq. 4.39 (upwardfacing plate) with g replaced by g(0, cos ~)max,and take the maximum of the three heat transfer rates. It is important to take maximum heat transfer rather than the maximum Nusselt number, because the Nusselt numbers are based on different length scales. For plates with small aspect ratio, such as shown in Fig. 4.9, follow the same procedure except use Eq. 4.37b in place of Eq. 4.33b. Vertical Isothermal Plate in Stably Stratified Ambient Correlation. For an isothermal vertical plate (see Fig. 4.6) in an ambient fluid whose temperature Too increases linearly with height x, the heat transfer depends on the stratification parameter S, defined by L dToo S  AT dx
(4.41)
where the mean temperature difference AT is also the value of Tw Too at the midheight of the surface. For 0 < S < 2, Tw  Too is positive over the entire plate; for S = 2, Tw  Too at x  L; and for S ~ oo the plate temperature is lower than Tooover the top half of the plate and greater than Tooover the bottom half. From laminar thinlayer analysis [41,225,226] the value of Nu T, corrected for stratification effects, is NuT= (1 + S/a)bCe Ra TM : cSl/4Cf
Ra TM
S2
(4.42b)
For gases: a = 1, b = 0.38, and c = 1.28; for water (Pr  6): a = 2, b = 0.5, and c = 1.19. For turbulent flow everywhere on the plate and for S _5, Aihara [1] has shown that the heat transfer coefficient is essentially the same as for the parallelplate channel (see the section on parallel isothermal plates). Also, as W/S ~ O, the heat transfer should approach that for a vertical fiat plate. Van De Pol and Tierney [270] proposed the following modification to the Elenbaas equation [88, 89] to fit the data of Welling and Wooldridge [283] in the range 0.6 < Ra < 100, Pr = 0.71, 0.33 < W/S < 4.0, and 42 < H/S < 10.6:
Ra{ F / 0 5 ~3/4]} Nu =   ~   1  e x p [  ~ ~ a a ) ]
(4.61a)
24(1 0.483e °'17/~*) W = {(1 + ix*/2)[1 + (1  e°83~*)(9.14~e 3  0.61)]} 3 
where
IV
t~p
IV iV
IV IV
/
'~T.
4sl* \,
,4sl,
,I ~ s j w
2WS r = 2W + S " a * = S/W
q"S Nu = ~( T .  T ~ ) k
q"r Nu : (TwToo) k
Ro 
IV
q"L Nu = ( T .  T = )
g.B(TwToo) S 3
r
va
L
(a)
k
g/3(T.Too) L 3
va
Re :
g/3 (TwToo) r 3
(4.61b)
va
Re :
(c)
(b)
/
,Is ~ s
.I,' '4
s F
(T.Too) k Ro = g,B (T.T~)S 3 va
(d)
H
Nu 
q"S (T,Too) k
Re =
(e)
gB (T.Too)S3 va
S 
H
Nu =
~S
(TwTm) k'~
FIGURE 4.23 Flow configurations and nomenclature for various open cavity problems.
d
= ~ . Re =
g# (T.T=) S3 S va
(f)
D
N A T U R A L CONVECTION
,0,..37
m
and where S is dimensionless and equal to 4.65S (for S in cm) or11.8S (for S in inches). Other nomenclature is defined in Fig. 4.23a. For a given base plate area there are two fin spacings, $1 and $2, of particular interest. If the fin spacing is decreased, starting from a large value, the heat transfer coefficient remains relatively constant until the spacing $1 (which corresponds to Ra = R a l ) is reached, at which it begins to fall rapidly because of fin interference. As the spacing is decreased within a specified volume, more fins are added to the base plate, thereby increasing the total surface area for heat transfer. Since total heat flow is proportional to the product of heat transfer coefficient and surface area, decreasing the spacing below S1 still improves the total heat transfer until the spacing $2 (which corresponds to Ra = Ramax) is reached; below $2, the total heat transfer falls. For long fins (0~* T= or downwardfacing for Tw < T=) has been measured by Jones and Smith [150], Starner and McManus [263], and Harahap and McManus [120]. For a given fin width, W = 0.254 m (0.833 ft), Jones and Smith were able to correlate their measured heat transfer to within about +_25 percent on an NuRa plot. The following equation closely represents this correlation over the data range 2 x 102 < Ra < 6 x 105, Pr = 0.71, 0.026 < H/W < 0.19, and 0.016 < S/W < 0.20: Nu =
[(Ra)m ]500
+
(0.081 Ra°39)mql/m
m = 2
(4.62)
This simple equation ignores the effect of the geometric parameters H/S and H/W. While H/S does not appear to play a strong role, H/W is known to have significant effect. A parametric study using Eq. 4.62 shows that, for a given base area and temperature difference, the curve of total heat transfer versus fin spacing displays a sharp maximum for high fins (large H) and a less welldefined peak for short fins. This results from the rapid increase in total surface area with decreasing fin spacing (i.e., as fins are added to the base plate) for high fins. Because the total heat transfer falls off very sharply for spacings below the optimum, a conservative design would use spacings larger than the optimum $2 (defined in the section on rectangular isothermal fins on vertical surfaces) calculated from Eq. 4.62.
Horizontal Corrugated Surfaces. A heated horizontal isothermal corrugated surface, such as that shown in Fig. 4.23c, can also be considered a finned surface. The heat transfer is from the top surface only. AIArabi and EIRefaee [3] have measured heat transfer rates to air (Pr =0.71) over the range 1.8 x 104 < Ra < 1.4 x 10 7 (the nomenclature is defined in Fig. 4.23c), and have provided the following correlations: (0.46 ) Nu = sin (W/2) 0.32 Ra m where
for 1.8
104 < Ra < Rac
(4.63)
for Ra~ < Ra < 1.4 x 107
(4.64)
×
Rac = (15.8 14.0 sin (W/2)) x 105 m = 0.148 sin (re/2) + 0.187
and
054 ) Ral/3 Nu = ( 0.090 + sin0 . (~/2~
The measurements and the above correlation were intended to represent the case where W is asymptotically large. The dependence of the heat transfer on W remains to be determined.
Vertical Triangular Fins. For the vertical fin array in Fig. 4.23d, S is the fin spacing measured at the midheight of the fin, so that SW is the crosssectional area of the flow channel formed by the sides of adjacent fins, the base of width S, and the vertical plane passing
4.38
CHAPTER FOUR
through the fin tips. Nu and Ra are defined in the figure. The correlation of Karagiozis et al. [152] is Nu = NUcoND + C~ Ra 1/4 1 +
RaO2~
+ 8 Nu
8 Nu = [(0.147 Ra °39 0.158 Ra°46), 0]max
(4.65a) (4.65b)
where NUCOND is the (conduction) Nusselt number for the entire fin array in the limit as Ra 4 0. A conservative design (i.e., the heat transfer is underestimated) results if NUcoND = 0 is used. Otherwise, to estimate NUCOND,suppose that triangular fins are mounted on a rectangular base plate of dimension L1 and H, where the base plate entirely covers the vertical surface on which it is mounted. The surface and base plate therefore have a r e a A b = L1 × H. Suppose further that the fin height is small compared to the base plate dimensions (i.e., W > L a n d W >> L, or D >> L). This section deals with situations covered by entry 10 in Table 4.6, with the additional proviso that either 0 = 0 or 0 = 180 °. When 0 = 180 °, the hot, light fluid lies above the cold, heavy fluid, so the stationary fluid layer (in which there is no fluid motion) is inherently stable, and Nu = 1 for all Ra. (In terms of the conduction layer model, for 0 = 180 ° both the conduction layers are infinite, so the conduction layers always overlap, and Nu = 1.) In the 0 = 0 ° orientation, hot, light fluid lies below the cold, heavy fluid, so the stationary fluid layer is inherently unstable. Despite this inherent instability, the fluid remains stationary provided Ra is less than a "critical Rayleigh number" denoted by Rat. The value of Rat for this particular geometry is 1708. For Ra > Rac, the instability leads to a steadystate convective motion, the form and strength of which depends on both Ra and Pr. For Ra only slightly greater than Rao it consists of steady rolls of order L in size, but as Ra is further increased, more complex flow patterns are observed, and eventually the flow becomes unsteady. At very high Ra it becomes fully turbulent. The heat transfer characteristics reflect the existence of these various flow regimes: for Ra < Rac the fluid is stationary, so Nu is unity; the cellular motion initiated at Rac produces a sharp rise in Nu with Ra, which ultimately becomes asymptotic to the relation Nu ~= Ra 1/3 at very large Ra. For 0 = 0 °, the recommended equation [140] for Nu is: N u = 1 + 1  1708 " Ra 1/3 11n(Ral/3/k2) Ra k~+2 k2 +
E J[ ( )
Ra 1/3_ 1 5803
1[( ) ]
(4.78)
where square brackets with dots indicate that only positive values of the argument are to be taken, i.e.,
NATURAL CONVECTION
4.45
[X]" = (IX1 +X)2
(4.79)
X being any quantity. Values of the parameters kl and k2, both functions of Pr, are tabulated in Table 4.7 for several values of Pr. This table also cites the experiments from which the values were inferred, and gives the range in Ra over which Eq. 4.78 has been tested for each Pr. The following equations fit the dependence of kl and k2 on Pr exhibited in Table 4.7. 1.44
kl = 1 + 0.018/Pr + 0.00136/Pr 2
(4.80)
k2 = 75 exp(1.5 Pr '/2)
(4.81)
The form of Eq. 4.80 resulted from an approximate analysis [118], but the constants in the denominator have been chosen to fit the values given in Table 4.7 for Pr = 0.7 and Pr = 0.024, and they are therefore based on limited data. Caution is advised in using this equation when Pr < 0.7, except when Pr = 0.024. Also, for some values of Pr, the narrow range of Ra over which Eq. 4.78 has been tested should be noted. The data of Kek and Mtiller's [158] recent experiments using liquid sodium with Pr = 0.0058 are fit reasonably well by Eqs. 4.784.81, but the fit is improved considerably if kl is set equal to 0.087 and the power on (Ra/5830) is changed from 1/5 to 1A. TABLE 4.7
Values of kl and k2 to Be Used in Eq. 4.78
Pr (approximate)
kl
k2
Range of Ra tested
Reference
0.02 0.7 6 34 100 200 3000
0.35 1.40 1.44 1.44 1.44 1.44 1.44
>200 >400 140 100 85 85 75
Ra < 108 Ra < 1011 1 0 3 < Ra < 2 x 105 103 < Ra < 105 1 0 3 ___Ra < 3 × 1 0 6 10 3 < Ra < 5 × 105 1 0 3 < Ra < 3 x 1 0 4
233 See 142 for list. See 142 for list; also 117 243 243 233 243
10 3 < 10 3
> H, differences of up to 25 percent occur. The equation from Smart et al. [253]
11"})
(4.90)
fits the data for W >> H better than Eq. 4.89, but in contrast to Eq. 4.89, it does not have the proper asymptote, as Ra > ~. Equation 4.90 agrees well with data for Ra < 100 Rac and L / H = 3, 5, and 10. It is not recommended for Ra > 100 Rac. Figure 4.31 shows a plot of Eq. 4.89 for a circular cylinder cavity with perfectly conducting walls and various values of D/L. As is clear from the graph, the Nusselt number rises very steeply with Ra after initiation of convection, and very rapidly approaches the value of Nu for the horizontally extensive cavity. This behavior is consistent with the conduction layer model: at high Ra, the conduction layers on the walls at the sides are so thin that they have no effect on the heat transfer; at sufficiently low Ra, they are so thick that they overlap (even though those on the horizontal plates do not), so that their presence governs the condition for a stationary fluid.
Heat
Transfer
in V e r t i c a l
Rectangular
Parallelepiped
Cavities:
9 = 90 °
Cavities with HIL > 5 and WIL >~5. In contrast to the horizontal cavity, for which there is flow only when Ra > Rac, the vertical cavity experiences flow for any finite Ra. At small Ra,
70 ,50 3020
z
7 ,53 2 !
1103
104
105
I 1 !11111
106
1
I I I lllll
I0 ?
L I I tltttl I0 e
1 I Atiitl I0 9
Ro
FIGURE 4.31 Relation between Nu and Ra for horizontal cylindrical cavities with perfectly conducting walls, for various values of D/L, as given by Eq. 4.89 with Pr = 6.0.
NATURAL CONVECTION
4.51
however, the velocities are small and essentially parallel to the plates, so that they contribute little to the heat transfer, and for all practical purposes, Nu = 1. These conditions constitute the conduction regime. The development of the flow as Ra increases beyond this regime depends on H/L. If H/L >~ 40, the conduction regime becomes unstable at a critical Rayleigh number Ra~, which is plotted as a function of Pr in Fig. 4.32 (from Ref. 162). Increases in Ra past Rac lead through a turbulent transition regime and finally into a fully developed turbulent boundary layer regime characterized by turbulent boundary layers on each plate and a wellmixed core between them in which there is a vertical temperature gradient of about 0.36(Th  Tc)/H [170]. If H / L ~40, the regimes encountered as Ra increases are first conduction, then turbulent transition, and then turbulent boundary layer; for H / L 10 the effect of wall properties is not expected to be important (see Table 4.6, entry 9). For fluids with Pr > 4, the recommended equations are based on the proposals of Seki et al. [248]. For Ra (H/L) 3 < 4 × 1012,
[
Nu = 1, 0.36 Pr °°51
()036
Ra °25, 0.084 Pr °°51
1
Ra °3
(4.96)
dmax
and for Ra (H/L) 3 > 4 x 1012, Nu = 0.039 Ra 1/3
(4.97)
These equations have been tested for values of H/L ranging from 5 to 47.5. The middle term in Eq. 4.96 has been tested for 3 < Pr < 40,000, and the last term for 3 ~ Pr ~< 200. Equation 4.97 has been tested only for Pr = 5, and may underpredict measurements by as much as 20 percent. For 5 _~ 10, the plates are extensive (Table 4.6, entry 9), and the wall thermal properties are not important. Vertical Cavities (0 = 90 °) with U H > 2 and WIL >~5. Except in an end region immediately adjacent to the two vertical plates, the flow in a cavity with L >> H is everywhere parallel to the horizontal walls, with hot fluid in the upper half of the cavity streaming toward the cold plate and cold fluid in the lower half streaming toward the hot plate (only at very high Rayleigh numbers, where turbulent eddies of a scale smaller than H are possible, will this simple flow pattern break down). The plates at temperatures Th and Tc deflect the streams into boundary layers on each vertical surface. The predictions of Bejan and Tien [16] for adiabatic walls are correlated to within 8 percent by their equation Nu=l+(['y1Ra2(~)8]m+[72Ral/5(~)E/5]m}l/m
(4.98)
in which m = 0.386, 71 = 2.756 × 106, and 3'2= 0.623. The analysis is for laminar flow and hence this equation is not recommended for large Ra (H/L) 3. Because of the dominance of the walls in this problem, departures from the adiabatic wall conditions can be expected to have a marked effect on Nu. Rectangular cavities of practical interest are very often not isolated cells but rather members of a multicellular array, such as that sketched in Fig. 4.33. When 0  0, the central plane of each partition forms an adiabatic plane of symmetry, so that each cell behaves like an isolated cell (of the type defined in the section on geometry and parameters for cavities without interior solids) having wall thickness b equal to onehalf the partition thickness. When 0 ~ 0, there is usually heat transfer between cells, the magnitude of which is established by the coupling parameter kwL/kb. Smart et al. [253] carried out an experimental study on multicellular arrays with air as the fluid and 0 = 90 ° and found that for Ra at least as high as 107, Eq. 4.98 fit their data provided
NATURAL CONVECTION
4.53
Plate at Tc
Detail A ,
2b
/Plate at T h
FIGURE 4.33 Sketch of a multicellular array in which many rectangular parallelepiped cavities such as sketched in Fig. 4.25 may be contained. the values of 71 and 72 were slightly altered. The altered values of 7~ and Y2, tabulated in Table 4.10, depended on the conductive and radiative wall properties, as noted in the table. W h e t h e r the changes in 71 and 72 were attributable to the multicellular array effect, the radiative effect, or both, cannot be resolved from the data. Vertical Cavities with 0.5 < I t / L < 5 a n d W / L > 5. In this intermediate range of H/L, the lowtomoderate Rayleigh number flow consists of a twodimensional roll. The problem, particularly with L / H = 1, has been the subject of many numerical studies, and indeed for the adiabatic wall case with Pr = 0.7, it has formed the basis of a "benchmark problem" [72] for computational fluid dynamic (CFD) codes (even though it is virtually impossible to duplicate this situation in the real world because real fluids with Pr = 0.7 can never be properly insulated). For both the perfectly conducting and the adiabatic boundary conditions, Table 4.11 gives a tabulation of Nu as a function of Ra for values of H / L of 0.5, 1, 2, and 5, as calculated by Catton et al. [35] (and reported by Catton [34]) for very large Pr, by Wong and Raithby [284] and Raithby and Wong [230] for Pr = 0.7, and by Le Qu6r6 [176]. The effect of Pr over
Values of Y1and Y2to Be Used in Eq. 4.98 for AirFilled Cavities in Multicellular Arrays [253]
TABLE 4.10
L/H
ew
eh
ec
kL/kwb
71 × 106
"Y2
3 5 5 5 5 10
0.13 0.13 0.9 0.9 0.9 0.13
0.065 0.065 0.065 0.9 0.9 0.065
0.065 0.065 0.065 0.065 0.9 0.065
100 166 42 42 42 332
1.274 1.324 0.970 1.524 4.76 3.952
0.415 0.474 0.594 0.430 0.511 0.502
See Fig. 4.33 for the meaning of b.
4.54
CHAPTER FOUR the r a n g e 0.7 < Pr < oo is seen to be quite modest. But for Pr < 0.7, the effect of Pr has b e e n f o u n d to be stronger: the effect of low P r a n d t l n u m b e r on cavities with adiabatic walls and H / L = 1 was investigated, using a C F D code, by L a g e and B e j a n [173]. T h e y f o u n d that at R a = 1 x 105, N u w e n t f r o m 4.9 at Pr = 1, to 3.35 at Pr = 0.1, and to 2.77 at Pr = 0.01. T h e s e w o r k e r s also give a criterion to establish w h e t h e r the flow is l a m i n a r or t u r b u l e n t . E x t e n s i o n to Table 4.11 to higher R a a p p e a r s only to have b e e n m a d e for the " b e n c h m a r k " configuration (i.e., adiabatic walls, Pr = 0.7). Thus K u y p e r et al. [171] c o r r e l a t e d their C F D results by the e q u a t i o n s N u = 0.171 R a °'282 for
104 < R a < 108
(4.99a)
N u = 0.050 R a °341 for
108 < R a < 1012
(4.99b)
E q u a t i o n 4.99b fits predictions o b t a i n e d using a t u r b u l e n c e model. H s i e h and W a n g [146] corr e l a t e d their high R a y l e i g h n u m b e r e x p e r i m e n t a l results on a d i a b a t i c  w a l l e d cavities by the equations N u = 0.321 and
R a 0"241
( H / L ) °'°95 Pr °'°53 for
N u = 0.133 R a °3°1 ( H / L ) °°95 Pr °°53 for
106 < R a < 1.4 x 107
(4.100)
R a > 1.4 x 107 ~ R a ~ 2 x 109
(4.101)
This e q u a t i o n pair was derived f r o m d a t a covering the range 0.7 < Pr < 464 and 3 < H / L < 5.
H e a t Transfer in Vertical Cavities With W l L ~ 5: A n Overview. Figure 4.34 p r e s e n t s a c o m p i l a t i o n of s o m e of the d a t a of the previous t h r e e sections in terms of N u versus H/L, with R a as a p a r a m e t e r , for adiabatic walls with ew = 0 and Pr = 0.7. T h e figure shows a p e a k that m o v e s to lower values of H / L as R a is increased. TABLE 4.11 Tabulation of Numerically Computed Nusselt Number for Vertical (0 = 90 °) Rectangular Parallelepiped Cavities Having W/L ~> 5 and 0.5 < H/L < 5.
H/L 2
5
Prandtl number
0.5 oo
0.7
1 oo
0.7
0.7
Reference
35
176, 230
35
230
230
1.05 ~ 1.77 2.50 3.66 ~
1.11 1.42 1.97 2.61 3.53 ~
1.05 1.28 1.81 2.45 3.30
1.12 2.24 3.16 4.52 ~ ~ ~
1.19 1.64 2.34 3.12 4.26 
1.09 1.39 2.00 2.72 3.68
Perfectly conducting walls Ra = 1 0 3 3 x 10 3 104 3 x 10 4 105 3 x 105
1.00 1.01 1.07 1.48 2.51 3.64
1.05 1.25 1.75 2.41 3.40 4.47 Adiabatic walls
Ra = 103 3 x 103 104 3 x 10 4 105
1.00 1.05 1.30 2.18 3.82
10 6

107 108
~ 
See Fig. 4.25 for meaning of symbols.
1.12 1.50 2.24 3.14 4.51 8.83 16.52 30.22
NATURAL CONVECTION
I
I I
I
I I I I 1 
1
I
I
1 I I 1I I
'
!
4.55
I 1 1 l__l
c
I0.I
p_La.rr'Y1.0
~
J'I"r~ Ji
I0
l
L
In 4 0 oo
H/L FIGURE 4.34 N u as a f u n c t i o n o f H / L f o r various values o f R a f o r a v e r t i c a l r e c t a n g u l a r p a r a l l e l e p i p e d cavity w i t h W/L >~ 5, Pr = 0.7, and ~w = 0 (no r a d i a t i o n effects). F o r H / L < 0.5, the p l o t is based on Eq. 4.98; f o r 0.5 _ 8 can resemble that in the corresponding horizontal cavity or that in the corresponding vertical cavity; it rarely combines the characteristics of both. Consequently, with few exceptions, the Nusselt number in the inclined cavity can be determined, to a reasonable approximation, from either the vertical or the horizontal Nusselt number relation, by means of simple angular scaling laws. In this section Null (Ra) will refer to the Nusselt numberRayleigh number relation for a horizontal cavity (as determined by methods given in the section on natural convection in these cavities) having the same values for all the other relevant dimensionless groups as the inclined cavity at hand. Similarly, Nuv (Ra) will refer to the Nu (Ra) relation for the corresponding cavity at 0 = 90 ° (as determined by methods given in the section on heat transfer in vertical rectangular parallelepiped cavities), while Nu0 (Ra) will be the sought Nu (Ra) relation at the angle 0.
A n g u l a r Scaling.
4.56
CHAPTER FOUR
The scaling laws are found to be slightly dependent on the Prandtl number; the laws will first be reported for Pr ~> 4 (nonmetallic liquids), then for Pr = 0.7 (gases).
Cavities with Pr >~4 and WItt >_ 8. For 90 ° < 0 < 180 °, i.e., cavities heated from above, the scaling law suggested by Arnold et al. [7], Nue (Ra) = 1 + (Nuv (Ra)  1) sin 0
(4.103)
has been experimentally validated by Arnold et al. [8] for cavities with H/L = 1, 3, 6, and 12. For 0 < O < 90 ° (heating from below), two scaling laws are particularly useful: the horizontal scaling law of Clever [65], Nue (Ra) = Null (Ra cos O)
(4.104)
Nu0 (Ra) = Nuv (Ra sin 0)
(4.105)
and the vertical scaling law,
for H/L _>6, the maximum of the values of Nu0 given by each is recommended, i.e., Nu0 (Ra) = [Null (Ra cos 0), Nuv (Ra sin
8O
6o 2O I
1
I
2
4
6
I
I
8 I0 H/L
I
I
I
12
14
16
F I G U R E 4.35 Plot of the crossover angle 0c governing the transition from horizontallike flow to verticallike flow in an inclined cavity containing nonmetallic liquids. Adapted from Arnold et al. [8]. 0c is in degrees.
0)]ma x
(4.106)
Equation 4.106 yields a single value of the crossover angle Oo defined so that for 0 < 0c the horizontal scaling law applies and for 0 > 0c the vertical scaling law applies. Angle 0c is obtained by equating Nun (Ra cos 0) to Nuv (Ra sin 0) and solving for 0. This angle (which also locates a minimum in Nu0 when Nu0 is plotted against 0 with Ra held constant) is plotted as a function of H/L in Fig. 4.35, as given by Arnold et al. [8]. Equation 4.106, with a different but very similar vertical scaling law, was validated by Arnold et al. [8] for H/L = 6 and 12 and for 104 _ 1.2, it can be similarly approximated by (I)(11)= (2 cosh 11 1)/(2 cosh 11). As was the case for cylinders, Nu is expected to be independent of E when Ra is large enough to make A i I mo < L  E. Using a modified conduction layer method, Raithby and Hollands [223] obtained an explicit relation for the Nusselt number Nu: namely Eq. 4.121 with NUCONDgiven by Eqs. 4.1254.127 (or given by unity if E = 0) and Nut given by
  [ L ~1/4 Nu/
1.16Cik~///""
Ral/4 [(Di/Do)315 _{_(Do[Di)4/5] 5/4
(4.128)
This equation was shown to closely fit the E =0 data of Scanlan et al. [241], which covered the ranges 1.3 x 103 < Ra < 6 × 108, 5 too), steadystate convection is achieved. These two regimes are called the conduction regime (0 < t < to) and the steadystate regime (t > t~). The transition regime lies in the range to < t < too. At time ti, the heat flow by conduction matches the steadystate convective heat transfer from the body. If conditions are met to initiate convection before ti (this will depend on Pr and Ra), the heat flow falls monotonically in the transition regime, as along path A in Fig. 4.37b. Otherwise convection will not be initiated until to > ti, so the heat flow will have fallen below the steadystate value and must therefore recover from the undershoot in the transition regime as shown by path B.
TI
I
~r"T" ~¢1"oo q"l i ~ to Steady_T,,Too'
2
V
,o,e \
xc; from the experiments [148] for water, Xc is given roughly by Eq. 4.161 with K = 155. Integrating the local heat transfer relations results in the following expression for average heat transfer: Nu = V~0.664Re lr2 P r =
[29Re 1 Gr2/3 , 1
1/3 + ( 1  ~ ) C U R a 1/3
Ra = Gr Pr
(4.162a) (4.162b)
min
where the minimum of the two quantities in brackets is to be used in Eq. 4.162b. The coefficient 29 in Eq. 4.162b will likely be somewhat dependent on Prandtl number, but data are not available to resolve this dependence. For laminar flow above a cooled surface or below a heated surface, the presence of buoyancy forces stabilizes the flow (inhibits transition) and tends to diminish the heat transfer. The analysis of Chen et al. [43] predicts that natural convection will alter the local convective heat transfer by less than 5 percent if IGrl(x/L)3rZ/Re 5r2 < 0.03 for Pr = 0.7. Robertson et al. [232] show that for IGr I/Re 5r2 > 0.8 and Pr = 0.7, buoyancy may inhibit the flow so strongly that a separation bubble may form over the surface. In turbulent flow, stable stratification significantly damps turbulence and reduces heat transfer in the vertical direction. H o r i z o n t a l Cylinders. For a heated horizontal cylinder in perpendicular cross flow, the angle of the approaching stream, ~ in Fig. 4.47, greatly affects the heat flow in the DV mixed convection regime. For ~ = 0 the forced flow assists the Re= T natural convection and the dependence of the average NusT gB (T.T®) D~ selt number on Re resembles path A in Fig. 4.44. For ~)= 90 ° Ro vii there is a sharper transition from natural to forced convecRa tion than when ~ = 0, while for opposed flow ((~ = 180 °) there Gr Pr is a minimum as shown by path B in Fig. 4.44. For a cooled g cylinder the same description applies except that ~) is measured from the vertical axis extending upward from the Tw > T~ ~ . . cylinder. Equating the Nusselt numbers for pure natural convecFIGURE 4.47 Perpendicular flow across a horizon tion and pure forced convection provides a good estimate of the RaRe curve along which mixed convection effects are tal circular cylinder in mixed convection. most important, as already discussed. After a careful study of available data, Morgan [198] proposed the following equation for forced convection heat transfer from a cylinder for cross flow in a lowturbulence airstream: Nu =
q"D (TwT~) k
NUF = a Re"
(4.163)
where a and n are given in Table 4.14. This equation can also be used as a first approximation for other fluids if the right side is multiplied by the factor (Pr/0.71) 1/3. If Eq. 4.163 is equated to Eq. 4.45 for NuN, the Re/versus Gri relation indicated by the solid curve in Fig. 4.48 is obtained, which denotes the approximate center of the mixed convection regime. The approximate bounds of this regime, based on a 5 percent deviation in heat transfer from pure forced convection and from pure natural convection, respectively, have been estimated by Morgan [198] to lie in the shaded bands.
NATURAL CONVECTION
4.77
Constants for Forced Convection Over a Circular Cylinder (Eq. 4.163)
TABLE 4.14
Re range
a n
104 to 4 x 1 0 3
4 × 1 0 3 to 9 x 102
9 X 1 0 2 to 1.0
1.0 to 35
35 to 5 × 10 3
5 X 1 0 3 to 5 × 10 4
5 x 104 to 2 x 105
0.437 0.0895
0.565 0.136
0.800 0.280
0.795 0.384
0.583 0.471
0.148 0.633
0.0208 0.814
A procedure for calculating the heat transfer in the mixed convection regime for the problem has also been proposed by Morgan [198] on the basis of work of B6rner [22] and Hatton et al. [131]. For a given Ra and Re, the value of NuN is computed from Eq. 4.45. For the given Re, the constants a and n are chosen from Table 4.14. The value of Re/is then found from Eq. 4.163 with NUF= NuN; that is, Re~=
a(Pr/0.71)l/3
(4.164)
An effective Reynolds number Re~, is then calculated from Nee. = [(Re/+ Re cos ,)2 + (Re sin (i))211/2
(4.165)
and Nu is computed by insertion of Re~. into Eq. 4.163 to obtain ( Pr /1/3 Nu = a RebUff\ 0.71 ] It will be seen that if Rei >> Re, the natural convection result is recovered, while if Rei 0.5), very little difference exists b e t w e e n the Nusselt n u m b e r for uniform wall t e m p e r a t u r e and the Nusselt n u m b e r for uniform wall heat flux in smooth circular ducts. However, for Pr < 0.1, there is a difference b e t w e e n NuT and NUll. Table 5.11 presents the fully developed t u r b u l e n t flow Nusselt n u m b e r in a s m o o t h circular duct for Pr > 0.5. The correlation p r o p o s e d by Gnielinski [69] is r e c o m m e n d e d for Pr > 0.5, as are those suggested by Bhatti and Shah [45]. In this table, the f in the equation is calculated using the Prandtl [52]von K~irm~in [53]Nikuradse [43]; Coleb r o o k [54]; F i l o n e n k o [55]; or Techo et al. [56] correlations shown in Table 5.8.
FORCED
0.10
~ "
0.09
~
0.08
~
o.o7
.~
:~
, ,_uli i1,11
INTERNAL
FLOW
'
IIIIIIII IIIIIIII

J ~
J .: ., ., . 
IIIIIIII
IIIIIIIII
~ ' ~ ~ . ~ ~ " ~ . .~. . ~
~.
5.23
IIIlllll
i ~

0.O6
IN DUCTS
IIIIIII
]
~ ii;=u~iii~
'n e
CONVECTION,

=
~
_
"" 
~,
I I I I l l l l l
iiiiiiii
o.o5
"'"'":
0.03
' ' " " l ' 0""
..
mumwmmmm
'
0.04
_
i:'~
i "
0.025
0.~
o.ol~


t .....
~
~
ii ii i iI i~ I II
~
 "

.
.
.
.
.. .
.
.
.

. .
!
"
. .
' '
. .
__
_ ' L.
__
~
i~
. . J .... i i Ji ,
i iii
~
~ "" 0.0o8 _'_" o.o,
,.~
! ~~" ~ ~
~
I ! i iii
.~
"
~ i
i
IL !
103 2 ( 1 # ) 3 4 S6
J i Ji lid
i 2(1~)3
I1#
2(1#))
4 $6

",,,,,,,, '"'"'
 
~ ~~ _ ~
.
.
.
, ~ "  :
IIIIIIII
0 0 ~
0.006
nnlmlgl
7,;;;;;;; ~,,~,,
mnnmmml l l l l l l l
0.002 O.®t
0.0008 :::::::: 0.0006 mnuummmu 0.0o04 nuumnnn llllllll
~ ~
IlUlUl

It l Il nIl lImIl nI I I
0.0001
0.000,05

~ ~::" r~' '
~ ,nummmu~.mn, ,,,,,,, o ooo.= mllNnmllllt: ": mUllllmnnnnnnm
I1#
~ 2(1~)3
Re
E
Dh
Illlillm
"
i: 4 s6
.,,n,,,,, nuummunnnum 0.01
::t:::t::
. ~ ~ : ~ ,,. ..,., ;,,.,. . . .
' __1
~!~!
o.oo9
_
~ ~ ....

~ 
I:

~
_ , : : ::
. . . .
~
~).o00.0,
4 S6 l l O ? ' ~ : t n u .
/
t = 0.000,001 ~,
t~= 0.000,005
FIGURE 5.9 Moody's [58] friction factor diagram for fully developed flow in a rough circular duct [45].
For liquid metal (Pr < 0.1), the most accurate correlations for NUT and NUll a r e those put forth by Notter and Sleicher [80]: NUT = 4.8 + 0.0156Re °'85 P r °'93
(5.76)
Null = 6.3 + 0.0167Re °85 P r °'93
(5.77)
These equations are valid for 0.004 < Pr < 0.1 and 10 4 < Re < 10 6. Heat Transfer in Rough Circular Ducts. The Nusselt number for a complete, rough flow regime in a circular duct is given in Table 5.12. The term f i n this table denotes the friction factor for fully rough flow. It is given by the Nikuradse [60] correlation shown in Table 5.9. The recommended equations for practical calculations are those correlations by Bhatti and Shah [45] shown in Table 5.12. Artificially roughed circular ducts are also often used to enhance heat transfer. The Nusselt numbers for artificially roughed ducts have been reviewed by Rao [59].
Hydrodynamically Developing Flow.
An analytical, closeform solution for hydrodynamically developing flow in rough circular ducts has been obtained by Zhiqing [87]. The velocity distribution in the hydrodynamic entrance region is given as u ~(y/8) in Umax   [ 1
Umax
"~
for O < y < 5 for ;5 < y < a
(5.78)
+
(5.79)
where ;5 is the hydrodynamic boundary layer thickness, which varies with axial coordinate x in accordance with the following relation:
5.24
CHAPTER FIVE
TABLE 5.9 Fully Developed Turbulent Flow Friction Factor Correlations for a Rough Circular Duct [48] (a = tube radius)
Investigators
Correlations
1
von Kfirm~in [46]
Mf
Nikuradse [60]
~
1
vf
Remarks
E
3.36 1.763 In a
This explicit theoretical formula is applicable for Re~ > 70.
e
 3.481.737 I n a
This experimentally derived formula renders very nearly the same results as the von Kfirm~in [46] formula.
1
Colebrook [54]
 ~ = 3.481.7372 In
Moody [58]
f = 1.375
Wood [61]
f = 0.08(~)°'225 + 0.265(ae) + 66.69(ae)°4 Re "
[ x 103
+ ReV~)
1 + 21.544
(~.
This implicit formula is applicable for 5 < Re~ < 70, spanning the transition, hydraulically smooth, and completely rough flow regimes.
100'~lc3] + ~] ]
Shows a maximum derivation of15.78% from the ColebrookWhite equation for 4000 < Re < 108 and 2 x 108 < e/a < 0.1.
/ e ~0.1M
where n = 1.778~a ) Swamee and Jain [62]
Churchill [63]
1 ~=
3.4769
1.7372 In [e
a

_
+
42.48] ReO.9
Shows a maximum deviation of 3.19% from the ColebrookWhite equation for 4000 < Re < 108 and 2 x 108 < rda < 0.1.
2[( 8 ~12÷ 1 11/12 f= L\Re ] (A1 + B1)3r2
Unlike other equations in the table, this equation applies to all three flow regimesmlaminar, transition, and turbulent. Its predictions for laminar flow are in agreement with f = 16/Re. The predictions for transition flow are subject to some uncertainty. However, the predictions for turbulent flow are comparable with those rendered by the preceding equation.
where A1 = {2.2088 + 2.457 In [ ea + 42"683 16R 09e ]}
Re
Chen [64]
Round [65] Zigrang and Sylvester
1
 ~ = 3.481.7372 In
E;  16Re"6 In A21
where (e/a) 11°98 (7.149) °~1 A 2 = 6.0983 + Re 1 [96.2963] ,   4.2146 1.5635 In e a Re vf
1
 ~ = 3.4769  1.7372 In

R~ In A3
[661 where ,43 = 7~  2.1802 In Zigrang and Sylvester
~=
3.4769  1.7372 In
Applicable only for rda > 2 x 105; shows a maximum deviation of 6.16% from the ColebrookWhite equation for 4000 < Re < 10a and 2 x 108 < e/a < 0.1.
This explicit equation is consistently in good agreement with the ColebrookWhite equation for 4000 < Re < 10a and 2 x 108 _<e./a < 0.1, the maximum deviation being 0.39%.
Comparable with Moody's equation.
]
Shows a maximum deviation of +0.96% from the ColebrookWhite equation for 4000 _< Re < 10s and 2 x 108 < rda < 0.1
 16.1332 In ,43  16.1332 In A4
[661 where e/a [cda 13] A, = ~.4  2.1802 In  ~ +
Predictions not substantially different from those of the preceding equation.
5.25
FORCED CONVECTION, INTERNAL FLOW IN DUCTS TABLE 5.9
Fully Developed Turbulent Flow Friction Factor Correlations for a Rough Circular Duct [48] (Continued)
Investigators
Correlations
Haaland [67]
1  ~ = 3.4735  1.5635 In
Serghides [68]
1 (As B2)2 V~  As  As  2B + C1 where
(e./a
As = 0.8686 In \ 7.4 +
Remarks
6,6 ,0] + Re
Shows a maximum deviation of +1.21% from the ColebrookWhite equation for 4000 < Re < 108 and 2 x 108 < rda < 0.1. Shows a maximum deviation of +0.14% from the ColebrookWhite equation for 4000 < Re < 108 and 2 x 108 < e./a< 0.1.
12 )
(rda) 2.5A5 B2 = 0.86868 In  ~ + Re
( e.la
C1 = 0.8686 In \ ~ Serghides [68]
1 X/f
4.781
+
2.51B2) Re
(As 4.781) 2 (4.781  2A5 B2)
Shows a maximum deviation of 0.45 % from the ColebrookWhite equation for 4000 < Re < 108 and 2 x 108 < rda < 0.1. ,
X/DhRe TM _
1.4039(~)5/4[1 + 0.1577(~)  0.1793(~) 2  0.0168(~) 3 + 0.0064(~) 4]
(5.80)
The axial pressure drop Ap*, the incremental pressure drop n u m b e r K(x), and the apparent Fanning friction factor fapp are given as follows [87]"
Ap* = ( Umax]21
(5.81)
\Um]
K(X) lapp Rel/4 =
X/Dh
Ap*  0.316 Re1/z Ap*
4x/(Dh Re °25)
(5.82)
(5.83)
TABLE 5.10 Average Roughness of Commercial Pipes
Material (new)
Roughness e (mm)
Riveted steel Reinforced concrete Wood Cast iron Galvanized steel Asphalted cast iron Bitumencoated steel Structural and forged steel Drawn tubing Glass
0.99 0.33 0.180.9 0.26 0.15 0.12 0.12 0.045 0.0015 Smooth
5.26
CHAPTER FIVE
]'ABLE 5.11 Fully Developed Turbulent Flow Nusselt Numbers in a Smooth, Circular Duct for Gases and Liquids (Pr > 0.5) [48] Investigators Dittus and Boelter [70]
Correlations ~0.024 Re °8 Pr °4 Nu = [0.026 Re °8 Pr °'3
Colburn [71]
Nu = (f/2) Re Pr 1/3 Nu = 0.023 Re °8 Pr ~/3
von K~irm~in [72]
Nu =
Application range
for heating for cooling
0.7 < Pr < 120 and 2500 < Re < 1.24 x 105, L/d > 60 0.5 < Pr < 3 and 1 0 4 < Re < 105
(f/2) Re Pr l+5(fi2)a~[Prl+ln(
0.5 < Pr < 10 and 5Pr+16
10 4
500.
Transition Flow
As seen in the previous section, flow is considered to be laminar when R e < 2300 and turbulent when Re > 104. Transition flow occurs in the range of 2300 < Re < 104. Few correlations or formulas for computing the friction factor and heat transfer coefficient in transition flow are available. In this section, the formula developed by Bhatti and Shah [45] is p r e s e n t e d to compute the friction factor. It follows: B f = A + Re1/~
(5.97)
E q u a t i o n 5.97 is applicable to the laminar, transition, and turbulent flow regions. For laminar flow (Re < 2100), A = 0, B = 16, and m  1. For transition flow, 2100 < Re _< 4000, A 0.0054, B  2.3 x 10 8, and m = 2/3. For turbulent flow, (Re > 4000) A  1.28 x 10 3, B = 0.1143, and m = 3.2154. Blasius's [49] formula (see Table 5.8) is also applicable for calculating the friction factor in the range of 4000 < R e < 10 s.
FORCED CONVECTION,INTERNALFLOWIN DUCTS NusseltNumber Ratios for a Smooth Circular Duct with Various Entrance Configurations for Pr = 0.7 [2]
TABLE 5.13
Entrance configurations Long calming section
Schematics Adiabaticsurface
q~,
C 0.9756 0.760
,..,.,~.,,,,.,,,,,,I I IiiI
"~'~"''~"''tti"ttt Square entrance
q~,
2.4254 0.676
I~tilliilliili [~ttttit
f't t f f
180° Round bend
0.9759 0.700
l ii I i i l il~ 1.0517 0.629 90°R°undbend
~~l
90° Elbow
~
~ I ~ i~
N
~ I I i
""~tt~ t ttt ti
2.0152 0.614
5.31
5.32
CHAPTER FIVE
Heat transfer results for transition flow are rather uncertain due to the fact that so many parameters are needed to characterize heataffected flow. In the range of 0 < Pr < ~ and 2100 _ 1, T i e d t [106] also d e m o n strated that 24 f R e = 1 + 1.5e .2
(5.157)
It s h o u l d b e n o t e d t h a t w h e n e* = 0, t h e e c c e n t r i c a n n u l a r d u c t is r e d u c e d to a c o n c e n t r i c a n n u l a r duct. C h e n g a n d H w a n g [108] a n a l y z e d t h e h e a t t r a n s f e r p r o b l e m in e c c e n t r i c a n n u l a r ducts. T h e N u s s e l t n u m b e r s f o r fully d e v e l o p e d f l o w in e c c e n t r i c a n n u l a r d u c t s w i t h t h e t ~ a n d t h e r m a l b o u n d a r y c o n d i t i o n s a r e given in T a b l e 5.26. F o r e c c e n t r i c a n n u l a r d u c t s w i t h b o u n d a r y c o n d i t i o n s d i f f e r e n t f r o m t h e f o u r d e s c r i b e d in t h e s e c t i o n e n t i t l e d " F o u r F u n d a m e n t a l
TABLE 5.26
Nusselt Number
Num and NUll2for Fully Developed
r*
e* = 0
0.01
0.05
0.1
0.2
0.25 0.50 0.75 0.90
7.804 8.117 8.214 8.232
7.800 8.111 8.208 8.226
~ ~
7.419 7.608 7.659 7.667
6.524 6.473 6.432 6.422
0.02 0.05 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 0.95
m m m ~ m ~ ~ 
m m ~ ~ ~ ~ ~
6.347 6.777 7.139 7.430 7.452 7.267 6.839 6.064 4.776 3.273 0.887 0.229
6.224 6.582 6.815 6.793 6.395 5.688 4.691 3.463 2.154 1.002 0.246 0.059
5.826 5.956 5.854 5.191 4.241 3.198 2.206 1.364 0.725 0.299 0.068 0.016
Laminar Flow in Eccentric Annular Ducts [1,108]
0.4
0.6
0.8
0.9
0.95
0.99
4.761 4.393 4.227 4.192
3.735 3.247 3.024 2.975
3.203 2.644 2.384 2.324
3.038 2.446 2.171 2.106
~ 
2.925 2.305 2.016 1.947
4.913 4.701 4.228 3.191 2.244 1.484 0.918 0.522 0.262 0.104 0.023 0.005
4.537 4.289 3.518 2.517 1.694 1.080 0.650 0.362 0.179 0.070 0.015 0.003
4.157 3.849 3.334 2.360 1.571 0.990 0.590 0.326 0.160 0.062 0.013 0.003
4.149 3.827 3.322 2.361 1.571 0.988 0.586 0.323 0.158 0.061 0.013 0.003
4.160 3.833 3.341 2.370 1.575 0.989 0.586 0.322 0.157 0.061 0.013 0.003
Num
N U l l 2
m
m
m
m
5.50
CHAPTERFIVE Thermal Boundary Conditions," caution must be taken in using the superposition technique. The reader is strongly recommended to consult the literature.
Turbulent Flow Presented in this section are the friction factor and Nusselt number for turbulent flow and heat transfer in concentric annular ducts. The effects of eccentricity on the friction factor and Nusselt number are also discussed. Critical Reynolds Number. For concentric annular ducts, the critical Reynolds number at which turbulent flow occurs varies with the radius ratio. Hanks [109] has determined the lower limit of Recrit for concentric annular ducts from a theoretical perspective for the case of a uniform flow at the duct inlet. This is shown in Fig. 5.16. The critical Reynolds number is within +_3 percent of the selected measurements for air and water [109].
Fully D e v e l o p e d
Flow. Knudsen and Katz [110] obtained the following velocity distributions for fully developed turbulent flow in a smooth concentric annular duct in terms of wall coordinates u ÷ and y+:
2500
Uo÷ = 3.0 + 2.6492 In yo+
for rm < r < ro
(5.158)
uT = 6.2 + 1.9109 In y~
for ri < r < rm
(5.159)
ii
iii
i
l
i
I
I
I
2400
2300 Recnt 2200
ro 2100
2000
I__ 0
0.2
m 0.4
I
I
0.6
0.8
1.0
r* FIGURE 5.16 Lower limits of the critical Reynolds numbers for concentric annular ducts with uniform velocity at the inlet [109].
F O R C E D C O N V E C T I O N , I N T E R N A L F L O W IN D U C T S
where
Uo+ 
U
uT 
Ut, o
Y +o 

and
ut,o(ro 
5.51
U Ut, i
r)
yi ÷=
V
Ut,o =
ut, i(r 
ri)
V
(5.160)
Ut, i =
The radius of maximum velocity r* in Eqs. 5.158 and 5.159 can be determined by the formula obtained by Kays and Leung [111]. It follows: r* = rm = r,0.343(1 + ro
r *0"657 
r*)
(5.161)
A critical review of the extensive friction factor data has been made by Jones and Leung [112]. The researchers recommend that the fully developed friction factor formulas for smooth circular ducts given in Table 5.8 be used for calculating the friction factor for concentric annular ducts by replacing 2a with the laminar equivalent diameter Dt for concentric annular ducts. The term Dt is defined by 1 + r .2 + (1  r 2*)/In r*
Dr= Dh
(1  r*) 2
(5.162)
where r* = ri/ro; Dh = 2 ( r o  ri) and r~ and ro are the radii of the inner and outer tubes, respectively. The fully developed Nusselt numbers Nuo and Nui at the outer and inner walls of a smooth concentric annular duct can be determined from the following relations for uniform wall heat fluxes qo' and q;' at the outer and inner walls:
where
Nuo
hoDh k 
Nui
hiDh k  1
q~= ho(To  Tm),
1


Nuoo ,, ,, , ( q i ]qo )0o
(5.163)
Nuii ,, ,, , (qo/qi )Oi
(5.164)
q;" = h~(Ti  Tm)
(5.165)
The terms To and T/denote the duct wall temperatures at the outer and inner walls. The temperature difference T o  Ti is given by Dh[ ( 1 0* / 7 ( 1~ 0* )] To  Ti = if q'o" Nuoo + Nuii ]  q \ Nuii + Nuoo
(5.166)
The Nusselt numbers Nuoo and Nuu., as well as the influence coefficients 0* and 0* in Eqs. 5.163, 5.164, and 5.166 are provided by Kays and Leung [111]. These are given in Table 5.27 for wide ranges of Re and Pr and for r* = 0.1, 0.2, 0.5, and 0.8. For r* = 1, the concentric annular duct is reduced to a parallel plate duct. The applicable results are given in Table 5.28, the simple Nu being used for the Nusselt number at the heated wall. It should be noted that for laminar flow (Re < 2300) in parallel plate ducts, Nu is equal to 5.385 and 0* is equal to 0.346 for all values of Pr. Dwyer [113] has developed semiempirical equations for liquid metal flow (Pr < 0.03) in a concentric annular duct (0 < r* < 1) with one wall subjected to uniform heat flux and the other
5.52
CHAPTER FIVE
TABLE 5.27 Nusselt Numbers and Influence Coefficients for Fully Developed Turbulent Flow in a Concentric Annular Duct with Uniform Heat Flux at One Wall and the Other Wall Insulated [111] r* = 0.10
Heating from outer wall with inner wall insulated Re
3
= 10 4
105
× 10 4
3 × 105
10 6
Pr
Nuoo
0*
Nuoo
0*
Nuoo
0*
Nuoo
0*
Nuoo
0*
0 0.001 0.003 0.01 0.03 0.5 0.7 1.0 3 10 30 100 1000
6.00 6.00 6.00 6.13 6.45 24.8 29.8 36.5 61.5 99.2 143.0 205.0 378.0
0.077 0.077 0.077 0.076 0.076 0.039 0.032 0.026 0.013 0.006 0.003 0.002
6.12 6.12 6.24 6.50 7.95 53.4 66.0 81.8 147.0 246.0 360.0 525.0 980.0
0.079 0.079 0.081 0.081 0.075 0.032 0.028 0.023 0.013 0.006 0.003 0.002
6.32 6.40 6.55 7.80 13.7 134 167 212 395 685 1030 1500 2850
0.081 0.082 0.083 0.077 0.065 0.028 0.024 0.021 0.012 0.006 0.003 0.002
6.50 6.60 7.34 12.1 28.2 320.0 409.0 520.0 1000.0 1780.0 2720.0 4030.0 7600.0
0.084 0.082 0.082 0.067 0.051 0.025 0.022 0.019 0.012 0.006 0.003 0.002
6.68 7.20 10.8 26.4 71.8 860.0 1100.0 1430.0 2830.0 5200.0 8030.0 12,100 23,000
0.085 0.082 0.071 0.052 0.036 0.022 0.020 0.017 0.011 0.006 0.003 0.002
r* = 0.10
Heating from inner wall with outer wall insulated Re = 104
3
105
X 10 4
3 x 105
10 6
Pr
Nuii
0'~
Nuii
0'~
Nuii
O*
Nuii
O*
Nuii
O*
0 0.001 0.003 0.01 0.03 0.5 0.7 1.0 3 10 30 100 1000
11.5 11.5 11.5 11.8 12.5 40.8 48.5 58.5 93.5 140.0 195.0 272.0 486.0
1.475 1.475 1.475 1.472 1.472 0.632 0.512 0.412 0.202 0.089 0.041 0.017 0.004
11.5 11.5 11.5 11.8 14.1 81.0 98.0 120.0 206.0 328.0 478.0 673.0 1240.0
1.502 1.502 1.475 1.442 1.330 0.486 0.407 0.338 0.175 0.081 0.039 0.015 0.003
11.5 11.5 11.7 13.5 21.8 191.0 235.0 292.0 535.0 890.0 1320.0 1910.0 3600.0
1.500 1.480 1.473 1.323 1.027 0.394 0.338 0.286 0.162 0.078 0.038 0.015 0.003
11.5 11.7 12.6 19.4 42.0 443.0 550.0 700.0 1300.0 2300.0 3470.0 5030.0 9600.0
1.460 1.462 1.391 1.090 0.760 0.339 0.292 0.256 0.152 0.078 0.038 0.016 0.004
11.6 12.3 17.0 39.0 103.0 1160.0 1510.0 1910.0 3720.0 6700.0 10,300.0 15,200.0 28,700.0
1.477 1.410 1.124 0.760 0.526 0.294 0.269 0.232 0.148 0.077 0.040 0.018 0.004
r* = 0.2
Heating from outer wall with inner wall insulated Re
3
= 10 4
105
× 10 4
3 × 105
10 6
Pr
Nuoo
0o*
Nuoo
0o*
Nuoo
0o*
Nuoo
0o*
Nuoo
0o*
0 0.001 0.003 0.01 0.03 0.5 0.7 1.0 3 10 30 100 1000
5.83 5.83 5.83 5.95 6.22 22.5 29.4 35.5 60.0 98.0 142.0 205.0 380.0
0.140 0.140 0.140 0.140 0.140 0.071 0.063 0.051 0.026 0.013 0.004 0.003 0.001
5.92 5.92 6.00 6.20 7.55 51.5 64.3 80.0 145.0 243.0 360.0 520.0 980.0
0.145 0.144 0.146 0.146 0.140 0.064 0.055 0.046 0.026 0.013 0.006 0.003 0.001
6.10 6.10 6.22 7.40 12.7 130.0 165.0 206.0 390.9 680.0 1030.0 1500.0 2830.0
0.151 0.151 0.150 0.144 0.125 0.055 0.049 0.042 0.024 0.012 0.006 0.003 0.001
6.16 6.30 6.90 11.4 26.3 310.0 397.0 504.0 980.0 1750.0 2700.0 4000.0 7500.0
0.152 0.154 0.150 0.131 0.098 0.049 0.044 0.039 0.024 0.012 0.006 0.003 0.001
6.35 6.92 10.2 24.6 80.0 823.0 1070.0 1390.0 2760.0 4980.0 7850.0 12,000.0 22,500.0
0.157 0.153 0.136 0.102 0.074 0.044 0.040 0.035 0.023 0.012 0.006 0.003 0.001
FORCED CONVECTION, INTERNAL FLOW IN DUCTS
5.53
T,~BI.I: 5.27 Nusselt Numbers and Influence Coefficients for Fully Developed Turbulent Flow in a Concentric Annular Duct with Uniform Heat Flux at One Wall and the Other Wall Insulated [111] (Continued) Heating from inner wall with outer wall insulated
r* = 0.2 Re
3
= 10 4
105
× 10 4
3 × 105
10 6
Pr
Nuii
Off
Nuii
O*
Nuii
O~
Nuii
0'~
Nuii
O~
0 0.001 0.003 0.01 0.03 0.5 0.7 1.0 3.0 10.0 30.0 100.0 1000.0
8.40 8.40 8.40 8.50 9.00 31.2 38.6 46.8 77.4 120.0 172.0 243.0 448.0
1.009 1.009 1.009 1.000 1.012 0.520 0.412 0.339 0.172 0.120 0.036 0.014 0.004
8.30 8.40 8.40 8.60 10.1 64.0 79.8 99.0 175.0 290.0 428.0 617.0 1400.0
1.028 1.040 1.027 1.018 0.943 0.398 0.338 0.284 0.151 0.074 0.034 0.014 0.002
8.30 8.30 8.50 9.70 15.8 157.0 196.0 247.0 465.0 800.0 1210.0 1760.0 3280.0
1.020 1.020 1.025 0.944 0.771 0.333 0.286 0.248 0.143 0.072 0.035 0.015 0.002
8.30 8.40 9.05 14.0 31.7 370.0 473.0 600.0 1150.0 2050.0 3150.0 4630.0 8800.0
1.038 1.014 0.980 0.796 0.600 0.295 0.260 0.229 0.137 0.073 0.036 0.016 0.004
8.30 8.90 12.5 33.6 81.0 980.0 1270.0 1640.0 3250.0 6000.0 9300.0 13,800.0 26,000.0
1.020 0.976 0.834 0.748 0.374 0.262 0.235 0.209 0.135 0.077 0.038 0.016 0.003
r* = 0.5
Heating from outer wall with inner wall insulated Re
3 x 104
= 10 4
105
3 x 105
10 6
Pr
Nuoo
0*
Nuoo
0*
Nuoo
0*
Nuoo
0*
Nuoo
0o*
0 0.001 0.003 0.01 0.03 0.5 0.7 1.0 3.0 10.0 30.0 100.0 1000.0
5.66 5.66 5.66 5.73 6.03 2.6 28.3 34.8 60.5 100.0 143.0 207.0 387.0
0.281 0.281 0.281 0.281 0.279 0.162 0.137 0.111 0.059 0.028 0.013 0.006 0.001
5.78 5.78 5.78 5.88 7.05 49.8 62.0 78.0 144.0 246.0 365.0 530.0 990.0
0.294 0.294 0.294 0.289 0.284 0.142 0.119 0.101 0.058 0.028 0.013 0.006 0.001
5.80 5.80 5.85 6.80 11.6 125.0 158.0 200.0 384.0 680.0 1030.0 1500.0 2830.0
0.296 0.296 0.294 0.289 0.258 0.123 0.107 0.092 0.055 0.028 0.014 0.006 0.001
5.83 5.92 6.45 10.3 24.4 298.0 380.0 490.0 960.0 1750.0 2700.0 4000.0 7600.0
0.302 0.302 0.301 0.264 0.214 0.111 0.097 0.085 0.054 0.028 0.014 0.006 0.001
5.95 6.40 9.00 22.6 64.0 795.0 1040.0 1340.0 2730.0 5030.0 8000.0 12,000.0 23,000.0
0.310 0.304 0.278 0.217 0.163 0.098 0.090 0.078 0.052 0.028 0.015 0.006 0.001
r* = 0.5
Heating from inner wall with outer wall insulated Re
3 x 104
= 10 4
105
3 x 105
106
Pr
Nuii
O*
Nuii
O*
Nuii
O*
Nuii
O*
Nuii
O*
0 0.001 0.003 0.01 0.03 0.5 0.7 1.0 3.0 10.0 30.0 100.0 1000.0
6.28 6.28 6.28 6.37 6.75 24.6 30.9 38.2 66.8 106.0 153.0 220.0 408.0
0.620 0.620 0.620 0.622 0.627 0.343 0.300 0.247 0.129 0.059 0.028 0.006 0.002
6.30 6.30 6.30 6.45 7.53 52.0 66.0 83.5 152.0 260.0 386.0 558.0 1040.0
0.632 0.632 0.632 0.636 0.598 0.292 0.258 0.218 0.121 0.059 0.027 0.006 0.002
6.30 6.30 6.40 7.30 12.0 130.0 166.0 212.0 402.0 715.0 1080.0 1600.0 3000.0
0.651 0.651 0.656 0.623 0.533 0.253 0.225 0.208 0.115 0.059 0.028 0.006 0.002
6.30 6.40 6.85 10.8 24.8 310.0 400.0 520.0 1010.0 1850.0 2850.0 4250.0 8000.0
0.659 0.659 0.637 0.540 0.430 0.299 0.206 0.183 0.114 0.059 0.031 0.007 0.002
6.30 6.75 9.40 23.2 35.5 835.0 1080.0 1420.0 2870.0 5400.0 8400.0 12,600.0 24,000.0
0.654 0.644 0.585 0.427 0.333 0.208 0.185 0.170 0.111 0.061 0.032 0.007 0.002
TABLE 5.27 Nusselt Numbers and Influence Coefficients for Fully Developed Turbulent Flow in a Concentric Annular Duct with Uniform Heat Flux at One Wall and the Other Wall Insulated [111] (Continued) r* = 0.8
Heating from outer wall with inner wall insulated Re = 104
3×
105
10 4
106
3 x 105
Pr
Nuoo
0o*
Nuoo
0o*
Nuoo
0o*
Nuoo
0o*
Nuoo
0o*
0 0.001 0.003 0.01 0.03 0.5 0.7 1.0 3.0 10.0 30.0 100.0 1000.0
5.65 5.65 5.65 5.75 6.10 22.4 28.0 34.8 61.3 100.0 146.0 209.0 385.0
0.379 0.379 0.379 0.381 0.388 0.225 0.192 0.159 0.083 0.039 0.019 0.008 0.002
5.70 5.70 5.70 5.85 6.90 48.0 61.0 76.5 142.0 243.0 365.0 533.0 1000.0
0.386 0.386 0.386 0.386 0.380 0.191 0.166 0.141 0.079 0.039 0.019 0.008 0.002
5.75 5.75 5.84 6.72 11.1 121.0 156.0 197.0 382.0 670.0 1040.0 1500.0 2870.0
0.398 0.398 0.397 0.390 0.339 0.169 0.150 0.129 0.078 0.039 0.020 0.009 0.002
5.80 5.88 6.35 9.95 23.2 292.0 378.0 483.0 960.0 1740.0 2720.0 4000.0 7720.0
0.407 0.406 0.407 0.361 0.290 0.153 0.136 0.120 0.076 0.040 0.021 0.009 0.002
5.85 6.25 8.80 21.0 62.0 790.0 1020.0 1330.0 2730.0 5050.0 8000.0 12,000.0 23,000.0
0.409 0.407 0.374 0.286 0.216 0.136 0.122 0.111 0.073 0.040 0.022 0.010 0.002
r* = 0.8
Heating from inner wall with outer wall insulated Re = 104
3 x 104
105
106
3 x 105
Pr
Nu,
0*
Nuii
0*
Nu,
0*
Nuii
0*
Nui~
0*
0 0.001 0.003 0.01 0.03 0.5 0.7 1.0 3.0 10.0 30.0 100.0 1000.0
5.87 5.87 5.87 5.95 6.20 22.9 28.5 35.5 63.0 102.0 147.0 215.0 393.0
0.489 0.489 0.489 0.485 0.478 0.268 0.244 0.200 0.108 0.051 0.027 0.010 0.002
5.90 5.90 5.90 6.07 7.05 49.5 62.3 78.3 145.0 248.0 370.0 540.0 1000.0
0.505 0.505 0.505 0.506 0.485 0.250 0.212 0.181 0.102 0.051 0.027 0.010 0.002
5.92 5.92 6.03 6.80 11.4 123.0 157.0 202.0 386.0 693.0 1050.0 1540.0 2890.0
0.515 0.515 0.485 0.493 0.445 0.214 0.186 0.166 0.097 0.052 0.027 0.010 0.002
5.95 6.00 6.40 10.0 23.0 296.0 384.0 492.0 973.0 1790.0 2750.0 4050.0 7700.0
0.525 0.518 0.504 0.452 0.357 0.193 0.172 0.154 0.096 0.051 0.029 0.011 0.002
5.97 6.33 8.80 21.7 61.0 800.0 1050.0 1350.0 2750.0 5150.0 8100.0 12,100.0 23,000.0
0.528 0.516 0.468 0.382 0.276 0.174 0.160 0.140 0.093 0.051 0.030 0.012 0.002
TABLE 5.28 Nusselt Numbers and Influence Coefficients for Fully Developed Turbulent Flow in a Smooth Concentric Annular Duct With r*  1 (Parallel Plates Duct With Uniform Heat Flux at One Wall and the Other Wall Insulated* [111] Re = 104
3 × 104
105
106
3 × 105
Pr
Nu
0"
Nu
0"
Nu
0*
Nu
0"
Nu
0*
0.0 0.001 0.003 0.01 0.03 0.5 0.7 1.0 3.0 10.0 30.0 100.0 1000.0
5.70 5.70 5.70 5.80 6.10 22.5 27.8 35.0 60.8 101.0 147.0 210.0 390.0
0.428 0.428 0.428 0.428 0.428 0.256 0.220 0.182 0.095 0.045 0.021 0.009 0.002
5.78 5.78 5.80 5.92 6.90 47.8 61.2 76.8 142.0 214.0 367.0 214.0 997.0
0.445 0.445 0.445 0.455 0.428 0.222 0.192 0.162 0.092 0.045 0.022 0.009 0.002
5.80 5.80 5.90 6.70 11.0 120.0 155.0 197.0 380.0 680.0 1030.0 1520.0 2880.0
0.456 0.456 0.450 0.440 0.390 0.193 0.170 0.148 0.089 0.045 0.022 0.010 0.002
5.80 5.88 6.32 9.80 23.0 290.0 378.0 486.0 966.0 1760.0 2720.0 4030.0 7650.0
0.460 0.460 0.450 0.407 0.330 0.174 0.156 0.138 0.087 0.045 0.023 0.010 0.002
5.80 6.23 8.62 21.5 61.2 780.0 1030.0 1340.0 2700.0 5080.0 8000.0 12,000.0 23,000.0
0.468 0.460 0.422 0.333 0.255 0.157 0.142 0.128 0.084 0.046 0.024 0.011 0.002
6.54
FORCED CONVECTION,INTERNALFLOW IN DUCTS
5.55
wall insulated. For the case of the outer wall's being heated, the semiempirical equations are as follows: Nuoo : Ao + Bo(~ Pe) n° 0.05
Ao = 5.26 + rg
where
Bo = 0.01848 + no = 0.78 
0.003154 0.0001333 r~r, 2 0.01333 0.000833 rT+ r,5~
1.82 13= 1  Pr (l~m/V)max 1.4
(5.167) (5.168)
(5.169)
(5.170)
(5.171)
where (l~m/V)maxcan be calculated from the relation 1
: 2" (V)max,c
(5.172)
An expression for (Em/V)max,c applicable to a circular duct (r* = 0) was developed by Bhatti and Shah [45]. It is given by
(ff~)max,c0.037Re V f
(5.173)
In Eq. 5.173, the friction factor f can be calculated from the explicit formula given by Techo et al. [56], which is shown in Table 5.8. For a concentric annular duct with the inner wall heated, the semiempirical equations developed by Dwyer [113] are applicable: Nuii = Ai + ni( ~ Pe) ni
where
0.686 rg
(5.175)
0.000043 r~
(5.176)
0.01657 0.000883 rZr,T
(5.177)
Ai = 4.63 +
Bi 0.02154 =
ni = 0.752 +
(5.174)
The values of 13for this case can also be calculated from Eqs. 5.171 to 5.173. Both Eqs. 5.166 and 5.173 are valid for Pe values above the critical values. For Pr = 0.005, 0.01, 0.02, and 0.03, the critical Pe values are 270, 300, 330, and 345, respectively. For liquid metals, only the heat transfer mode for Pe < Pe~t is molecular conduction.
Hydrodynamically Developing Flow. Hydrodynamically developing turbulent flow in concentric annular ducts has been investigated by Rothfus et al. [114], Olson and Sparrow [115], and Okiishi and Serouy [116]. The measured apparent friction factors at the inner wall of two concentric annuli (r* = 0.3367 and r* = 0.5618) with a square entrance are shown in Fig. 5.17 (r* = 0.5618), where 3] is the fully developed friction factor at the inner wall. The values of f equal 0.01, 0.008, and 0.0066 for Re = 6000, 1.5 x 104, and 3 x 104, respectively [114].
5.56
CHAPTERFIVE
3.5 Re 3.0 tot
104
lapp,
i
X X 10 4 10 4
X X 104 X 104
2.5 I
ri
(
, 2.0
1.5
1.0
0
~
I~
1~
2~
2~
xlD h
FIGURE 5.17 Normalized apparent friction factors for turbulent flow in the hydrodynamic entrance region of a smooth concentric annular duct (r* = 0.5168) [114].
Having determined fapp, i from Fig. 5.17, the apparent friction factor fapp,o at the outer wall can be determined from fapp,o fapp~
r*(1  r .2 ) r*2 _ r*Z
(5.178)
where r* is given by r* = r*°343(1 + r *0"657 r*) Having identified both fapp,o and calculated as follows:
fapp,i, the
(5.179)
circumferentially averaged friction factor can be
£ ro + Lpp~ri lapp = ~pp,o
(5.180)
ro + ri Thermally Developing Flow. Kays and Leung [111] present experimental results for thermally developing turbulent flow in four concentric annular ducts, r* = 0.192, 0.255, 0.376, and 0.500, with the boundary condition of one wall at uniform heat flux and the other insulated, that is, the fundamental solution of the second kind. In accordance with this solution, the local Nusselt numbers Nu~,o and Nux,i at the outer and inner walls are expressed as
Nu~o= 1 '
Nux,oo "* . . . .
 Vx, oqi I q o '
1
Nuxi= ' 1  Ox,i i  Ox,mi
(5.181)
where q~ and q7 are the uniform heat fluxes at the outer and inner walls. Both q~ and q7 are positive whenever heat is added to the fluid and negative whenever heat is transferred out of
5.57
FORCED CONVECTION, INTERNAL FLOW IN DUCTS
the fluid. The Nusselt numbers Nux,oo and Nux, ii and the influence coefficients 0*x,oand 0~i are given by: 1 Nux, oo = 0 ....  0 . . . .
Ox~,o
1
O x ' m i  Ox'm° = 0 . . . .  0 . . . .
Ox'm°  Ox'i° O x, ii __ O x, m i
(5.183)
4r*(x/Dh) Ox,mi = Re Pr (1 + r*)
(5.184)
Ox~,,i =
4(x/Dh) and
(5.182)
N u x , ii ~ Ox, ii _ Ox, m i
0 . . . . = Re Pr (1 + r*)
The nondimensional temperatures Ox,oo, Ox,,, Ox, oi, and Ox,io for r* = 0.192 and 0.5 are presented in Fig. 5.18 as an example. Additional graphical results for r* = 0.192, 0.255, and 0.376 are available in Kays and Leung [111].
0.05
0.05
0.04
0.04
Oz,°° 0.03
#x,// 0.03
0.02
0.02
= 11,010
0.01 0
Re= I0,80~70_~ ..i ro ~ . ~
O=,io 0 . 0 1
O0
30:420..~\~ ~
'
20
40 x/D h
60
1
0
J O=.ai 0.01
80
f
0(~
Re= 11,010 15 240 221410,,\\ 3 0 , 8 41, I
I
20
,
40
r
ri /
~
I
60
'
•
'i .j
'
80
x/D h
FIGURE 5.18 0.... Ox,io, Ox,ii, Ox.oifor use with Eqs 5.182 and 5.183 for thermally developing flow in a smooth concentric annular duct with r* = 0.5 and Pr = 0.7 [111].
The preceding solution is restricted to a fluid with Pr = 0.7, 104 < Re < 1.61 x 105, and 0.192 < r* < 0.5. Cross plotting and interpolation can be employed to increase the application range of the results in terms of Re and r*. For Pr = 0.01 and Pr = 1000, an eigenvalue solution to the fundamental problem of the second kind for four concentric annular ducts (r* = 0.02, 0.1067, 0.1778, and 0.3422) can be found in Q u a r m b y and A n a n d [117]. Developing Flow. Little information is available on simultaneously developing turbulent flow in concentric annular ducts. However, the theoretical and experimental studies by Roberts and Barrow [118] indicate that the Nusselt numbers for simultaneously developing flow are not significantly different from those for thermally developing flow.
Simultaneously
5.58
CHAPTERFIVE
Effects of Eccentricity.
Jonsson and Sparrow [119] have conducted a careful experimental investigation of fully developed turbulent flow in smooth, eccentric annular ducts. The researchers have provided the velocity measurements graphically in terms of the wall coordinate u ÷ as well as the velocitydefect representation. From their results, the circumferentially averaged fully developed friction factor is correlated by a powerlaw relationship of the following type: C f = Re n
(5.185)
where C is a strong function of e*, a relatively weak function of r*, and independent of the Reynolds number, which is given in Fig. 5.19. A single value, n = 0.18, has been suggested by Jonsson and Sparrow [119] for all r*, e*, and Re. More details regarding the friction factors j~ and fo for each of the two surfaces are also available [120]. Other investigations of fully developed turbulent flow in eccentric annular ducts have been conducted by Lee and Barrow [121], Deissler and Tayler [122], Yu and Dwyer [123], and Ricker et al. [124]. 0.20
r*=0.281
0.16 0.14 0.12 C
0.10 O.08 0.06
m
0.02
r
•
~

0 0
I 0.1
I 0.2
1 0.3
I 0.4
1 0.5
I 0.6
_J 0.7
I 0.8
I 0.9
1.0
e*
FIGURE 5.19 Empiricalconstant C in Eq. 5.185 [119]. The effects of the eccentricity on turbulent heat transfer in eccentric annular ducts have been investigated by Judd and Wade [125], Leung et al. [126], Lee and Barrow [121], and Yu and Dwyer [123] for the boundary condition of a uniform wall heat flux on the inner or outer surfaces while the other wall is insulated. The results were obtained under specific conditions. Further details can be found in the previously mentioned references. Few investigations have been conducted on hydrodynamically developing flow in eccentric annular ducts. Jonsson [120] has obtained experimental information on the pressure gradient in hydrodynamically developing flow and provided the hydrodynamic lengths Lhy/Dh for 1.8 x 104 < Re < 1.8 × 105. These are presented in Table 5.29.
FORCED CONVECTION, INTERNAL FLOW IN DUCTS
5.59
TABLE 5.29 Turbulent Flow Hydrodynamic Entrance Lengths for Smooth, Eccentric Annular Ducts [120]
Lhy/Dh r*
e* = 0
0.5
0.9
1.0
0.281 0.561 0.750
29 26 28
32 38 50
38 59 69
38 78 91
Few results that can be used in practice are available for thermally developing flow and simultaneously developing flow in eccentric annuli. According to the discussion in Bhatti and Shah [45], the Nusselt numbers may be estimated from the corresponding results for concentric annuli (e* = 0).
PARALLEL PLATE DUCTS Parallel plate ducts, also referred to as flat ducts or parallel plates, possess the simplest duct geometry. This is also the limiting geometry for the family of rectangular ducts and concentric annular ducts. For most cases, the friction factor and Nusselt number for parallel plate ducts are the maximum values for the friction factor and the Nusselt number for rectangular ducts and concentric annular ducts.
Laminar Flow
Laminar flow and heat transfer in parallel plate ducts are described in this section. The friction factor and Nusselt number are given for practical calculations.
Fully Developed Flow.
For a parallel plate duct with hydraulic diameter D h = 4b (b being the halfdistance between the plates) and the origin at the duct axis, the velocity distribution and friction factor are given by the following expression: u _ 3 1
Um Um = 
(5.186)
2
l( x)
~
b 2,
fRe
= 24
(5.187)
Similar to the four fundamental thermal boundary conditions for concentric annuli, the four kinds of fundamental conditions for parallel plate ducts are shown in Fig. 5.20. The fully developed Nusselt numbers for the four boundary conditions follow [1]: First kind:
NUl = Nu2 = 4
(5.188)
Second kind:
Nul = 0
Nu2 = 5.385
(5.189)
Third kind:
Nul = 0
Nu2 = 4.861
(5.190)
Fourth kind:
NUl = Nu2 = 4
(5.191)
5.60
CHAPTER FIVE
z=O
z=O I
I I I I
=z
a
I I
f"
I I I Seoond kind
Plrst kind ==0
z=O
:
Wall 1
I I
0
I '
I
\,
I I
fT.
I
I I I
I I I Third kind
F I G U R E 5.20
F o u r t h kind
Four fundamental boundary conditions for a parallel plate duct [2].
Examples of the application of these fundamental solutions to obtain the fully developed Nusselt number for a duct with three different boundary conditions follow. The Nusselt numbers are defined as
q"wjOh
(5.192)
Nuj = k ( T j  Tm) where ] denotes wall I or 2, and Tj is the temperature of the jth wall. Uniform Temperatureat Each Wall. When the temperatures on two walls are equal, Twl = /'wE, then NUl = Nu2 = NUT. The value of NUT is given by Shah and London [1] as follows: NUT = 7.541
(5.193)
When the temperatures on two walls are different, Twl ~: Tw2 , then NUl = Nu2 = 4, as shown in Eq. 5.188. When the effect of viscous dissipation is considered, the following formulas developed by Cheng and Wu [127] for the case T~I > Tw2are used to compute the Nusselt numbers: 4(1  6ar) Nul  1  48/35Br
4(1 + 6ar)
(5.194)
Nu2 = 1 + 48/35Br
When the effects of viscous dissipation and flow work are considered together for the case of Twl = Tw2, Ou and Cheng [128] have shown that for fully developed flow, NuT = 0 and the dimensionless temperature distribution is as follows:
T~T
9
[
(y~2]~
T.~e = 8 Br 1  \ b ] J '
TwTm
27 T~~  35 Br
(5.195/
FORCED CONVECTION, INTERNAL FLOW IN DUCTS
5.61
Taking the fluid axial condition into account, Pahor and Strand [129] and Grosjean et al. [130] have obtained the following asymptotic formulas for the Nusselt number in the case of
Twl = Tw2: NuT =
7.540(1 + 3.79/Pe 2 +..)
for Pe >> 1
8.11742(1 0.030859Pe + 0.0069436Pe 2 . . . . )
for Pe 0.001 for x* < 0.001 for 0.001 < x* < 0.01 for x* > 0.01
(5.212)
(5.213)
It has been concluded that except in the neighborhood of the duct inlet (x* < 102), the effect of the fluid axial conduction is negligible for Pe > 50 [134, 135]. Convective Boundary Condition at Both Walls or One Wall The solutions for the convective boundary condition on both walls or one wall are reviewed in Shah and London [1], where more detailed descriptions are available.
Simultaneously Developing Flow.
The results for simultaneously developing flow in parallel plate ducts are provided for the following thermal boundary conditions. Equal and Uniform Temperatures at Both Walls. For simultaneously developing flow in a parallel plate duct with fluids of 0.1 < Pr < 1000, the following equations are recommended for the computation of the local and mean Nusselt numbers [2, 136, 137]: NUx,T = 7.55 +
0.024x*114[0.0179Pr °'17x *°'64  0.14] [1 + 0.0358Pr °17 x*°64]z
(5.214)
0 . 0 2 4 X *l J4 NUm,T =
7.55 + 1 +
0 . 0 3 5 8 P r °'17 x *°'64
(5.215)
The thermal entrance length L*th,T has been found to be 0.0064 for 0.01 < Pr < 10,000 [138, 139]. A detailed description can be found in Shah and London [1] for the solutions for Pr = oo and Pr = 0. When one duct wall is insulated and the other is at a uniform temperature, the local and mean Nusselt numbers for simultaneously developing flow have been obtained for fluids of 0.1 < Pr < 10. These follow [1]: NUx,T = 4.86 +
0.0606x*12[0.0455Pr °ATx *°'7  0.2] [1 + 0.0909Pr °17 x*°7] 2
(5.216)
0.0606x *1.2
NUm,T = 4.86 + 1 +
0 . 0 9 0 9 P r °'17 x *°'7
(5.217)
Uniform and Equal Heat Flux at Both Walls. The local Nusselt number for heat transfer of laminar flow in a parallel plate duct with uniform and equal heat flux at both walls is displayed in Fig. 5.22 for different Prandtl numbers, Pr = 0 [34] and Pr = 0.01, 0.7, 1, 10, and oo [136]. The thermal entrance lengths obtained from the results presented in this figure are 0.016, 0.030, 0.017, 0.014, 0.012, and 0.0115, for Pr = 0, 0.01, 0.7, 1, 10, and 0% respectively.
5.64
CHAPTER FIVE
25 Pr=0
,
]
20 2b
Nu=,H 15
12
10
8.2353
SLI
sxlo
103
2
4
6
8 102
2
5 X 102
X* FIGURE 5.22 Local and mean Nusselt numbers for simultaneously developing flow in a parallel plate duct with the @ boundary condition [34, 136]. The local Nusselt n u m b e r is displayed in Fig. 5.23 for Pr = 0.0, 0.01, 0.7, 10, and o~ when one wall of the parallel plate duct is insulated and the other wall is subjected to uniform heat flux heating [140]. Included in Fig. 5.23 are the results for Pr = 0% obtained from the concentric annular duct corresponding to r* = 1. The local and m e a n Nusselt n u m b e r s for Pr = 0 were obtained by Bhatti [34]. In addition, N g u y e n [141] has obtained the apparent friction factor and Nusselt n u m b e r s for low Reynolds n u m b e r simultaneously developing flow in a parallel plate duct with a con35
30
PrO 25
0.01
2O
2b
"""ffff/'/" T
0.7
Nux, H
ttttttt
10
15
6.00
10
5.39
51 104
103
102
10t
100
X* FIGURE 5.23 Local and mean Nusselt numbers for simultaneously developing flow in a fiat duct with uniform heat flux at one wall and the other wall insulated [34, 140].
FORCED CONVECTION, INTERNAL FLOW IN DUCTS
5.65
stant temperature and constant wall heat flux thermal boundary conditions. Results are presented for Pr = 0.7 and Reynolds numbers between 1 and 20. The results for Pr = 0.2, 0.7, 2, 7, 10, and 100 as well as for the Reynolds numbers between 40 and 200 have been numerically determined by Nguyen and MaclaineCross [142].
Turbulent Flow The characteristics of turbulent flow and heat transfer in parallel plate ducts are discussed in this section. The friction factor for transition flow is also addressed.
Transition Flow. The lower limit of the critical Reynolds number Rec,it for a parallel plate duct is reported to be between 2200 and 3400, depending on the entrance configurations and disturbance sources [143]. The following friction factor formula developed by Hrycak and Andrushkiw [144] is recommended for transition flow in the range of 2200 < Re < 4000: f =  2 . 5 6 x 103
+
4.085 x 106 Re  5.5 x 1010 Re 2
(5.218)
The mean Nusselt number in the thermal entrance region of a parallel plate duct with uniform wall temperature at both walls in the range of 2300 < Re < 6000 is given by Hausen [20] as follows:
[
(
NUm,T= 0.116(Re a3 160t Pr '/3 1 + \D7]
J
(5.219)
Fully Developed Flow.
Beavers et al. [145] obtained the following friction factor for fully developed turbulent flow in a parallel plate duct for 5000 < Re < 1.2 x 10 6 from very accurate experimental data: f=
0.1268 Re0.3
(5.220)
For 1.2 x 104 < Re < 1.2 x 10 6, Dean [146] has developed the following equation based on a comprehensive survey of the available data: 0.0868 f=
Re1/4
(5.221)
Comparisons of precision using Eqs. 5.220 and 5.221 and Blasius's formula (Table 5.8) in which the diameter of circular duct 2a is replaced by hydraulic diameter 4b, b being the halfspace between two plates, have been conducted by Bhatti and Shah [45]. In the range of 5000 < Re < 3 x 10 4, Eq. 5.220 is recommended; otherwise, Eq. 5.221 should be used to obtain the friction factor for fully developed turbulent flow in a parallel plate duct. However, use of the hydraulic diameter to substitute for the circular duct diameter in the Blasius equation is reasonable for the prediction of the fraction factor [45]. Kays and Leung [111] analyzed turbulent heat transfer in a parallel flat plate duct for arbitrarily prescribed heat flux on the two duct walls. The fully developed Nusselt number Null can be obtained from the following expression: Nu
Null  1  ~,e*
(5.222)
where ), is the ratio of the prescribed heat fluxes on the two duct walls. The Nusselt number Nu and the influence coefficient 0* in Eq. 5.222 are given in Table 5.28 for different Re and Pr numbers. It should be noted that ~,= 0 signifies that one wall is heated and the other is insulated; ? = 1 indicates that uniform heat fluxes of equal magnitudes are applied to both walls;
5.66
CHAPTER FIVE
and 7 = 1 refers to heat transfer into one wall and out of the other wall, while the absolute values of the heat fluxes at both walls are the same. Bhatti and Shah [45] and Sparrow and Lin [133] have performed a comparison of Nusselt numbers predicted using Eq. 5.222 or other equations for parallel plate ducts and the Nusselt number calculated using the equation for circular ducts replacing 2a with the hydraulic diameter of the parallel plate duct. It was concluded that the Nusselt number for parallel plate ducts can be determined using the circular duct correlations. Analogous to circular ducts, the fully developed turbulent Nusselt numbers for uniform wall temperature and uniform wall heat flux boundary conditions in parallel plate ducts are nearly identical for Pr > 0.7 and Re > 105. This is also true for the Nusselt number of turbulent thermally developing flow in a parallel plate duct [147]. For liquid metal, when one wall of the parallel plate duct is heated and the other is adiabatic, the following empirical equation is recommended for Pr < 0.03 by Duchatelle and Vautrey [148]:
(5.223)
NUll = 5.14 + 0.0127 Pe °'8
Fully developed fluid flow and heat transfer results for rough parallel plate ducts can be predicted using the results for rough circular ducts with the use of hydraulic diameter [45].
Hydrodynamically Developing Flow. Hydrodynamically developing flow in smooth parallel plate ducts with uniform velocity at the duct inlet has been analyzed by Deissler [92] by means of an integral method. The apparent friction factors fapp in the hydrodynamic entrance are presented in Fig. 5.24.
0.020 t t t 0.015
ll\ f.~
=1o,
I•I . ~ ~ ~ x
I I
3×104 I
5
o.o~o
0.005
0.000
0
4
8
12
16
20
24
~Da FIGURE 5.24 Turbulent flow apparent friction factors in the hydrodynamic entrance region of a parallel plate duct with uniform inlet velocity [45].
FORCED CONVECTION,INTERNALFLOW IN DUCTS
5.67
Thermally Developing and Simultaneously Developing Flow. Thermally developing turbulent flow in a parallel plate duct with uniform and equal temperatures at both walls has been solved by Sakakibara and Endo [149] and by Shibani and Ozisik [150]. A discussion of the solution can be found in Bhatti and Shah [45]. Hatton and Quarmby [151] and Sakakibara and Endo [149] have obtained the solution for a thermally developing turbulent flow problem in a flat duct with one wall at uniform temperature and the outer wall insulated (i.e., the fundamental solution of the third kind). Hatton and Quarmby [151], Hatton et al. [152], and Sakakibara [153] have analyzed thermally developing turbulent flow in a flat duct with uniform heat flux at one wall and the other insulated (i.e., the fundamental solution of the second kind). Ozisik et al. [154] have obtained the solution of the thermal entry region heat transfer of turbulent developing flow in a parallel plate duct with uniform wall temperature. Detailed discussions of these solutions can be found in the previously mentioned references. Few investigations have been conducted for simultaneously developing flow in parallel plate ducts. Therefore, no correlations are provided for practical usage. RECTANGULAR DUCTS Rectangular ducts are also often used in the design of heat transfer devices such as compact heat exchangers. Unlike circular and parallel plate ducts, twodimensional analysis is required to obtain the friction factors and Nusselt numbers for rectangular ducts.
Laminar Flow In this section, the friction factors and Nusselt numbers for fully developed, hydrodynamically developing, thermally developing, and simultaneously developing laminar flows in rectangular ducts are presented.
Fully Developed Flow Velocity Distribution and the Friction Factor. Marco and Han [155] have obtained the fully developed velocity distribution in a rectangular duct with crosssectional dimensions 2a and 2b. It follows:
16(dp)a2 u =  ~ ~ where the pressure gradient
~l3 (1)(n1)/2(cosh(n~y/2a))
~ ......
n3
1  cosh
(n~z I (n~b/2a) cos \  ~ a ]
(5.224)
dp/dx is related to Umas follows:
Urn=~ ~
~ [1~
/ ......
~ t a n h \[ nrcb 2a l] ]]
(5.225)
The origin of the Cartesian coordinate is at the center of the rectangular duct. To avoid computational complexity, the following simple approximations have been suggested [156]:
u Umax [1 y " (z)ml Umaxum(mX)(nl)m n Natarajan and Lakshmanan [157] have provided the relation for the values of rn and m = 1.7 + 0.50[*1"4
5226, 5227, n: (5.228)
5.68
CHAPTERFIVE
n = where ~* =
[~ + 0.3(~*  V~)
for a* < 1,6 for ~* > 1/3
(5.229)
b/a is the aspect ratio. The exact expression for the fully developed friction factor is 24 1 +
1
~1~,50~, . . .1,3, ..
n5
However, for ease in practical calculations, the following empirical equation suggested by Shah and L o n d o n [1] is used to a p p r o x i m a t e Eq. 5.230: f R e = 24(1  1.3553~* + 1.9467~ .2  1.7012~ .3 + 0.9564~ .4  0.2537~ .5)
(5.231)
Heat Transfer The fully d e v e l o p e d Nusselt n u m b e r s NUT for the case of the uniform temp e r a t u r e at four walls are a p p r o x i m a t e d by the following formula [1]: NUT = 7.541(1  2.610~* + 4.970~ .2  5.119~ .3 + 2.702~ .4 + 0.548C~.5)
(5.232)
For rectangular ducts with uniform t e m p e r a t u r e at one or m o r e walls, the Nusselt numbers, available in Shah and L o n d o n [1], are displayed in Table 5.30. For rectangular ducts with a uniform wall heat flux at four walls u n d e r the ~ b o u n d a r y condition, the fully d e v e l o p e d Nusselt n u m b e r s N u m can be c o m p u t e d with the following formula [1]:
NUll1 = 8.235(1  2.0421tx* + 3.0853tx .2  2.4765~t .3 + 1.0578t~ .4  0.18610~ .5)
(5.233)
For rectangular ducts with one or m o r e walls subjected to the t~ b o u n d a r y condition with the other wall insulated, the fully d e v e l oped Nusselt numbers NUll1 are displayed in Table 5.30.
TABLE 5.30 Nusselt Number for Fully Developed Laminar Flow in Rectangular Ducts With One Wall or More Walls Heating
_~
I,~2a*l
1//////
0~*
NUT
NUll1
NUT
NRH1
NUT
Num
NUT
NUll1
NUT
NUll1
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 2.0 2.5 5.0 10.0 o,
7.541 5.858 4.803 4.114 3.670 3.383 3.198 3.083 3.014 2.980 2.970 3.383 3.670 4.803 5.858 7.541
8.235 6.785 5.738 4.990 4.472 4.123 3.895 3.750 3.664 3.620 3.608 4.123 4.472 5.738 6.785 8.235
7.541 6.095 5.195 4.579 4.153 3.842 . 3.408 . . 3.018 2.602 2.603 2.982 3.590 4.861
8.235 6.939 6.072 5.393 4.885 4.505
7.541 6.399 5.703 5.224 4.884 4.619 . 4.192 . . 3.703 2.657 2.333 1.467 0.843 0
8.235 7.248 6.561 5.997 5.555 5.203 . 4.662 . . 4.094 2.947 2.598 1.664 0.975 0
0 0.457 0.833 1.148 1.416 1.647
0 0.538 0.964 1.312 1.604 1.854
2.023
2.263
2.437 3.185 3.390 3.909 4.270 4.861
2.712 3.539 3.777 4.411 4.851 5.385
4.861 3.823 3.330 2.996 2.768 2.613 2.509 2.442 2.401 2.381 2.375 2.613 2.768 3.330 3.823 4.861
5.385 4.410 3.914 3.538 3.279 3.104 2.987 2.911 2.866 2.843 2.836 2.911 3.279 3.914 4.410 5.385
.
. 3.991
. .
. . 3.556 3.146 3.169 3.636 4.252 5.385
. . .
i
FORCED CONVECTION, INTERNAL FLOW IN DUCTS
5.69
TABLE 5.31 Fully Developed fRe, NUT,NUll1,and NUll2for Laminar Flow in Rectangular Ducts With All
Four Walls Transferring Heat [1]
ix*
fRe
NUT
NUll1
NUll2
ct*
fRe
NUT
Num
1.000 0.900 1/1.2 0.800 0.750 1/1.4 0.700 ~6 0.600 0.500 0.400 1,6 0.300
14.227 14.261 14.328 14.378 14.476 14.565 14.605 14.701 14.980 15.548 16.368 17.090 17.512
2.970 2.980 ~ 3.014 ~ 3.077 3.083 3.117 3.198 3.383 3.670 3.956 4.114
3.60795 3.62045 3.64531 3.66382 3.70052 3.73419 3.74961 3.79033 3.89456 4.12330 4.47185 4.79480 4.98989
3.091
0.250 0.200
18.233 19.071 19.702 20.193 20.585 20.904 21.169 21.583 22.019 22.477 23.363 24.000
4.439 4.803 5.137 5.597 5.858 7.541
5.33106 5.73769 6.04946 6.29404 6.49033 6.65107 6.78495 6.99507 7.21683 7.45083 7.90589 8.23529
3.02 2.97
1//7 0.125 1/'9 1/10 1/12 1/15 1/20 1/50 0
NUH2 2.94 2.93 2.93 2.94 2.94 2.94 2.95
8.235
For rectangular ducts with four walls h e a t e d u n d e r the @ thermal b o u n d a r y condition, the fully d e v e l o p e d Null2 can be d e t e r m i n e d from Table 5.31 [1]. The f R e , NUT, and NUH] are also given in Table 5.31 for convenience of usage.
Hydrodynamically Developing Flow.
Shah and L o n d o n [1] have reviewed and c o m p a r e d several analytical and e x p e r i m e n t a l investigations of hydrodynamically developing flow in rectangular ducts. They concluded that the numerical results r e p o r t e d by Curr et al. [158] and the analytical results r e p o r t e d by Tachibana and I e m o t o [159] best fit the experi ment al data. The a p p a r e n t friction factors obtained by Curr et al. [158] are shown in Fig. 5.25. 80
i
70
2b
a* = 2b
60
 2o
t
50 lapp Re 0 02. 0.5 1
40
30
15 0.002
0.005
0.01
0.02
0.05
0.1
x+ FIGURE 5.25 Apparent Fanning friction factors for hydrodynamically developing flow in rectangular ducts [158].
5.70
CHAPTER FIVE T h e r m a l l y D e v e l o p i n g Flow. Wibulswas [160] and A p a r e c i d o and Cotta [161] have solved the thermal entrance p r o b l e m for rectangular ducts with the thermal b o u n d a r y condition of uniform t e m p e r a t u r e and uniform heat flux at four walls. However, the effects of viscous dissipation, fluid axial conduction, and thermal energy sources in the fluid are neglected in their analyses. The local and m e a n Nusselt n u m b e r s NUx,T, NUm,T, and Nux, m and NUm,H1 obtained by Wibulswas [160] are p r e s e n t e d in Tables 5.32 and 5.33. For square ducts, a* = 1, the Nusselt n u m b e r s for the ~ , ~ , and @ thermal b o u n d a r y conditions have b e e n obtained by C h a n d r u p a t l a and Sastri [162]. As r e c o m m e n d e d by Shah and L o n d o n [1], the results obtained by C h a n d r u p a t l a and Sastri [162], shown in Table 5.34, are m o r e accurate than those p r e s e n t e d by Wibulswas [160]. The thermal entrance lengths for rectangular ducts in the 0) b o u n d a r y condition L *th,T are d e t e r m i n e d to be 0.008, 0.054, 0.049, and 0.041 for a*  0 , 0.25, 0.5, and 1, respectively [1]. The thermal entrance lengths in the ~ b o u n d a r y condition L *th,nl are found to be 0.0115, 0.042,
TABLE 5.32 Local and Mean Nusselt Numbers in the Thermal Entrance Region of Rectangular Ducts With the 03 Boundary Condition [160] 1
NUm,T
NUx,T
x*
a* = 1.0
0.5
1/3
0.25
0.2
V6
1.0
0.5
1/3
0.25
0.2
0 10 20 30 40 60 80 100 120 140 160 180 200
2.65 2.86 3.08 3.24 3.43 3.78 4.10 4.35 4.62 4.85 5.03 5.24 5.41
3.39 3.43 3.54 3.70 3.85 4.16 4.46 4.72 4.92 5.15 5.34 5.54 5.72
3.96 4.02 4.17 4.29 4.42 4.67 4.94 5.17 5.42 5.62 5.80 5.99 6.17
4.51 4.53 4.65 4.76 4.87 5.08 5.32 5.55 5.77 5.98 6.18 6.37 6.57
4.92 4.94 5.05 5.13 5.22 5.40 5.62 5.83 6.06 6.26 6.45 6.63 6.80
5.22 5.25 5.34 5.41 5.48 5.64 5.86 6.07 6.27 6.47 6.66 6.86 7.02
2.65 3.50 4.03 4.47 4.85 5.50 6.03 6.46 6.86 7.22 7.56 7.87 8.15
3.39 3.95 4.46 4.86 5.24 5.85 6.37 6.84 7.24 7.62 7.97 8.29 8.58
3.96 4.54 5.00 5.39 5.74 6.35 6.89 7.33 7.74 8.11 8.45 8.77 9.07
4.51 5.00 5.44 5.81 6.16 6.73 7.24 7.71 8.13 8.50 8.86 9.17 9.47
4.82 5.36 5.77 6.13 6.45 7.03 7.53 7.99 8.39 8.77 9.14 9.46 9.79
5.22 5.66 6.04 6.37 6.70 7.26 7.77 8.17 8.63 9.00 9.35 9.67 10.01
TABLE 5.33 Local and Mean Nusselt Numbers in the Thermal Entrance Region of Rectangular Ducts With the ~ Boundary Condition [160] 1
x* 0 10 20 30 40 60 80 100 120 140 160 180 200
NUx, H1
a* = 1.0 3.60 3.71 3.90 4.18 4.45 4.91 5.33 5.69 6.02 6.32 6.60 6.86 7.10
NUm, H1
0.5
½
0.25
1.0
0.5
½
0.25
4.11 4.22 4.38 4.61 4.84 5.28 5.70 6.05 6.37 6.68 6.96 7.23 7.46
4.77 4.85 5.00 5.17 5.39 5.82 6.21 6.58 6.92 7.22 7.50 7.76 8.02
5.35 5.45 5.62 5.77 5.87 6.26 6.63 7.00 7.32 7.63 7.92 8.18 8.44
3.60 4.48 5.19 5.76 6.24 7.02 7.66 8.22 8.69 9.09 9.50 9.85 10.18
4.11 4.94 5.60 6.16 6.64 7.45 8.10 8.66 9.13 9.57 9.96 10.31 10.64
4.77 5.45 6.06 6.60 7.09 7.85 8.48 9.02 9.52 9.93 10.31 10.67 10.97
5.35 6.03 6.57 7.07 7.51 8.25 8.87 9.39 9.83 10.24 10.61 10.92 11.23,,,
FORCED CONVECTION, INTERNAL FLOW IN DUCTS
5.71
TABLE 5.34 Local and Mean Nusselt Numbers in the Thermal Entrance Region of a Square Duct With the 03, (~, and @ Boundary Conditions [162] 1 X*
NUx,T
NUm,T
NRx,H1
NUm,H1
NUx,H2
NRm,H2
0 10 20 25 40 50 80 100 133.3 160 200
2.975 2.976 3.074 3.157 3.432 3.611 4.084 4.357 4.755 5.412
2.975 3.514 4.024 4.253 4.841 5.173 5.989 6.435 7.068 8.084
3.612 3.686 3.907 4.048 4.465 4.720 5.387 5.769 6.331 6.730 7.269
3.612 4.549 5.301 5.633 6.476 6.949 8.111 8.747 9.653 10.279 11.103
3.095 3.160 3.359 3.471 3.846 4.067 4.654 4.993 5.492 5.848 6.330
3.095 3.915 4.602 4.898 5.656 6.083 7.138 7.719 8.551 9.128 9.891
0.048, 0.057, and 0.066 for o~* = 0, 0.25, 1/3, 0.5, and 1, respectively [1]. T h e r m a l l y developing flow in rectangular ducts with one wall or m o r e insulated is reviewed in Shah and L o n d o n [1].
Simultaneously Developing Flow.
Table 5.35 presents the results for simultaneously developing flow in rectangular ducts; these w e r e obtained by Wibulswas [160] for the 03 and (~ b o u n d a r y conditions for air (Pr = 0.72). Transverse velocity is neglected in this analysis. H o w ever, C h a n d r u p a t l a and Sastri [163] include transverse velocity in their analysis for a square duct with the ~ b o u n d a r y condition. The NUx,H1 and NUm,H1 obtained by C h a n d r u p a t l a and Sastri [163] are illustrated in Table 5.36. It should be n o t e d that in Table 5.36, Pr = 0 corresponds to slug flow, w h e r e a s Pr = oo corresponds to hydrodynamically d e v e l o p e d flow.
TABLE 5.35 Local and Mean Nusselt Numbers for Simultaneously Developing Flow in Rectangular Ducts With the 03 and ~ Boundary Conditions [160] 1
Nux,m
x*
0~* = 1.0
5 10 20 30 40 50 60 80 100 120 140 160 180 200 220
. 4.18 4.66 5.07 5.47 5.83 6.14 6.80 7.38 7.90 8.38 8.84 9.28 9.69 .
.
.
NUrn,H1
0.5
1A
0.25
1.0
. 4.60 5.05 5.40 5.75 6.09 6.42 7.02 7.59 8.11 8.61 9.05 9.47 9.88 .
. 5.18 5.50 5.82 6.13 6.44 6.74 7.32 7.86 8.37 8.84 9.38 9.70 10.06 .
5.66 5.92 6.17 6.43 6.70 7.00 7.55 8.08 8.58 9.05 9.59 9.87 10.24
4.60 5.43 6.60 7.52 8.25 8.90 9.49 10.53 11.43 12.19 12.87 13.50 14.05 14.55 15.03
0.5 5.00 5.77 6.94 7.83 8.54 9.17 9.77 10.73 11.70 12.48 13.15 13.79 14.35 14.88 15.36
NUm,T
½
0.25
1.0
5.58 6.27 7.31 8.13 8.85 9.48 10.07 11.13 12.00 12.78 13.47 14.10 14.70 15.21 15.83
6.06 6.65 7.58 8.37 9.07 9.70 10.32 11.35 12.23 13.03 13.73 14.48 14.95 15.49 16.02
. 3.75 4.39 4.88 5.28 5.63 5.95 6.57 7.10 7.61 8.06 8.50 8.91 9.30 9.70
0.5 .
½ .
4.20 4.79 5.23 5.61 5.95 6.27 6.88 7.42 7.91 8.37 8.80 9.20 9.60 10.00
. 4.67 5.17 5.60 5.96 6.28 6.60 7.17 7.70 8.18 8.66 9.10 9.50 9.91 10.30
0.25 . 5.11 5.56 5.93 6.27 6.61 6.90 7.47 7.98 8.48 8.93 9.36 9.77 10.18 10.58
5.72 6.13 6.47 6.78 7.07 7.35 7.90 8.38 8.85 9.28 9.72 10.12 10.51 10.90
5.72
CHAPTERFIVE
Local and Mean Nusselt Numbers for Simultaneously Developing Flow in a Square Duct (ix* = 1) With the ~ Boundary Condition [163]
TABLE 5.36
NUx,H1
Num,rtl
x*
Pr = 0.0
0.1
1.0
10.0
~
0.0
0.005 0.0075 0.01 0.0125 0.02 0.025 0.04 0.05 0.1 **
14.653 12.545 11.297 10.459 9.031 8.500 7.675 7.415 7.051 7.013
11.659 9.597 8.391 7.615 6.353 5.883 5.108 4.826 4.243 3.612
8.373 7.122 6.379 5.877 5.011 4.683 4.152 3.973 3.687 3.612
7.329 6.381 5.716 5.480 4.759 4.502 4.080 3.939 3.686 3.612
7.269 6.331 5.769 5.387 4.720 4.465 4.048 3.907 3.686 3.612
0.1
21.986 19.095 17.290 16.003 13.622 12.647 10.913 10.237 8.701 7.013
17.823 15.391 13.781 12.620 10.475 9.601 8.043 7.426 5.948 3.612
1.0 13.390 11.489 10.297 9.461 7.934 7.315 6.214 5.782 4.783 3.612
10.0 11.200 9.737 8.823 8.181 7.010 6.533 5.682 5.347 4.580 3.612
11.103 9.653 8.747 8.111 6.949 6.476 5.633 5.301 4.549 3.612
Turbulent Flow Entrance configuration is the key factor affecting flow transition in rectangular ducts. The lower limit of the critical Reynolds numbers Rent along with entrance configuration has been investigated by Davies and White [164], Allen and Grunberg [165], Cornish [166], and Hartnett et al. [167]. The lower limits of the critical Reynolds numbers for a smooth rectangular duct with two entrance configurations are given in Table 5.37. For most engineering calculations of friction factors and Nusselt numbers for fully developed flow in rectangular ducts, it is sufficiently accurate to use the circular duct correlations by replacing the circular duct diameter 2a with the hydraulic diameter Dh = 4ab/(a + b) or with Dr, defined by the following equations to consider the shape effect [168]: 2
(
Dr= ~ Dh(1 + 0~*)2 1 
192tx*~o 1 (2n + 1 ) u c t * ) ( 5 . 2 3 4 ) n~ .= '~ (2n + 1) tanh 2
An approximate expression for Dt is: 2 Dh + "11 Dt = "~ ~ o~*(2  tx*)
(5.235)
which yields D~ values within +_2 percent of those given by Eq. 5.234. TABLE 5.37
Lower Limits of the Critical Reynolds Numbers for Smooth Rectangular Ducts
Entrance configurations Smooth entrance
Schematics t ................ _,. g
Abrupt entrance
l
m

~ " . . . . . . . . "~"
Aspect ratio ct*
Critical Reynolds number Recnt
0 0.1 0.2 0.3333 1.0
3400 4400 7000 6000 4300
0 0.1 0.2 0.2555 0.3425 1.0
3100 2920 2500 2400 2360 2200
FORCED CONVECTION,INTERNALFLOW IN DUCTS
5.73
The fully developed friction factor and heat transfer coefficients for turbulent flow in an asymmetrically heated rectangular duct have been reported by Rao [59]. In this investigation, the experimental region of the Reynolds number was from 104 to 5 x 104. For fully developed Nusselt numbers for the turbulent flow of liquid metals in rectangular ducts, a simple correlation has been derived for the 03 and ~ boundary conditions [169]. This correlation follows: Nu = ~Nus~ug + 0.015Pe °8
(5.236)
where Nus~ugis the Nusselt number corresponding to slug flow (Pr = 0) through rectangular ducts, which is given in Fig. 5.26 as a function of or* for rectangular ducts under the 03 and (~ boundary conditions.
14
'
I
I
I
'
2 10
Nuaut
I\
4 0
FIGURE [169].
\
~
/ T ~
boundary condition
L boundarycondi ( ~ tion~ 0.25
0.50
0.75
t.O0
5.26Slug flow Nusselt numbers for rectangular ducts
TRIANGULAR DUCTS The flow and heat transfer characteristics of triangular ducts, as shown in Fig. 5.27, are explained in this section. The coordinates shown in Fig. 5.27 are used in the presentation of the results.
Laminar
Flow
In this section, the laminar flow and heat transfer characteristics are explained for different triangular ducts.
5.74
CHAPTERFIVE
I
2
I
Lz
z
2a
), , f
(a)
,
2b
(b)
(c)
(e)
(0
I
2b
l (d)
FIGURE 5.27 Triangularducts: (a) equilateral; (b) equilateral with rounded comers; (c) isosceles; (d) and (e) right; and (f) arbitrary.
Fully Developed Flow Equilateral Triangular Ducts. For equilateral triangle ducts as shown in Fig. 5.27a, the fully developed laminar flow velocity profile and friction factor have been obtained by Marco and Han [155]:
U15Um _
y
z 2
8 [ ( b ) 3 3(~)(~) 2(b) 2 2(b)2~  ~~]
Um= 1"~
f R e = .40. = 13.333
(5.237)
(5.238)
When the three walls of the equilateral triangles are subjected to the uniform wall temperature boundary condition @, the fully developed Nusselt number Nux is equal to 2.49 [170]. However, when the uniform wall heat flux with the uniform circumferential wall temperature boundary condition ~ is applied, the Nusselt number Num is determined by the following equation [6]:
28/9 NUll1 = 1 + 1A2S*+ 4°AIBr'
(5.239)
The internal heat source and viscous dissipation effects are considered in Eq. 5.239. The Nusselt number for uniform wall heat flux in both the axial and circumferential directions under the ~) boundary condition Num is found to be 1.892 [171]. Since sharp triangular ducts are rarely seen in practical use, triangular ducts with rounded corners, such as that presented in Fig. 5.27b, have been investigated by Shah [172]. His results are presented in Table 5.38, in which the y and Ymaxrefer to the distances measured from the duct base to the centroid and to the point of maximum fluid velocity, respectively.
FORCED CONVECTION,INTERNALFLOW IN DUCTS
5.75
TABLE 5.38 Fully Developed Laminar Flow and Heat Transfer Characteristics of Equilateral Triangular Ducts With Rounded Comers [172]
Dh/2a y/2a Ymax[2a Umax~Urn K(oo) L~y f Re Num NUH2
No rounded corners
One rounded comer
Two rounded comers
Three rounded corners
0.57735 0.28868 0.28868 2.222 1.818 0.0398 13.333 3.111 1.892
0.59745 0.26778 0.28627 2.172 1.698 0.0659 14.057 3.401 2.196
0.62115 0.30957 0.29117 2.115 1.567 0.0319 14.899 3.756 2.715
0.64953 0.28868 0.28868 2.064 1.441 0.0284 15.993 4.205 3.780
Isosceles Triangular Ducts. For isosceles triangular ducts like those shown in Fig. 5.27c, the velocity distribution and friction factors for fully developed laminar flow are expressed by the following set of equations suggested by Migay [173]: 1 (dp)y2z2tan2*[(z) B2 u =~
~
1 _ tanZ ~
2b2(dP)
u~ =   ~
~
] 1
(B 2) tan2 ~
dxx (B + 2)(1

tan
2
~)
12(B + 2 ) ( 1  tan 2 ¢) f R e = (B  2)[tan ¢ + (1 + tan 2 ,)1/212 B = [4 + 5/z(cot2 ¢  1)]1/2
(5.240)
(5.241)
(5.242) (5.243)
Apparently, when 2~ = 90 °, f Re is indeterminate from Eq. 5.242. Migay [173] obtained another relation for 2~ = 90 °, as follows: 12(B + 2)(1  3 tan z ~) f R e = 13/2tan ~[4 tan z ~ + 5/2(1 tan 2 ~)]1,1 _ 2}[tan ~ + (1 + tan 2 ,)1/212
(5.244)
The remaining flow and heat transfer characteristics, represented by K(oo), L+hy,NUT, NUll1, and NUll2, together with f R e , are given in Table 5.39 [1]. The fully developed Nusselt numbers NUT and NUll1 for laminar flow in isosceles triangular ducts with one wall or more heated are given in Table 5.40. Right Triangular Ducts. For right triangular ducts, shown in Fig. 5.27d and e, the fully laminar developed flow and heat transfer characteristics f R e , K(oo), Num, and Null2 are given in Fig. 5.28 [2]. The data for this figure were taken from HajiSheikh et al. [170], Sparrow and HajiSheikh [174], and Iqbal et al. [175]. Arbitrary TriangularDucts. For triangular ducts with arbitrary angles such as that shown in Fig. 5.27f, the fully developed friction factors and Nusselt numbers are presented for fully developed laminar flow in Figs. 5.29 and 5.30 [2].
Thermally and Simultaneously Developing Flows. Hydrodynamically developing laminar flow in triangular ducts has been solved by different investigators as is reviewed by Shah and London [1]. Wibulswas [160] obtained a numerical solution for the problem of simultaneously
TABLE 5.39 Flow and Heat Transfer Characteristics for Fully Developed Laminar Flow in Isosceles Triangular Ducts [1]
2b/2a
2~
K(oo)
Lh~
oo 8.000 5.715 4.000 2.836 2.000 1.866 1.500 1.374 1.072 1.000 0.866 0.750 0.714 0.596 0.500 0.289 0.250 0.134 0.125 0
0 7.15 10.00 14.25 20.00 28.07 30.00 36.87 40.00 50.00 53.13 60.00 67.38 70.00 80.00 90.00 120.00 126.87 150.00 151.93 180.00
2.971 2.521 2.409 2.271 2.128 1.991 1.966 1.898 1.876 1.831 1.824 1.818 1.824 1.829 1.860 1.907 2.165 2.235 2.543 2.574 2.971
0.1048 0.0648 0.0590 0.0533 0.0484 0.0443 0.0436 0.0418 0.0412 0.0401 0.0399 0.0398 0.0399 0.0400 0.0408 0.0421 0.0490 0.0515 0.0644 0.0659 0.1048
f
Re
NuT
Num
Num
12.000 12.352 12.474 12.636 12.822 13.026 13.065 13.181 13.222 13.307 13.321 13.333 13.321 13.311 13.248 13.153 12.744 12.622 12.226 12.196 12.000
0.943 1.46 1.61 1.81 2.00 2.22 2.26 2.36 2.39 2.45 2.46 2.47 2.45 2.45 2.40 2.34 2.00 1.90 1.50 1.47 0.943
2.059 2.348 2.446 2.575 2.722 2.880 2.910 2.998 3.029 3.092 3.102 3.111 3.102 3.095 3.050 2.982 2.680 2.603 2.325 2.302 2.059
0 0.039 0.080 0.173 0.366 0.747 0.851 1.22 1.38 1.76 1.82 1.89 1.84 1.80 1.59 1.34 0.62 0.490 0.156 0.136 0
TABLE 5.40 Fully Developed N u t and Num for Heat Transfer in Isosceles Triangular Ducts With One Wall or More Walls Heated [1 ] NUT
NUll1
A A A
2b
A A A
2a
~ degrees
5.000 2.500 1.667 1.250
0 5.71 11.31 16.70 21.80
1.885 m 2.058 2.227 2.312
0.000 0.822 1.268 1.525 1.675
1.215 1.416 1.849 2.099 2.237
1.215 1.312 1.573 1.724 1.802
2.059 2.465 2.683 2.796 2.845
0 1.003 1.515 1.807 1.978
1.346 1.824 2.274 2.541 2.695
1.346 1.739 1.946 2.074 2.141
1.000 0.833 0.714 0.625 0.556
26.56 30.96 34.99 38.66 41.99
2.344 m 2.311 ~ 
1.758 ~ 1.812 ~ ~
2.301 2.319 2.306 2.274 2.232
1.831 1.822 1.787 1.735 1.673
2.849 w 2.778 ~ ~
2.076 w 2.146 ~ ~
2.773 2.801 2.792 2.774 2.738
2.161 2.146 2.107 2.053 1.989
0.500 0.450 0.400 0.350 0.300
45.00 48.01 51.34 55.01 59.04
2.162 m ~ 1.923 ~
1.765 1.633 
2.183 2.127 2.055 1.968 1.861
1.606 1.529 1.433 1.315 1.173
2.594 w 2.332 ~
2.111 ~ 1.991 w
2.696 2.646 2.583 2.505 2.412
1.921 1.843 1.746 1.628 1.486
0.250 0.200 0.150 0.100 0.050
63.43 68.20 73.30 78.69 84.29
1.671 1.512 1.330 1.126 0.895
1.471 1.361 1.229 1.071 0.878
1.733 1.581 1.401 1.182 0.893
1.004 0.805 0.578 0.332 0.106
2.073 1.917 1.748 1.576 1.418
1.843 1.746 1.635 1.515 1.398
2.301 2.174 2.032 1.881 1.737
1.316 1.114 0.874 0.587 0.244
0
90.00
0.6076
0.608
~

1.346
1.346
~
5.76
FORCED CONVECTION, INTERNAL FLOW IN DUCTS 3.0
5.77
3.0 ~ 13.2
NUT 2.5
NUHI
13.0

2.0
2.5 " 12.8
NUHI
fRe
fRe
I2.,.,I
1.5
a* =
K ( = )  12.6
2b/2a
K(=) 1.0

Null2
Null2 0.5
2.0
~

12.4 I

12.2
~
! o
1.5 i  12.0 1.0
o
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
0l s
FIGURE 5.28 Fully developed fRe and Nu for right triangular ducts [2].
13.4 a* = 1.0 13.2 0.5
13.0 12.8 fRe 12.6
12.4
2c > 2= a*'
12.2
2a
I~ 2,= .~1 12.0 0
10
20
30
40
50
60
70
80
90
24,, d e g
FIGURE 5.29 Fully developed friction factors for arbitrary triangular ducts [2].
developing laminar flow for equilateral triangular and rightangled isosceles triangular ducts for Pr = 0, 0.72, and ~. His results for uniform wall temperature and axial uniform wall heat flux with circumferential uniform wall temperature boundary conditions are presented in Tables 5.41 and 5.42. Since Pr = ~ implies that the flow is hydrodynamically developed, the results for Pr = oo can be applied to all fluids in thermally developing laminar flow.
5.78
CHAPTER FIVE 3.25 a* = 1.0
3.00
_ NUHI 

 
NuT
0.5
2..50 ~
~
~
m
m
., ~
I
...~
I
~
0.7
Nu 2.00
/
'
/
__
L 1.501~
/I/ ,// Ix
I
/
I
S _/
s S s
..~~. ~ ~
0.3
/ss S s
~,"s
I../¢~,J/S~,~ S
/"/
0
~
2c>_2.
_
~ " 2.~"]
1.00 F
0.751
q,¢, / I
I
I
I
I
I
10
20
30
40
50
.... I 60

I
I
70
80
)0
2~, de8 F I G U R E 5.30
Turbulent
Fully developed Nusselt numbers for arbitrary triangular ducts [2].
Flow
T h e l o w e r limit of Recrit is c o n s i d e r e d to be a p p r o x i m a t e l y 2000 in t r i a n g u l a r ducts [45]. N o reliable results for the friction factor and Nusselt n u m b e r are available for transition flow in t r i a n g u l a r ducts. In this section, the t u r b u l e n t flow and h e a t t r a n s f e r characteristics for equilateral, isosceles, and right t r i a n g u l a r ducts are p r e s e n t e d .
TABLE 5.41 Local and Mean Nusselt Numbers for Thermally and Simultaneously Developing Laminar Flows and Equilateral Triangular Ducts [160]
1
Nux,'r
Nu,,,~
NUx,H1
Num,m
x*
Pr = oo
0.72
0
oo
0.72
0
oo
0.72
0
oo
0.72
0
10 20 30 40 50 60 80 100 120 140 160 180 200
2.57 2.73 2.90 3.08 3.26 3.44 3.73 4.00 4.24 4.47 4.67 4.85 5.03
2.80 3.11 3.40 3.67 3.93 4.15 4.50 4.76 4.98 5.20 5.40 5.60 5.80
3.27 3.93 4.46 4.89 5.25 5.56 6.10 6.60 7.03 7.47 7.88 8.20 8.54
3.10 3.66 4.07 4.43 4.75 5.02 5.49 5.93 6.29 6.61 6.92 7.18 7.42
3.52 4.27 4.88 5.35 5.73 6.08 6.68 7.21 7.68 8.09 8.50 8.88 9.21
4.65 5.79 6.64 7.32 7.89 8.36 9.23 9.98 10.59 11.14 11.66 12.10 12.50
3.27 3.48 3.74 4.00 4.26 4.49 4.85 5.20 5.50 5.77 6.01 6.22 6.45
3.58 4.01 4.41 4.80 5.13 5.43 6.03 6.56 7.04 7.50 7.93 8.33 8.71
4.34 5.35 6.14 6.77 7.27 7.66 8.26 8.81 9.30 9.74 10.17 10.53 10.87
4.02 4.76 5.32 5.82 6.25 6.63 7.27 7.87 8.38 8.84 9.25 9.63 10.02
4.76 5.87 6.80 7.57 8.20 8.75 9.73 10.60 11.38 12.05 12.68 13.27 13.80
6.67 8.04 9.08 9.96 10.65 11.27 12.35 13.15 13.82 14.46 15.02 15.50 16.00
FORCED CONVECTION, INTERNAL FLOW IN DUCTS
5.79
Local and Mean Nusselt Numbers for Thermally and Simultaneously Developing Laminar Flows in RightAngled Isosceles Triangular Ducts [160]
TABLE 5.42
1
Nux,T
NUm,T
Nux,m
NUm,H1
x*
Pr = oo
0.72
0
oo
0.72
0
oo
0.72
0
oo
0.72
0
10 20 30 40 50 60 80 100 120 140 160 180 200
2.40 2.53 2.70 2.90 3.05 3.20 3.50 3.77 4.01 4.21 4.40 4.57 4.74
2.52 2.76 2.98 3.18 3.37 3.54 3.85 4.15 4.43 4.70 4.96 5.22 5.49
3.75 4.41 4.82 5.17 5.48 5.77 6.30 6.75 7.13 7.51 7.84 8.10 8.38
2.87 3.33 3.70 4.01 4.28 4.52 4.91 5.23 5.52 5.78 6.00 6.17 6.33
3.12 3.73 4.20 4.58 4.90 4.17 5.69 6.10 6.50 6.82 7.10 7.33 7.57
4.81 5.85 6.48 6.97 7.38 7.73 8.31 8.80 9.18 9.47 9.70 9.94 10.13
3.29 3.58 3.84 4.07 4.28 4.47 4.84 5.17 5.46 5.71 5.95 6.16 6.36
4.00 4.73 5.23 5.63 5.97 6.30 6.92 7.45 7.95 8.39 8.80 9.14 8.50
5.31 6.27 6.85 7.23 7.55 7.85 8.37 8.85 9.22 9.58 9.90 10.17 10.43
4.22 4.98 5.50 5.91 6.25 6.57 7.14 7.60 8.03 8.40 8.73 9.04 9.33
5.36 6.51 7.32 7.95 8.50 8.99 9.80 10.42 10.90 11.31 11.67 12.00 12.29
6.86 7.97 8.68 9.20 9.67 10.07 10.75 11.32 11.77 12.14 12.47 12.75 13.04
Fully Developed Flow Equilateral Triangular Ducts. The friction factor for an equilateral triangular duct has been measured by Altemani and Sparrow [176]. Their data are fitted by the following equation in the region of 4000 < Re < 8 × 104: f=
0.0425 Re0. 2
(5.245)
These researchers have also obtained the fully developed Nusselt numbers for air (Pr = 0.7) in the range of 4000 < Re < 8 × 104 in an equilateral triangular duct with the ~ boundary condition on two walls and the third wall insulated as follows: Nun1 = 0.019Re °'781
(5.246)
Isosceles Triangular Ducts. Bhatti and Shah [45] r e c o m m e n d e d that the friction factor for fully developed turbulent flow in isosceles triangular ducts can be determined using different correlations. For 0 < 2~ < 60 °, the circular duct correlations in Table 5.8 can be used with Dh replaced by D~, as defined by B a n d o p a d h a y a y and A m b r o s e [177]: D~ = ~
3 In cot ~ 
2 In tan ~ 
In tan
(5.247)
where 0 = (90 °  ~)/2. For 2~ = 60 °, the circular duct correlations in Table 5.8 should be used with Dh replaced by D~, which is equal to V3a. For 60 ° < 2~ < 90 °, the use of circular duct correlations with Dh is probably accurate enough. For 2~ = 90 °, the circular duct correlations in Table 5.8 can be used. For 2~ > 90 °, no definite r e c o m m e n d a t i o n can be made at this moment. Right Triangular Ducts. The fully developed turbulent friction factor for two rightangled triangular ducts and an equilateral triangular duct are shown in Fig. 5.31 [45]. The data are from Nikuradse [178] and Schiller [179]. Also provided in this figure are the correlations for computing the friction factor for each triangular duct.
Thermally Developing Flow. Altemani and Sparrow [176] have conducted experimental m e a s u r e m e n t s of the thermally developing flow of air (Pr = 0.7) in an equilateral triangular duct with the ~) boundary condition on two walls and the third wall insulated. The local Nusselt numbers Nux, m and the thermal entrance lengths from their results are given in Figs. 5.32 and 5.33, respectively.
0.100 0.050
(a)
f = 0.079 Re"°.25
0.020 0.010 0.0O5 _ 0 Nikuradse [178] • Schiller [1791 0.002 
/ / ~
(=) t 2 X 102 4
I 6
(b) II 103
! 2
,, t
to) ~ ,I 6 104 Re
, 4
I 2
,t 4
6 10s
2
FIGURE 5.31 Fully developed friction factor for turbulent flow in smooth rightangled and equilateral triangular ducts [45]. 140 0 Measurements [176]
120 Adiabaticwall Re= 59,130
100
80
NUx,HI
0
0
40,
0 JI
60
28,570
20,130 40 13,860 9,730 20 ~ 10
0
.
20
.
.
40
.
.
60
6,740
80
x/Dh FIGURE 5.32 Local Nusselt numbers Nux,m for thermally developing and hydrodynamically developing turbulent airflow (Pr = 0.7) in a smooth equilateral triangular duct [176]. 5.80
FORCED CONVECTION,INTERNALFLOW IN DUCTS 50
I
I
' I
I. . . . . .
5.81
I
O Measurements [176] 40
Lth, HI
D,
(1
,,,,,,
30.
20I
I
lO
....
L
o
I
.... I
2 x 1(#
4 x 1(#
,
I
6 x 1(#
Re
FIGURE 5.33 Thermal entrance lengths for thermally developing and hydrodynamically developing turbulent airflow (Pr = 0.7) in a smooth equilateral triangular duct [176].
For equilateral triangular ducts having rounded corners with a ratio of the corner radius of curvature to the hydraulic diameter of 0.15, Campbell and Perkins [180] have measured the local friction factor and heat transfer coefficients with the @ boundary condition on all three walls over the range 6000 < Re < 4 x 104. The results are reported in terms of the hydrodynamically developed flow friction factor in the thermal entrance region with the local wall (T~) to fluid bulk mean (Tm) temperature ratio in the range 1.1 < Tw/Tm< 2.11, 6000 < Re < 4 x 104, and 7.45 < X/Dh < 72. These data were correlated by
fo
( Zw lO'40+(X/Oh)"0"67  \ T,, /
(5.248)
where f, so denotes the friction factor for isothermal flow, which can be calculated from either the Blasius formula or the PKN formula presented in Table 5.8. In these calculations, kinematic viscosity v entering Re = UmDh/Vmust be evaluated at the duct wall temperature Tw. The following correlation has been obtained by Campbell and Perkins [180] from their measurements for local Nusselt numbers: Nux,H1 = 0.021Re °'8 Pr °'4 (\~m] Tw / 0.7¢
(5.249)
For 6 < X/Dh < 50, the correction factor • is given by
•=
l+\oh]
\Tm/ J
(5.250)
and for X/Dh> 50, (I)= 1. Equation 5.249 is valid for 6 <X/Dh < 123 and 1.10 < Tw/Tm< 2.11. The fluid properties v, tx, and k entering Eq. 5.249 must be taken at the fluid bulk mean temperature T,,.
5.82
CHAPTERFIVE
ELLIPTICAL DUCTS Elliptical ducts can be thought as a family of ducts, including geometries ranging from lenticular to circular ducts. The major and minor axes lengths are represented by 2a and 2b, respectively, in this section. The origin of the coordinate is the intersection of the major and minor axes.
Laminar
Flow
Presented in this section are the friction factors and Nusselt numbers for laminar flow in elliptical ducts. Fully Developed Flow. The velocity distribution for fully developed laminar flow in elliptic ducts with major axis 2a and minor axis 2b is given by Shah and London [1] as follows: u
I
Um
Um
~
1+
"~X
o~. 2
fRe
'
2(1
=
E(m)
+ o~ . 2 )
(5.252)
In the preceding equations, m = 1 + ct.2, and E(m) is the complete elliptic integral of the second kind. The hydraulic diameter of the elliptical duct is Dh = nb/E(m), and the crosssectional area is Ac = nab. The expression for hydrodynamic entrance length L~y developed by Bhatti [181] is as follows: 0.5132 ( E ( m ) ) 2 Lhy  1 + O~:~2 n
(5.253)
+
The fully developed incremental pressure drop number K(oo) for elliptic ducts has been found to be independent of the duct aspect ratio o~* = 2b/2a [187]. The value of K(oo) is recommended to be 1.26 for practical calculations [2]. The fully developed Nusselt numbers NuT, Nun1, NUn2 are displayed in Fig. 5.34. The values of the data used in this figure are derived from the results by Tyagi [6], Tao [182], Iqbal et al. [175], and Dunwoody [183]. 6.0
i
i
I
1
I
i
1
i
I'1
I
I
I
I
I
I
['
I
I
m=
5.0 ~
NUHI
~

4.0 
NUT
.

i
Nu
~
~
~ = " ~ = j
ii
u m
3.0 ,B
u
2.0
m,
i 1.0
m
a* = 2 b / 2 a
0.0,,,'J'
I
0.0
oa
F I G U R E 5.34
I
Fully
I
0.2
i
i
0.3
I
I
a
0.4
developedNussclt
I
0.s
I
I
0.6
I .I
0.7
I
I
o.s
I
u
I
019
numbers for elliptical ducts [2].
J 1.0
FORCEDCONVECTION,INTERNALFLOWINDUCTS
5.83
Hydrodynamically Developing Flow. Bhatti [181] has analyzed hydrodynamically developing flow in elliptic ducts. The apparent friction factors and incremental pressure drop numbers can be expressed as: lapp R e 
Ap* 4x+,
Ap* =
K(x) =
2(1  11)(1 + 311)  (1 + 1"1)2 In 1"13 3(1 + rl) 2
(31]3 + 9rl 2 + 211] + 7)(1  rl) 6(1 + 11)2
(5.254)
(5.255)
where 11 is a boundary layer parameter that refers to the fraction of the duct cross section carrying inviscid flow. The term rl is determined from the following equation implicitly as a function of x+: 16(1 + a , 2 )
E(m)
(5.256)
x+ = r12  1  In 112
Thermally Developing Flow. The local Nusselt number for thermally developing flow in elliptic ducts with uniform wall temperature Nux,v was obtained by Dunwoody [183] in terms of a double infinite series. These results are considered the most accurate as x* > 0.005. Dunwoody's formula for calculating the mean Nusselt number is as follows: NU.,,T = ~ 1 +  ~
(5.257)
The ;~ and C values for Eq. 5.257 are given in Table 5.43. The Constants ~, and C for Eq. 5.257 [183]
TABLE 5.43
a*
k
C
0.0625 0.125 0.25 0.5 0.8
14.59 14.90 15.17 14.97 14.67
0.0578 0.0388 0.0239 0.0158 0.0138
For x* < 0.005, the following formula obtained by Richardson [184] is recommended:
3((e(x*)2+(l(x*) NUm.T= F(a/3)(9x+)~,3 1 +
3)
36
(5.258)
For elliptic ducts subjected to the (~ thermal boundary condition, Someswara et al. [185] have solved thermally developing flow. The mean Nusselt number Num,m can be computed using the following expression: 2.61F Num.nl x,1/3
(5.259)
where factor F is a function of 0~* and can be calculated by the following [2]: F = 1.2089 0.795(x* 4.3011(x .2 + 23.8465(x . 3  44.7053(x .4 + 37.0874(x . 5  11.4809(x .6 (5.260) Equation 5.260 is accurate within +3 percent error to the original data given by Someswara et al. [185].
5.84
CHAPTERFIVE
Turbulent Flow
The friction factors for fully developed turbulent flow have been measured by Barrow and Roberts [186] in elliptical ducts with co* = 0.316 and 0.415 in the range of 103 < Re < 3.105 and by Cain and Duffy [187] in elliptical ducts with ct* = 0.5 and 0.667 in the range 2 x 104 < Re < 1.3 x 105. Based on the data presented by Barrow and Roberts [186] and Cain and Duffy [ 187], Bhatti and Shah [45] have derived the following correlation to calculate the friction factor: f = 0.4443 + 2.2168o~*  2.0431ct .2 + 0.3821c~.3
(5.261)
fc
where fc is the friction factor for a smooth circular duct (c~* = 1), which can be obtained from the Blasius equation in Table 5.8. It should be noted that Eq. 5.261 is valid for 0.136 < c~* < 1. Heat transfer in fully developed turbulent flow in elliptical ducts has been determined in several investigations. A comparison of the different results has been presented by Bhatti and Shah [45]. It was concluded that the Gnielinski correlation for circular ducts can confidently be used to calculate the fully developed Nusselt number for elliptical ducts for fluids of Pr ___0.5. For liquid metals, the fully developed Nusselt numbers can be determined using Eq. 5.236 for elliptical ducts with the @ boundary condition. The values of Nus~ugrequired in Eq. 5.236 are given in Fig. 5.35. 10.0
9.5
Nudug 9.0
8.5
8.0
0
0.2
0.4
0.6
0.8
1.0
(X* FIGURE 5.35 Slug flow Nusselt numbers for elliptical ducts with the ~ boundary condition [169].
CURVED DUCTS AND HELICOIDAL PIPES The most prominent characteristic of flow in curved ducts and helicoidal pipes is the secondary flow induced by centrifugal force due to the curvature of the pipe. Consequently, the friction factor is higher in curved pipes than that in straight pipes for the same Reynolds number. The pitch of the helicoidal pipe also has an effect on the flow. As a result, the heat trans
FORCED CONVECTION, INTERNAL FLOW IN DUCTS
5.85
fer rate is higher in curved or helicoidal pipes than in straight pipes. Therefore, curved or helicoidal pipes are widely used in engineering applications. Spiral coils are curved ducts with varying curvature. The friction factor and heat transfer rate for spiral coils are also included in this section. In addition to the dimensionless parameters used in straight pipes, the following parameters are particularly useful in the case of curved ducts or helicoidal pipes: the Dean number De; the helical number He, and the effective radius of curvature Re. These are defined as follows: De = Re
(5.262)
[ { b t2]m ]
He = Re(~c) ~/2= De 1 + \  ~ j
RcRll
(5.263)
+ ( 2n~R)]
(5.264)
where a denotes the radius of a circular pipe, b represents the coil pitch, and R is the radius of the coil. In this section, emphasis will be given to the correlations used for calculating the friction factors and Nusselt numbers for laminar and turbulent flows in curved ducts, helicoidal pipes, and spiral ducts. These will be presented as the ratio of the friction factor in curved ducts to the friction factor in straight ducts and the ratio of the Nusselt number in curved ducts to the Nusselt number in straight ducts Nuc/Nu, in most cases. The subscript c represents curved ducts or helicoidal pipes, while the subscript s denotes straight pipes of the same shape.
fc/fs
Fully Developed Laminar Flow Dean [188, 189] first studied the velocity profile of flow in helicoidal pipes using perturbation analysis. His result is valid only for De < 20, where the velocity distribution is almost identical to that found in straight ducts. Mori and Nakayama [190] have obtained the solution for De > 100 for a coil with R > a. Their results are in agreement with the experimental data reported by Mori and Nakayama [190] and Adler [191] and numerical simulations [192]. The friction factors for fully developed flow in helicoidal pipe proposed by Srinivasan et al. [193] in the range of 7 < < 104 follow:
R/a fc [1  =/0.419De °275 fs [0.1125De °5
for D e < 3 0 for 30 < De < 300 for De > 300
(5.265)
However, after reviewing the available experimental data and theoretical predications, Manlapaz and Churchill [194] recommended the following correlations, which contain both the Dean number De and radius ratio of
R/a:
fc_ I(1 . 0 
f, 
018
[1 + (35/De)2] °5
+ 1.0 + ~
88.33
(5.266)
where m = 2 for De < 20; m = 1 for 20 < De < 40; and m = 0 for De > 40. It can be observed that Eq. 5.265 does not include the parameter R/a and will not be used for all the range of R/a. After a comparison of Eq. 5.266 with experimental data, Shah and Joshi [195] suggested that Eq. 5.265 be used for the coils with R/a < 3 and that either Eq. 5.265 or Eq. 5.266 be used for coils with R/a > 3.
5.86
CHAPTER FIVE
The friction factors for spiral coils can be obtained using the following correlation [193], which is valid for 500 < Re (b/a) °5 < 20,000 and 7.3 < b/a < 15.5. fc =
0"62(n°7  n°7)2 Re0.6(b/a)O.3
(5.267)
where nl and n2 represent the numbers of turns from the origin to the start and the end of a spiral. The critical Reynolds number, which is used to identify the transition from laminar to turbulent flow in curved or helicoidal pipes, has been recommended for design purposes by Srinivasan et al. [193]: Recrit = 210011+ 12(R) °5]
(5.268)
However, for spiral pipes, no single Recrit exists because of varying curvature. The minimum value of Recrit can be obtained using Rmax to replace R in Eq. 5.268, and the maximum value of Recnt can be determined using Rmin tO replace the R in Eq. 5.268. The fully developed Nusselt numbers for laminar flow in helicoidal pipes subjected to the uniform wall temperature have been obtained theoretically and experimentally by Mori and Nakayama [196], Tarbell and Samuels [197], Dravid et al. [198], Akiyama and Cheng [199], and Kalb and Seader [200]. A comparison of these results has been made using the ManlapazChurchill [201] correlation. In Fig. 5.36, experimental and theoretical results [196200] are compared with the following ManlapazChurchill [201] correlation based on a regression analysis of the available data: NuT =
where
[(4"343) 3.657 +
3 xl
957 )2, x3 = 1.0 + De 2 Pr
( De/3/211/3 + 1.158 ~ \x2/ J x2 = 1.0 +
0.477 Pr
(5.269) (5.270)
It can be seen that the prediction calculated from Eq. 5.269 agrees well with the experimental data in most cases, except for Pr = 0.01 and 0.1 at intermediate He values. The fully developed Nusselt numbers for spiral coils with uniform wall temperature are suggested by Kubair and Kuldor [202, 203], as follows: NuT 1.98 + 1.8 which is valid in the ranges of 9 ,", 0
¢, 0
D
D
rt ,",
o
z
"
"
~®
~'~
FORCED CONVECTION, INTERNAL FLOW IN DUCTS TABLE 5.44
5.89
Numerically Calculated NUll2 for Helical Coils of Circular Cross Sections [201] NUp,H2
R/a
b/R
5.0
0.0 0.5 1.0
5.0
0.0 0.5 1.0
10.0
0.0 0.5 1.0
5.0
0.0 0.5 1.0
10.0
0.0 0.5 1.0
5.0
0.0 0.5 1.0
Re 9.196 9.197 9.194 46.70 47.72 46.79
De
He
4.113 4.113 4.112
4.113 4.100 4.061
Pr = 0.1
0.3162
1.0
10.0
4.642 4.462 4.462
4.639 4.639 4.640
4.633 4.633 4.634
4.620 4.620 4.621
4.936 4.934 4.929
8.447 8.438 8.414
20.88 20.89 20.93
20.88 20.83 20.67
4.769 4.768 4.765
4.759 4.758 4.755
392.6 393.0 394.4
124.14 124.29 124.72
124.14 123.90 123.17
5.604 5.602 5.596
7.541 7.535 7.518
402.5 403.1 404.9
180.01 180.28 181.07
180.01 179.71 178.82
6.058 6.078 6.071
9.312 9.307 9.292
1008 1009 1013
318.8 319.1 320.5
318.8 318.1 316.5
7.120 7.114 7.103
1043 1045 1051
466.6 467.4 469.8
466.6 465.9 464.0
9.680 9.600 9.588
14.30 14.27 14.23
The fully developed Nusselt numbers NUll2 for helicoidal pipes have been obtained numerically by Manlapaz and Churchill [201]. Their results are listed in Table 5.44. In Table 5.44, it can be seen that the pitch of the helicoidal coil has almost no influence on the Nusselt number. However, the studies by Yang et al. [206, 207] have shown a positive effect of the pitch on the Nusselt number when Pr > 1. In addition, the experiments conducted by Austen and Soliman [208] indicated that the Nusselt number for the laminar flow of water (3 < Pr < 6) in the uniformly heated helicoidal pipe is in good agreement with the prediction from Manlapaz and Churchill [201]. To consider the effect of variable viscosity, the viscosity ratio (~.l,m/~.[w) 0"14 is applied. The use of Eqs. 5.269 and 5.272 with their right sides multiplied by (gm/gw) °14 is recommended. The density change of fluids leads to natural convection; consequently, heat transfer is normally enhanced. An experimental correlation has been obtained by AbulHamayel and Bell [209] to account for the density and viscosity variations in helicoidal pipe. Experiments with water, ethylene glycol, and nbutyl alcohol in a helicoidal pipe with the @ boundary condition were conducted. The following equation was derived from the measurement data: Num=
/ Gr ~3.94]f / / G r \z78 ][ 4 . 3 6 + 2 . 8 4 ~ ~ e 2 ) ] [ 1 + 0.9348~~e2 ) x s ] l + [ 0.0276 De °75 Pr °'97 (Urn/°"'4]
\Uw/
J
(5.274) where
xs=exp 
1.33 G r ' ) De:
(5.275)
This correlation is valid for 92 < Re < 5500, 2.2 < Pr < 101 and 760 < Gr' < 106. It reduces to the straight circular duct forced convection Nusselt number value of 4.36 upon neglecting the coil
5.90
CHAPTER FIVE
effect (De ) 0). Equation 5.274 is recommended for those fluids whose densities are strongly dependent on temperature.
Developing Laminar Flow Hydrodynamically developing laminar flow, thermally developing laminar flow, and simultaneously developing laminar flow in helical coils are still under investigation [210212]. Accurate formulas for engineering applications are limited. However, the entrance region of a helical coil is approximately 20 to 50 percent shorter than that of a straight tube. For most engineering applications, when De > 200, the design can be based on fully developed values without significant errors [195].
Turbulent Flow in Coils With Circular Cross Sections The research conducted by Hogg [213] has indicated that turbulent flow entrance length in coils with circular cross sections is much shorter than that for laminar flow. Turbulent flow can become fully developed within the first halfturn of the coil. Therefore, most of the turbulent flow and heat transfer analyses concentrate on the fully developed region. Ito [214] has proposed the following correlation to calculate the friction factor for turbulent flow in helicoidal coils: f~a)
= 0.00725 + 0.076 Re
for 0.034 < Re
< 300
(5.276)
However, Srinivasan et al. [193] has obtained another formula for the turbulent friction factor, as follows: f~2~
= 0.084 Re
for Re
< 700
and
7 <   < 104 a
(5.277)
Either Eq. 5.276 or Eq. 5.277 can be used for design purposes since they are very similar and agree quite well (within +10 percent) with the experimental data for air [215] and water [216] and with the numerical predictions by Patankar et al. [217]. The friction factor for spiral coils can be obtained using the following experimental correlation [193]: 0.0074(n
fc =
0.9 _ nO.9) 1.5
[Re (b/a)°5] °2
(5.278)
Equation 5.278 is valid for 40,000 < Re (b/a) °5 < 150,000 and 7.3 < b/a < 15.5. Since the turbulent Nusselt numbers are independent of the thermal boundary condition for Pr > 0.7, the Nusselt numbers that appear in this section will not be specified with thermal boundary conditions. The following correlation, developed by Schmidt [218] to calculate the turbulent Nusselt number, is suggested for 2 x 104 < Re < 1.5 x 105 and 5 < R/a < 84:
10 36[ Equation 5.279 was obtained using air and water flow in coils. For low Reynolds numbers, Pratt's [219] correlation is recommended: N U t = l + 3 . 4 ~a Nus
for 1.5
x
103 < Re < 2 x 104
This correlation was obtained from experiments using water and isopropyl alcohol.
(5.280)
FORCED CONVECTION, INTERNAL FLOW IN DUCTS
5.91
When the variable thermal properties of the fluid are considered, Orlov and Tselishchev [220] recommend the following correlation: Nuc ( al(Prml °'°25 Nu~  1 + 3.54 R/\~r~ /
R for a > 6
(5.281)
The pitch effect in helicoidal circular pipe has been considered in the investigation conducted by Yang and Ebadian [221]. These researchers have concluded that the effect of pitch is minimum on heat transfer.
Fully Developed Laminar Flow in Curved, Square, and Rectangular Ducts The following formulas are suggested by Shah and Joshi [195] to compute the friction factor for fully developed laminar flow in curved square ducts: (fRe)c  0.1520De °5 (1.0  0.216De °5 + 0.473De 1 + l l l . 6 D e ~5 256.1De 2) (fRe)s for De < 100
(5.282)
(fRe)c  0.2576De °'39 (fRe)s
for 100 < De < 1500
(5.283)
(fRe)c  0.1115De °5 (fRe)s
for De > 1500
(5.284)
The preceding three equations were obtained through the comparison of theoretical investigations [222224] and experimental measurement [225]. The influence of the pitch of coil on the friction factor has been found to be negligible [226, 227]. Friction factors for curved rectangular ducts are provided by Cheng et al. [222], as follows:
f~ fs
Co De *°5 (1.0 + C1 De lr2 + C2 De 1 + C3 De 3/2 + C4 De 2)
(5.285)
Equation 5.284 is valid for De < 700. Co, C1, C2, C3, and C 4 in Eq. 5.284 are constants given in Table 5.45 and De is defined as Re (Dh/R) ~/2. The following correlation, obtained by Cheng et al. [228], is recommended for curved square ducts: Nuts2 = NuT = 0.152 + 0.627(1.414De) °5 Pr °'25
(5.286)
Equation 5.286 is valid for 0.7 < Pr < 5 and 20 _ 0.7. Equation 5.286 can also be used for the 0), ~ , and @ thermal boundary conditions. Furthermore, the appropriate correlation for circular cross section coiled tubes can be adopted with the substitution of the appropriate hydraulic diameter for 2a to calculate the Nusselt number when the parameters are out of the application range as is the case in Eq. 5.286.
Fully Developed Turbulent Flow in Curved Rectangular and Square Ducts For curved rectangular ducts as well as square ducts, when Re* < 8000, the fully developed friction factors can be computed from the following correlation obtained by Butuzov et al. [230] and Kadambi [231]: fc _ 0.435 x 10_3 Re,0.96 [ R__R_/°22 fs \ d* ]
(5.287)
where d* is the shortside length of the rectangular duct and Re* is defined as umd*/v. The term f~ refers to the friction factor in a straight duct with the same aspect ratio as that of curved coil. For Re* > 8000, Eq. 5.276 or Eq. 5.277 for circular ducts can be used with a replaced by 0.5Dh, where D h is the hydraulic diameter of the rectangular duct. The Nusselt numbers for turbulent flow in curved rectangular ducts have been studied by Butuzov et al. [230] and Kadambi [231]. The correlation suggested by Butuzov et al. [230] is as follows: Nuc = 0.117 × 10_2 Re,O.93 [ R_R~°24
Nus
\ d* /
(5.288)
This correlation is valid for 450 < Re* (R/d*) °5 < 7500 and 25 < R/d* < 164. The term Nus in Eq. 5.288 is fully developed Nusselt number for a straight duct.
Laminar Flow in Coiled Annular Ducts Xin et al. [232] experimentally investigated the laminar flow and turbulent flow in coiled annular ducts. The pressure drop was measured for air and water flows. Based on these experimental measurements, the friction factor data can be correlated for laminar and turbulent flow as follows: f = 0.02985 + 75.89[0.5  a tan ( D e77.56  39"88
)/~'](d°Ddi)l45
(5.289)
where D is coil diameter. This equation is valid in the region of De = 3520,000, do/di = 1.611.67, and D/(do  d i ) = 2132. For the heat transfer in laminar flow in coiled annular ducts, Garimella et al. [233] experimentally obtained the following correlation to calculate the heat transfer coefficient:
00
0e094 r069
(5.290)
This equation indeed shows that the Dean number represents the heat transfer in laminar flow; the coil ratio (do di)/D is another factor to affect the heat transfer.
Laminar Flow in Curved Ducts With Elliptic Cross Sections Dong and Ebadian [234] numerically obtained the friction factor for laminar flow in curved elliptic ducts. The friction factor ratio fc/fs is represented by the following expression:
5.93
F O R C E D C O N V E C T I O N , I N T E R N A L F L O W IN D U C T S
= 1 + 0.0031~X.3 De 1"°7
(5.291)
f, where f~ is the friction factor for straight elliptic ducts and a* is the ratio of the minor axis to the major axis of the elliptic duct. In subsequent research [235], thermally developing flow in curved elliptic ducts is analyzed for different 0~* and Prandtl numbers. The local Nusselt numbers along the flow direction are shown in graph form, and the asymptotic values of the Nusselt numbers have been obtained, as is shown in Table 5.46. In a related study, the effects of buoyancy on laminar flow in curved elliptic ducts are discussed by Dong and Ebadian [236]. The Asymptotic Values of the Nusselt Number for Curved Elliptic Ducts [235]
TABLE 5.46
Pr
0~*
R/Dh
Re
De
0.1
0.7
5
50
0.2
4 10 100
849.16 105.15 1977.6
424.6 33.3 197.8
9.70 3.81 6.31
19.22 4.18 11.55
26.65 7.48 16.51
52.79 11.68 37.79
0.5
4 10 100
1271.3 1058.0 1514.1
635.7 334.6 151.4
8.92 6.73 5.07
23.23 15.20 9.38
35.93 23.10 13.83
75.51 54.58 32.84
0.8
4 10 100
881.7 1336.4 118.6
440.8 422.6 11.9
6.57 6.33 3.68
18.07 16.76 3.75
28.99 27.58 4.62
64.91 61.62 9.46
LONGITUDINAL FLOW BETWEEN CYLINDERS Longitudinal flow between cylinders is encountered in the fuel elements of nuclear power reactors, shellandtube heat exchangers, boilers, and condensers, among other applications. A cylinder is considered to be a long circular pipe or rod. The flow and heat transfer characteristics between the cylinders are dependent on their arrangement (e.g., triangular array, square array, etc.) as well as the Reynolds number. In this section, the fully developed friction factor and Nusselt number for longitudinal flow between the cylinders in triangular and square arrays are introduced. For longitudinal flow in other channels formed by the cylinders and the walls, the reader is encouraged to refer to Shah and London [1] and R e h m e [237]. Laminar Flow
The friction factor and Nusselt number for longitudinal laminar flow between a triangular array and a square array are discussed in this section. Triangular Array. A triangular array is shown in Fig. 5.38. The fluid longitudinally flows in the virtual channel formed by the triangular array. The friction factor for fully developed laminar flow in this configuration has been proposed by R e h m e [237] as follows:
5.1777(P/D 1) o.404 36.713(P/D 1) 0.24 f R e = 36.947(P/D  1) 0•372
I /
16(r, 2  1 )3 7~_2     ~~_4SL/_2 1,4r, In r,  3r, + 4r,  1
for 1.02 < P/D < 1.12 for 1.12 < P/D < 1.6 for 1.6 < P/D < 2.0
(5.292) (5.293) (5.294)
for P/D > 2.1
(5.295)
5.94
CHAPTER FIVE
P/DforfRe 16
1.0
1.4
1.8
2.2
2.6
3.0
14
70
J#L
12
60 "NUT
10 Nu
Null1
"d' /
[
fRe
~
,,s
oo'"
50 40 fRe 30 20
Null2
& Rarnachandra[238],Nut
10
0 i._
1.0
1.2
1.4
1.6
1.8
2.0
P/D for NuH!, NUH2,NUT
F I G U R E 5.38 Fully developed f Re and Nusselt numbers for longitudinal laminar flow between cylinders in a triangular array [237].
where
P x/2X/~ Pr, = ~  n = 1.05 D
(5.296)
Equations 5.292 through 5.295 were obtained as a result of comparison with numerous investigations such as those by Rosenberg [239], Sparrow and Loeffler [240], Axford [241, 242], Shih [243], Rehme [244, 245], Johannsen [246], Malfik et al. [247], Ramachandra [238], Mikhailov [248], Subbotin et al. [249], Dwyer and Berry [250], Rehme [251], and Cheng and Todreas [252]. The f Re calculated from Eqs. 5.292 through 5.294 is shown in Fig. 5.38. The fully developed Nusselt number for longitudinal flow in a triangular array with uniform cylinder temperature has been analyzed by Ramachandra [238] and is shown in Fig. 5.38. The fully developed Nusselt numbers for the @ and @ boundary conditions have been studied by Sparrow et al. [253], Dwyer and Berry [250], Hsu [254], and Ramachandra [238]. The differences for NuHi and NUll2 reported by these investigators are small (1 percent). The fully developed Null1 and NUll2 are shown in Fig. 5.38. Miyatake and Iwashita [255] conducted a numerical analysis to determine the characteristics of developing laminar flow between a triangular array of cylinders with a uniform wall temperature and various ratios of pitch to diameter (P/D). The relationships between the local Nusselt number NUx,Tand local Graetz number Gz/and between the logarithmic mean Nusselt number NUtm,Tand Graetz number Gz were obtained as follows: for P/D = 1.01.1"
NUx,T = 9.26(1 + 0.0022Gzxl46) TM NUtm,T = 9.26(1 + 0.0179Gz146) TM
for P/D = 1.14.0
NUx, T = (a 2 +
b 2 Gz2/3) 1/2
NUtm,T = [a 2 + (3b/2) z Gz2/3] 1/2
(5.297) (5.298) (5.299) (5.300)
FORCED CONVECTION, INTERNAL FLOW IN DUCTS
where
a= b=
8.9211 + 2 . 8 2 ( P / D  1)] 1 + 6.86(P/D  1)5/3
5.95
(5.301)
2.3411 + 2 4 ( P / D  1)] [1 + 36.5(P/D  1)5/4][2V~(P/D  1) 2  re]1/3
(5.302)
Gzx = rftcp/kx
(5.303)
G z = mcp/kL
(5.304)
NUx.T = hxD/k
(5.305)
NUtm.T= hmD/k = rftcp( Th  To)x=L ItLA Ttm
(5.306)
(V~ To)(T~ V.)~:~
(5.307)
ATom = In [(Tw To)/(T~ Tb)x=L]
In Eqs. 5.306 and 5.307, To, Tw, and Th are the inlet, wall, and fluid bulk temperatures, respectively; L is the length of the cylinder. It is noted that the fully developed Nusselt number can be calculated using Gz ) 0 (L ~ oo) in the corresponding equations. In the study of Miyatake and Iwashita [256], the relationship of local Nusselt number and Graetz number is formulated for developing longitudinal flow between a triangular array of cylinders with a uniform heat flux and various pitchtodiameter ratios. For P/D = 1.011.1: b Gzx1/3 a NUx.H2= 1 +451Gzx [154P/°14937] + 1 NUx.H2= a; For P/D = 1.14.0: where
when Gzx> (a/b) 3
when Gzx < (a/b) 3
(5.309)
NU~.H2= (a 2 + b 2 Gz2/3)1/2 a=
(5.308)
(5.310)
3 . 1 ( P / D  1)°1+ 3 2 4 ( P / D  1) '.6 1 + 69.5(P/D  1 )2.4
1.53611 + 8.24(P/D  1)°39] b = [2V~(P/D  1) 2  rt]l/3[1 + 6.37(P/D

1)0"73]
(5.311)
Square Array. A square array is displayed in Fig. 5.39. The fully developed friction factor for longitudinal flow in such a virtual channel has been investigated by Sparrow and Loeffler [240], Shih [243], Rehme [244, 245], Mal~ik et al. [247], Meyder [257], Kim [258], Ramachandra [238], and Ohnemus [259]. The f R e is given in Fig. 5.39. It can be approximated by the following equation [237]:
f R e = 40.70( P 
1) 0.435
(5.312)
Equation 5.312 is valid in the range of 1.05 < P/D < 2.0. The fully developed Nusselt numbers for the ~ and @ boundary conditions in square arrays have been analyzed by Kim [258], Ramachandra [238], Ohnemus [259], and Chen et al. [260]. The fully developed Null1 and NUll2 are shown in Fig. 5.39. Miyatake and Iwashita [255] also investigated the developing longitudinal laminar flow between a square array of cylinders with uniform wall temperature. The local and logarithmic Nusselt number can be obtained using the following correlations:
5.96
CHAPTER FIVE P/D for f Re 1.2
1.0 16 r,,
1.4
1.6
1.8
2.0
2.2
2.4
2.6
2.8
3.0
D
14
70
12
60
10
50
8 I'
NUHI
40
fRe
Nu
~"_o"rf
6 I
4 
,pS
 t 30 Null2
it
fRe
I
2 ~s
O Chen et al. [260], Null2 •
Ohnemus [259], Num
A
Ramachandra[238],Nu m
0
 20 " 10
0 1.0
1,2
1.4
1.6
1.8
2.0
P/D for NUHt, NUH2 FIGURE 5.39 Fully developed f Re and Nusselt numbers for longitudinal laminar flow in a square array [237].
F o r P/D = 1.01.2:
Nux, T = 4.08(1 + 0.0058Gz146) TM
(5.313)
NUtm,T = 4.08(1 + 0.0349Gzx]46) TM
(5.314)
For P/D = 1.24.0, the s a m e e q u a t i o n s as Eqs. 5.299 and 5.300 are used, but the a and b are different:
a=
4.0011 + 0.509(P/D  1)] l+0.765(P/D_l)5/3
(5.315)
1.6911 + 9.1(P/D 1)] b = [1 I lO.8(e/o 1)5/4][4(e/o 1) 2  2] 1/3
(5.316)
F o r d e v e l o p i n g longitudinal l a m i n a r flow b e t w e e n a s q u a r e array of cylinders with a unif o r m wall heat flux, the local Nusselt n u m b e r c o r r e l a t i o n s w e r e m a d e by M i y a t a k e and I w a s h i t a [256] as follows: F o r P/D = 1.011.2:
b Gzx]/3  a Nux, m = 1 + 94Gzx [7"66P/°7379] + a Nux, m = a
F o r P/D = 1.24.0:
w h e n Gzx < (a/b) 3
Nux, H2 = (a 2 + b 2 Gzff3) 1/2
w h e n Gzx > (a/b) 3
(5.317) (5.318) (5.319)
FORCED CONVECTION, INTERNAL FLOW IN DUCTS
where
a= b
Fully Developed
Turbulent
=
5.97
3.6(P/D 1) 0.2 4 32.2(P/D 1) 15 1 + 9.1(P/D 1)2.2
(5.320)
1.224[ 1 + 4.40(P/D  1 )0.39] [4(P/D 1) 2/I;]1'311 + 2.66(P/D 1) °"73]
(5.321)
Flow
Fully developed turbulent flow and heat transfer in triangular and square arrays have been analyzed by Deissler and Taylor [261,262]. The friction factors for longitudinal flow between the cylinders in a triangular and a square array are given in Fig. 5.40. Correspondingly, the Nusselt numbers, in terms of the Stanton number, defined as Nu/(Re Pr), are given in Fig. 5.41, where the cylinders are considered to be uniformly heated.
I
e I 
  ' 
F
I
I
i
I IF'~i
I
I

I }
Expedmental data, Pld = 1.12 Kays and Perkins [263] Ct~& ~1~
O.OtO F
0 008 P/d
0.006 ~W,,, ~
~
.. _.
0.004
'
"F
0.002
i
i
.
0.001 o.o~o
~
0.008 ;.>...~ =L__
~
,
,
J~i~'[
"'~r'..~
1__

l
.
.
.
L
,;~
, , ; : , . :
, i:i;. l
.I
~
..... . J
.
,,'x~
~
;
oooo ......
0.002
~..L~~
o o
I. 12 ~ Experimemldate, P/d 1.20 J
   ,  ,  Kays and Perldns [263]       Circular tube
0.001
.. . 104

I
_L_,[
I
I.~
,
~
~
",~
L
" '...w... iii I
l I I I I ..... IO s
IO s
Re
F I G U R E 5.40 Fully developed friction factor for longitudinal turbulent flow between a triangular and a rectangular array [261].
5.98
CHAPTER FIVE
102
" :'" ' "1'"'
' "~~,'
~ , ~ , . _ , , , j Re ,~" x r u"4
4==
" '"]
60~~
\\
1
10"3 o
St
St o
2.0
.=.
! 103
104 = =
Re =3 X lOS.el
lOS •
,
~
' !i IO'~. I
/ L
•
!
! ,,
Ilil
,,,,
1,, lO
I
i
,,,,J IO0
I ,.lt.,,oS I,O00
Pr IO'Z
i
I
;
;i;I
"
l

I
1"i' • 1
! ~1
103
{
~
~
Q 0 =
:
t4
~t~
I J I
I! !,
St
10z
I ""
St
P/d
z.O 1.0
I0 "4
~,~, 2.o I
,
= 103 "
"~'
!~~_., ~
"
105
10 4
"J
!
I
I
!
1 1 1 IIII
Illl
I
I 1 I I111
I0
I00
I I I
=. 10S 1,000
Pr FIGURE 5.41 Fully developed heat transfer characteristics for longitudinal turbulent flow between a triangular array and a square array [261].
FORCED CONVECTION, INTERNAL FLOW IN DUCTS IOO
I
'
" r
"
'
W
I
''
I~
'J
I
I ......
5.99
I
'°t 60
40
J
Nu 30 Pld 20  2.20 1.75o
.~ ._._ o
o
I
100z
2
o
.
o
° o
o
o
~
I_
I
I
3
5
7
t
i0 ~
I
2
3
,
I
,.
,5
t
J
7
~04
Pe FIGURE 5.42 Nusselt numbers for fully developed longitudinal flow between cylinders in a triangular array [263].
Maresca and Dwyer [264] have analyzed the heat transfer of liquid metal flow in a triangular array with uniform longitudinal heat flux. The Nusselt number resulting from their analysis is given in Fig. 5.42.
INTERNALLY FINNED TUBES
Internally finned tubes are ducts with internal longitudinal fins. These tubes are widely used in compact heat exchangers. The friction factorReynolds number product and the Nusselt number for such internally finned tubes, designated as ( f Re)d and NUb+,d, respectively, are computed from the following definitions: (5.322)
aOh,finless=(Uml(Oh I Red =
~
\~'] \'~cc ]finless
q"Dh,nnles~ q" (D~]  ~ = NUb~d k(t~  tin) 4k(t~ tm) \ Ac /f~nl~ss
(5.323)
(5.324)
where Dh, finlessis the hydraulic diameter corresponding to finless ducts. Based on actual geometry, the Dh, finned is used in the f R e and NUbc. The relationships between (fRe)d, NUbc,d, and f Re, NUbc are given in the following expressions:
D hfinless NUb¢
d =
(5.325)
Zcfio+)
NUbc( Dh'finned)2( Dh finless Acfinned
(5.326)
5.100
CHAPTER
FIVE
Circular Ducts With Thin Longitudinal Fins H u and Chang [265] have o b t a i n e d the friction factors and Nusselt n u m b e r s for fully develo p e d laminar flow t h r o u g h a circular duct having longitudinal rectangular fins equally spaced along the wall. The fin's efficiency was treated as 100 percent, while its thickness was t r e a t e d as zero. The fully d e v e l o p e d ( f R e ) a and Num.a for laminar flow in a circular duct with longitudinal fins are given in Table 5.47, in which l* and n are relative fin length and the n u m b e r of fins, respectively. Prakash and Liu [266] have numerically analyzed laminar flow and heat transfer in the e n t r a n c e region of an internally finned circular duct. In this study, the fully d e v e l o p e d f R e is c o m p a r e d with those r e p o r t e d by H u and C h a n g [265] and Masliyah and N a n d a k u m a r [267]. The incremental pressure drop K(oo) and h y d r o d y n a m i c e n t r a n c e length Z+hy t o g e t h e r with f R e are given in Table 5.48, in which the term n refers to the n u m b e r of fins, while l* d e n o t e s the relative length of the fins. TABLE 5.47 The Fully Developed (fRe)d and NuH2.afor Laminar Flow in a Circular Duct With Longitudinal Fins [1] (fRe)a n
l* =0.2
0.4
0.6
0.7
2 8 12 16 20
17.28 21.22 . 25.99 .
20.83 42.87
27.42 101.10 . 219.54 .
31.89 139.55
22 24 28 32
. . . 30.43
n
.
. 69.57 .
. .
. . .
. 348.86
0.795
35.68 161.03
35.98 162.03 286.66 439.37 616.52
36.64 164.84 ~ 448.43 632.11
712.76 813.67 1025.6 1251.6
732.60 838.23 1062.7 1298.7
1546.8
0.795
0.8
0.9
6.16 30.10 53.65 73.48 83.60
6.23 30.65 ~ 71.06 80.41
6.93 27.26 u 31.85
86.82 85.00 75.32 62.43
84.02 83.70 78.06 67.05

434.40 607.72
. . .
0.79
701.75
91.65
. . 372.37
. . 773.69
l* = 0.2
0.4
0.6
0.7
2 8 12 16 20
4.25 4.27 . 4.12 ~
4.32 4.67
4.88 8.66 . 7.29 
5.38 16.79
6.11 29.49
21.65
72.66 81.89
22 24 28 32
. . 3.84
1221.0
0.8
0.9 40.54 172.70 481.12
NUH2,d
.
. 4.04 ~

TABLE 5.48
n

.
84.11
. .
3.39
.

. .
0.79
. .
4.10
8.62
55.76
25.15
Flow Characteristics for the Entrance Problem in an Internally Finned Circular Duct 8
1"
(fRe)
K(oo)
0 0.3 0.6 1
15.96 27.88 97.37 171.8
1.25 2.44 2.85 1.58
16 Lh+y 0.0415 0.0443 0.0320 0.00524
(fRe) . 39.18 208.1 477.4
24
Lh+y
K(oo) .
. 4.11 10.7 1.79
. 0.0438 0.0540 0.00235
(fRe)
K(oo)
46.00 293.0 933.8
5.40 23.5 1.93
Lh+r
. 0.0417 0.0622 0.00136
5.101
FORCED CONVECTION, INTERNAL FLOW IN DUCTS
TABLE 5.49
Heat Transfer Characteristics for Fully Developed Flow in a Finned Circular Duct
n
8
l*
NUT,d
0 0.3 0.6 1.0
3.658 4.110 8.779 33.25
LT, d
0.0421 0.0392 0.00549 0.00574
16 NUHI,d
4.371 5.245 17.26 41.58
+
LHI.d
NUT,d
LT, d
24 NUHI.d
+
LHI.d
NUT,d
+
LT, d
+
NUHI,d
LHI,d
0.0571 . . . . . . 0.0658 3.993 0.0301 5.107 0.0653 3.859 0.0258 4.830 0.0618 0.0848 6.545 0.0411 13.32 0.121 5.313 0.0116 9.154 0.116 0.00774 80.55 0.00336 106.5 0.00379 143.7 0.00246 2 0 0 . 0 0.00227
The fully developed Nusselt numbers for the thermal boundary conditions of uniform wall temperature and axial uniform wall heat flux with circumferential uniform temperature obtained by Prakash and Liu [266] are given in Table 5.49, along with the corresponding thermal entrance lengths. The term n in Table 5.49 denotes the number of fins, whereas l* represents the relative length of the fins.
Square Ducts With Thin Longitudinal Fins Gangal and Aggarwala [268] have analytically obtained the f R e and NUll1 for fully developed flow in a square duct with four equal internal fins, as that shown in Fig..5.43. The fins were treated as having zero thickness and 100 percent efficiency. The results o f f Re and NUHI,dfor fully developed flow are provided in Table 5.50.
TABLE 5.50 Longitudinal Four Thin Fins Within a Square Duct: (fRe)d and NUHI,d for Fully Developed Laminar Flow [268]
[
r
2a
LI
FIGURE 5.43 A square duct with four equal longitudinal thin fins.
l*
(fRe)d
NUm,d
0 0.125 0.250 0.375 0.500 0.625 0.750 1
14.261 15.285 18.281 23.630 31.877 42.527 52.341 56.919
3.609 3.721 4.160 5.172 7.309 11.096 14.025 14.431
Rectangular Ducts With Longitudinal Thin Fins from Opposite Walls The fully developed ( f Re)d and NUHI,d for rectangular ducts with two fins and four fins on opposite walls have beenobtained by Aggarwala and Gangal [269] and Gangal [270]. These are shown in Fig. 5.44.
Circular Ducts With Longitudinal Triangular Fins Nandakumar and Masliyah [271] and Masliyah and Nandakumar [267] have analyzed fully developed laminar flow in a circular duct with equally spaced triangular fins, as shown in the inset in Fig. 5.45. The flow area and wetted perimeter for this type of duct are given by Ac, finned "/T,a2  n[aZ¢~  a(a  l) sin ~] Pfinned  2ha + 2 n l ' 
2n~a
(5.327) (5.328)
CHAPTERFIVE
5.102
6.0
~'
I
~
1
~
~
~
I
~.o _ i _ ~ T
"
Nu""d
6.0
~
where
,~;,~/ 5.0
k__d2
,.o.
H~oH , J iiL".'Y _ L L . . . . . . ~2 0
3.0_
o i i..L T ...........
7/
The hydraulic diameter can be calculated from Oh, finned = and Figs. 5.45 and 5.46, in which the case of 2~ = 0 ° represents longitudinal fins of zero thickness.
4.o
A

/ 1
t~Re);
,  , 3.0 "",,.,,_~.:
/ /
(5.329)
4(Ac/P)finned. T h e results of ( f Re)d and NUHI,d are provided
K / /
r = [a 2 + (a  I) 2  2a(a  1) cos ~]la
Circular
~o);/' / ...~t2.o ~ . / ( r " o ) 7  2.0
Ducts
With
Twisted
Tape
The enhancement of heat transfer inside a circular duct is often achieved by inserting a thin, metal tape in such a way 1.0 ~ * t 1.0 NuHl'd t / R e ) ; = (fRe)d that the tape is twisted about its longitudinal axis, as indicated in Fig. 5.47. Swirl flow is created in this manner. The 0.0 I I1 I I I I I I l 0.0 width of the tape is usually the same as the internal diameter 0.0 0.2 0.4 0.6 0.8 1.0 of the duct. The tape twist ratio X~ is defined as H / d . When l/a XL approaches infinity, the circular duct with the twisted FIGURE 5.44 Friction factor and Nusselt number tape becomes two semicircular straight ducts separated by for fully developed laminar flow in rectangular ducts the tape. with longitudinal thin fins from opposite walls [1]. Manglik and Bergles [272, 273] made an extensive review on the study of laminar and turbulent flow in circular ducts with inserted tape. For laminar flow, the dimensional swirl parameter S w was incorporated in the correlation of friction factor. This parameter considers the thickness of inserted tape & 
I00 80
/
.=±
/
60
/ /////
40
/~
40
'
f'" .
30
20 2o
.
"
' .
.
I
'
I
'
_ . 1. ~  ' 
I
2~ =0 °  24 = 3* /~"'~'\'~"'"  24 =6 ° /f.~'"
'
"
\\

1"=
/ ' i ' , " / / / /
,J
16 I0
8
NUH~d 6 ~4.364
4
[
3  !
/
.....
2~ =6 °
2
4
8
12 n
16
20
""
FIGURE 5.45 Friction factors for fully developed laminar flow in a circular duct with longitudinal triangular fins [1].
l =0.6
///
"" ~ £ = o . ,  . . . j
L/,"
"80

"'~
....
24
•
""
/
I.": .'li~ ..... 'V,'"
/
2~ = 0", 3",6" I0""
. ' ~ ~ 
/
4
, , _ ~ ~ . _ =u.z  , , ,
Ft 8
12
16
20
~
I , 24
FIGURE 5.46 Nusselt numbers for fully developed laminar flow in a circular duct with longitudinal triangular fins [1].
FORCED CONVECTION, INTERNAL FLOW IN DUCTS
H~
A
5.103
TAPE
SECTION AA F I G U R E 5.47
A circular tube with a twisted tape inserted [1].
the twist ratio Xz,, and the helicoidally twisting flow velocity. The definition of dimensional swirl parameter Sw is given as: Sw = Resw V ~ L
1+
n  45/d
(5.330)
For most applications, the following equation is recommended for the calculation of friction factor in laminar flow in circular ducts with inserted tape: f Re = 15.767rl:(rl: + 2  ~)2(rl:  4 ~)3(1 + 2~t )(1 + 106Sw255)1/6
(5.331)
where f R e is based on the empty tube diameter d. The above equation can be applied in the range 0 < (8/d) < 0.1 and 300 ___Sw < 1400. The mean Nusselt number for the laminar flow in isothermal circular ducts with inserted tape can be obtained from the following equation suggested by Manglik and Bergles [272]: Num = 4.612{[(1 + 0.0951Gz°894) z5 + 6.413 x 109(Sw • Pr°391)3853] °2
+ 2.132 x
1014(Re~x
•
Ra)223}°1( \ ~~l'm w // 0"14 (5.332)
Where Gz, Re~x, and Ra are the Graetz number, the Reynolds number based on axial velocity, and the Rayleigh number, respectively. Their definitions are expressed as follows: Gz = m c p / K L
(5.333)
rnkt Re~ = n,d/4  8~
(5.334)
Ra = pgd3~SATw ~ta
(5.335)
For the turbulent flow in circular duct with inserted tape, it was proposed by Manglik and Bergles [273] that the friction factor can be calculated by the following equation:
0.0791( )1.75(2.752)
f = °2TRe n  48/d
1 d Sl.29
(5.336)
It was found that the flow rates with Re > 104 can be considered as fully developed turbulent flow. Therefore, the above equation is a more generalized correlation that covers a broad database of available empirical data for turbulent flow [273].
5.104
CHAPTER FIVE
TABLE 5.51 The Fully Developed (fRe)d and Nud Values for Forced Convection of Laminar Flow in a Semicircular Duct With Internal Fins [274]
Fin length (l*) n
0.1
0.2
0.3
0.4
0.5
1 3 5 8 11 17
43.307 46.836 49.207 
46.129 53.345 60.205 69.103 75.927 84.656
50.369 84.306 105.634 
55.717 87.943 122.503 167.660 201.532 243.454
61.674 m 175.639 265.071 
0.6
0.7
0.8
0.9
1.0
67.462 140.762 234.651 391.297 547.204 815.772
72.148 280.714 506.379 
75.047 170.601 303.208 568.948 910.726 1798.80
76.175 309.012 585.854 
76.314 173.382 309.420 587.054 953.762 1959.23
11.159 24.816 38.581 41.329 32.084 19.041
11.765 m 47.001 77.458 ~
11.839 25.843 44.127 79.918 124.440 211.243
11.839 42.667 76.329 ~
11.821 25.180 42.558 76.021 118.145 226.977
(fRe)d
Num.a 1 3 5 8 11 17
6.806 6.878 6.904 m ~
7.196 7.531 7.668 7.627 7.467 7.176
7.895 9.383 8.954 ~
8.896 12.171 13.104 11.760 10.253 8.808
10.086 m 21.892 19.148 m ~
A g e n e r a l i z e d c o r r e l a t i o n of m e a n Nusselt n u m b e r for t u r b u l e n t h e a t transfer in an i s o t h e r m a l circular duct with i n s e r t e d t a p e was d e v e l o p e d by M a n g l i k and B e r g l e s [273] b a s e d on the e x p e r i m e n t a l data. It is e x p r e s s e d as:
Nu =O.O23(l + O.769/XL) Re°8pr°4(
rt )°8( ~ + 2 + 28/d ) °2 n  48/d n  45/d ~
(~*'l'minor(rm) m
where
~ = \g~/
(5.337)
(5.338)
n  0.18 for liquid heating and 0.30 for liquid cooling; m = 0.45 for gas h e a t i n g and 0.15 for gas cooling.
Semicircular Ducts With Internal Fins D o n g and E b a d i a n [274] have used a very fine grid to p e r f o r m a n u m e r i c a l analysis of fully d e v e l o p e d l a m i n a r flow in a semicircular duct with internal l o n g i t u d i n a l fins. T h e fins are considered to h a v e z e r o thickness, and the n u m b e r of fins n and relative fin length l* = / / a are y t a k e n into account. The (~ t h e r m a l b o u n d a r y c o n d i t i o n is applied. Their results are given in Table 5.51.
Elliptical Ducts With Internal Longitudinal Fins 1 t
xa
 X
FIGURE 5.48 An elliptical duct with internal fins.
A n elliptical duct with f o u r internal l o n g i t u d i n a l fins m o u n t e d on the m a j o r and m i n o r axes, as s h o w n in Fig. 5.48, has b e e n a n a l y z e d by D o n g and E b a d i a n [275] for fully d e v e l o p e d l a m i n a r flow and h e a t transfer. In this analysis, the fins are c o n s i d e r e d to have z e r o thickness. T h e ~ t h e r m a l b o u n d a r y c o n d i t i o n is a p p l i e d to the duct wall, and l* is defined as a ratio of Ha/a = Hb/b. The friction factors and Nusselt n u m b e r s for fully d e v e l o p e d l a m i n a r flow are given in Table 5.52.
FORCED CONVECTION, INTERNAL FLOW IN DUCTS
5.105
TABLE 5.52 Friction Factors and Nusselt Numbers for Fully Developed Flow in an Elliptical Duct With Internal Fins [275] o~*
1"
(f Re)d
NU.,.d
(f Re)
Num
0.0 0.5 0.8 0.9 1.0
72.20 108.61 270.63 301.73 313.62
3.78 3.78 12.51 16.04 45.85
72.20 45.36 69.34 71.53 67.26
3.78 1.58 3.20 3.80 3.40
0.5
0.0 0.5 0.9 1.0
67.26 133.39 297.78 309.20
3.75 4.97 16.31 16.11
67.26 50.02 68.78 61.71
3.75 1.86 3.76 3.21
0.8
0.0 0.5 0.9 1.0
64.33 135.53 391.06 303.38
3.67 5.36 16.19 15.96
64.33 50.72 67.34 58.91
3.67 2.00 3.74 3.10
1.0
0.0 0.5 0.9 1.0
63.99 140.57 294.18 301.82
3.67 5.60 16.14 15.96
63.99 51.46 65.89 58.40
3.67 2.05 3.61 3.09
0.25
OTHER SINGLY CONNECTED DUCTS The fluid flow and heat transfer characteristics for 14 types of singly connected ducts are described in this section.
Y
Sine Ducts
I...2,,,.1
A sine duct with associated coordinates is shown in Fig. 5.49. The characteristics of fully d e v e l o p e d laminar flow and heat transfer in such a duct are given in Table 5.53. These results are based on the analysis by Shah [172].
r..2
FIGURE 5.49 A sine duct.
TABLE 5.53 Fully Developed Fluid Flow and Heat Transfer Characteristics of Sine Ducts [172]
b/a
K(oo)
Lh+y
f Re
Nux
Num
NUll2
oo 2 3/2 1 V/3/2
3.218 1.884 1.806 1.744 1.739
0.1701 0.0403 0.0394 0.0400 0.0408
15.303 14.553 14.022 13.023 12.630
0.739 m 2.60 2.45
2.521 3.311 3.267 3.102 3.014
0 0.95 1.37 1.555 1.47
3A ½ ¼ 1/8 0
1.744 1.810 2.013 2.173 2.271
0.0419 0.0464 0.0553 0.0612 0.0648
12.234 11.207 10.123 9.746 9.600
2.33 2.12 1.80 m 1.178
2.916 2.617 2.213 2.017 1.920
1.34 0.90 0.33 0.095 0
5.106
CHAPTER FIVE
Trapezoidal Ducts A trapezoidal duct is displayed in the inset of Fig. 5.50. Fully developed laminar flow and the heat transfer characteristics of trapezoidal ducts have been analyzed by Shah [172]. The fully developed f Re, Nulls, and NUll2 are given in Figs. 5.50 and 5.51. Farhanieh and Sunden [276] numerically investigated the laminar flow and heat transfer in the entrance region of trapezoidal ducts. The fully developed values of f Re, K(oo), and Nu were in accordance with the results from Shah [172].
24.0
I
_l_J..
i
I
/
'i
¢=
I
I
l
I
I
I
I
i
,~ "  , ,
I
1
I
I
I /
        fRe .... K(==)
2.2
2.0
1.8  , , ,,
22.0
6(7'
%% %
75*
\ ',
" ~ "
",
% %
20.0 fRe
1.6
1.4 K(oo)
/
1B.0
1.2
16.0 1.0 14.0 0.8
12o
_3°.
11.0 0.0
_ _
• * = 2a/2b ~
0.2
0.4
0.6
0.8
a* = 2b/2a
1.0
0.8
0.6
0.4
0.6 0.0
0.2
1/$
FIGURE 5.50
Fully developed f Re and K(o~) for laminar flow in a trapezoidal duct [172].
9.0
3.2
85"
8.0 NUH] ....
7.0
Nufl2
/
60*
'U "l
6.0
2.8
75" 2.4
"!
p,~
2.0
1.6 NUH~
NUll1 5.0 .
.
.
.
.
.
.
.
.
4.0
1.2 ~ . _ _ r
3.0
2.°V 1.0 0.0
FIGURE
30~
" S
0.8
0.4 =* = 2=/2b ~ I
5.51
= 0.2
l
1 0.4
i
I 0.6
i
I 0.8
I
=* = 2b/2= I 1.0 all
I
I 0.8
I
I 0.6
I
I 0.4
I
I 0.2
l
0.0 0.0
F u l l y d e v e l o p e d Nusselt n u m b e r s for l a m i n a r f l o w in a t r a p e z o i d a l pipe [172].
FORCED CONVECTION, INTERNAL FLOW IN DUCTS
5.107
Chiranjivi and Rao [277] experimentally obtained a correlation for laminar and turbulent flow in trapezoidal ducts with one side heated, which is expressed as:
Nu = a Reb pr°52(~)
(5.339)
where a = 6.27 and b = 0.14 for laminar flow, and a  0.79 and b  0.4 for turbulent flow of Reynolds n u m b e r from 3000 to 15,000.
Rhombic Ducts A rhombic duct is depicted in Fig. 5.52. The fully developed flow and heat transfer characteristics of rhombic ducts obtained by Shah [172] are shown in Table 5.54.
_j i iii__ FIGURE 5.52 A rhombic duct.
Quadrilateral Ducts A quadrilateral duct is shown schematically in Fig. 5.53. N a k a m u r a et al. [278] analyzed fully developed laminar flow and heat transfer in arbitrary polygonal ducts. Their results are presented in Table 5.55.
Regular Polygonal Ducts Fully developed flow and heat transfer in a regular polygonal duct with n equal sides, each subtending an angle of 360°/n at the duct center, have been reviewed by Shah and L o n d o n [1]. The f Re and Nu are given in Table 5.56.
FIGURE 5.53 A schematic drawing of a quadrilateral duct.
TABLE 5.54
Fully Developed Laminar Flow and Heat Transfer Characteristics of Rhombic Ducts [172]
0 10 20 30 40 45
K(oo)
Lh+y
2.971 2.693 2.384 2.120 1.925 1.850
0.1048 0.0732 0.0570 0.0477 0.0419 0.0397
f Re
NUll1 NUll2 ¢
K(oo)
Lh+y
12.000 12.073 12.416 12.803 13.193 13.381
2.059 2.216 2.457 2.722 2.969 3.080
1.778 1.673 1.603 1.564 1.551
0.0380 0.0353 0.0336 0.0327 0.0324
0 0.070 0.279 0.624 1.09 1.34
50 60 70 80 90
f Re
NUll1 NUll2
13.542 13.830 14.046 14.181 14.227
3.188 3.367 3.500 3.581 3.608
1.62 2.16 2.64 2.97 3.09
TABLE 5.55 Fully Developed Friction Factors, Incremental Pressure Drop Numbers, and Nusselt Numbers for Some Quadrilateral Ducts [278] (Ih (deg)
(I)2(deg)
(I)3(deg)
(1)4(deg)
fRe
K(~)
NUll1
NUll2
60 50 60 60
70 60 30 30
45 30 45 60
32.23 21.67 71.57 79.11
14.16 14.36 14.69 14.01
1.654 1.612 1.522 1.707
3.45 3.55 3.72 3.35
2.80 2.90 3.05 2.68
5.108
CHAPTERFIVE Fully Developed Laminar Flow Characteristics of Regular Polygonal Ducts [1]
TABLE 5.56
n
fRe
NUHI
NUll2
NUT
3 4 5 6 7
13.333 14.227 14.737 15.054 15.31
3.111 3.608 3.859 4.002 4.102
1.892 3.091 3.605 3.862 4.009
2.47 2.976
8 9 10 20
15.412 15.52 15.60 15.88 16.000
4.153 4.196 4.227 4.329 4.364
4.100 4.159 4.201 4.328 4.364
oo
m 3.657 ,,,
For practical calculations, Schenkel [279] proposed the following formula to compute the f Re in regular polygonal ducts:
(
n2
)4
f Re = 16 0.44 + n 2
(5.340)
The values of the predictions from Eq. 5.337 agree well (within +1 percent) with the tabulated values in Table 5.56.
Circular Sector Ducts
FIGURE 5.54 A circular sector duct.
A schematic drawing of a circular sector duct is presented in Fig. 5.54. The fully developed f Re and Nu for circular sector ducts have been obtained by Eckert and Irvine [280], Sparrow and HajiSheikh [174], Hu and Chang [265], and BenAli et al. [281]. The results are summarized in Table 5.57. Soliman et al. [282] numerically analyzed the problem of laminar flow development in circular sector ducts. Their fapp Re and L~,y results for 2~ = 11.25, 22.5, 45, and 90 ° are presented in Table 5.58. Furthermore, the hydrodynamic entrance lengths are 0.235, 0.144, 0.108, and 0.0786 for hydrodynamically developing flow in circular sector ducts with 2~ = 11.25, 22.5, 45, and 90 °, respectively.
Circular Segment Ducts
\\/
I
\\ \\
FIGURE 5.55 A circular segment duct.
A circular segment duct is depicted in Fig. 5.55. The fully developed flow and heat transfer characteristics obtained by Sparrow and HajiSheikh [283] are given in Table 5.59. Hong and Bergles [284] have analyzed the thermal entrance solution of heat transfer for a circular segment duct with 2~ = 180 ° (i.e., a semicircular duct). Two kinds of thermal boundary conditions are used: (1) a constant wall heat flux along the axial flow direction with a constant wall temperature along the duct circumference, and (2) a constant wall heat flux along the axial flow direction and a constant wall temperature along the semicircular arc, with zero heat
FORCED CONVECTION, INTERNAL FLOW IN DUCTS
TABLE 5.57
5.109
Fully Developed f R e , K(oo), and Nu for Laminar Flow in a Circular Sector Duct
2t~
f Re
K(oo)
NUll1
NUll2
NUT
2t~
f Re
K(oo)
Null1
NUll2
NUT
0 5 8 10 15 20
12.000 12.33 12.411 12.504 12.728 12.98
2.971 2.480 ~ 2.235 ~
2.059 2.245 2.384 ~ 2.619 2.742
0.018 0.081 0.195 0.354
80 90 100 120 150 160
14.592 14.79 14.929 15.200 15.54 15.611
1.530
1.423 ~ 1.692 1.901 2.073
3.671 3.730 3.806 3.906 3.999 4.04
2.984 2.898 2.995 ~
m 3.060
30 36 40 45 50 60 72
13.310 13.510 13.635 13.782 13.95 14.171 14.435
1.855 ~ 1.657 ~ 1.580 ~
3.005 ~ 3.27 3.337 3.479 
0.838 1.174 1.400 1.667 1.990 2.421 2.608
2.341 2.543 ~ 2.70 2.822 ~
180 210 240 270 300 330 350
15.767 15.98 16.15 16.29 16.42 16.54 16.62
4.089 4.127 4.171 4.208 4.244 4.280 4.304
2.923 2.871 2.821 2.781 2.749 2.723 2.708
TABLE 5.58
1.504 1.488 1.468 1.463
3.191 3.268
3.347 3.370 3.389 3.407 3.427 3.443
Flow Parameters for Hydrodynamically Developing Flow in Circular Sector Ducts [282]
,,
x÷
2¢~ = 11.25 °
22.5 °
45 °
90 °
Lhy
fapp Re
K(x)
fapp Re
K(x)
fapo Re
K(x)
fapp Re
K(x)
0.001 0.003 0.006 0.010 0.020 0.030 0.040 0.050 0.070 0.100 0.150 0.200 0.250 0.300 0.350 0.400 0.450 0.500 0.600 0.700 0.800 0.900 1.000
109.3 66.09 48.27 39.52 28.93 25.09 22.70 21.18 19.90 17.34 15.82 14.95 14.47 14.07 13.93 13.59 13.51 13.35 13.19 13.03 12.95 12.90 12.87
0.207 0.335 0.456 0.538 0.758 0.887 0.991 1.076 1.211 1.364 1.540 1.664 1.756 1.828 1.885 1.930 1.970 1.997 2.051 2.091 2.118 2.139 2.156
147.9 81.62 60.18 48.30 35.64 30.20 27.02 24.92 22.20 19.91 17.81 16.65 15.89 16.35 14.95 14.65 14.41 14.22 13.95 13.76 13.63 13.52 13.47
0.177 0.281 0.377 0.469 0.628 0.739 0.830 0.901 1.019 1.153 1.313 1.429 1.517 1.588 1.646 1.695 1.735 1.770 1.825 1.868 1.902 1.930 1.951
180.4 98.27 63.69 53.91 39.59 33.47 29.89 27.51 24.41 21.78 19.40 18.05 17.09 16.54 16.06 15.66 15.34 15.11 14.79 14.55 14.39 14.23 14.15
0.171 0.266 0.352 0.432 0.567 0.662 0.738 0.801 0.903 1.021 1.165 1.269 1.350 1.412 1.464 1.510 1.546 1.577 1.625 1.661 1.689 1.710 1.728
226.0 115.7 78.29 60.26 43.93 37.37 33.50 30.88 27.44 24.40 21.58 19.95 18.87 18.07 17.51 17.05 16.70 16.41 15.96 15.67 15.46 15.31 15.19
0.154 0.241 0.314 0.380 0.492 0.571 0.635 0.689 0.778 0.881 1.007 1.100 1.171 1.229 1.275 1.315 1.347 1.375 1.418 1.451 1.475 1.494 1.509
TABLE 5.59
The f R e , K(oo),
NUll1, and NUll2for Fully Developed Laminar Flow in Circular Segment Ducts [283]
2~
f Re
K(oo)
NUll1
NUll2
2~
f Re
K(oo)
NUll1
NUll2
0 10 20 40 60 80
15.555 15.558 15.560 15.575 15.598 15.627
1.740 1.739 1.734 1.715 1.686 1.650
3.580 3.608 3.616 3.648 3.696 3.756
0 0.013 0.052 0.207 1.456 0.785
120 180 240 300 360
15.690 15.767 15.840 15.915 16.000
1.571 1.463 1.385 1.341 1.333
3.894 4.089 4.228 4.328 4.364
1.608 2.923 3.882 4.296 4.364
5.110
CHAPTER FIVE
TABLE 5.60 Local Nusselt Numbers in the Thermal Entrance Region of a Semicircular Duct [284] NUx.H1
NUx,H2
x*
BC1
BC2
x*
BC1
BC2
0.000458 0.000954 0.00149 0.00208 0.00271
17.71 13.72 11.80 10.55 9.605
17.43 13.41 11.37 10.08 9.141
0.0279 0.0351 0.0442 0.0552 0.0686
4.767 4.562 4.429 4.276 4.217
4.339 4.037 3.830 3.686 3.543
0.00375 0.00493 0.00627 0.00777
7.475 7.723 7.137 6.556
8.127 7.375 6.788 6.312
0.0849 0.105 0.130 0.159
4.156 4.124 4.118 4.108
3.425 3.330 3.265 3.208
0.00946 0.0128 0.0168 0.0217
6.300 5.821 5.396 5.077
5.912 5.368 4.935 4.579
0.196 0.241 0.261 oo
m m
3.171 3.161 3.160 3.160
4.089
flux along the diameter. The local Nusselt n u m b e r s for these two b o u n d a r y conditions are p r e s e n t e d in Table 5.60. The terms BC1 and BC2 in Table 5.60 refer to the previously mentioned first and second b o u n d a r y conditions.
Annular Sector Ducts A n annular sector duct is displayed in the inset of Fig. 5.56. Shah and L o n d o n [1] have calculated the f R e value to a high d e g r e e of accuracy using the analytical solution p r o p o s e d by S p a r r o w et al. [285]. The results of f Re are p r e s e n t e d in Fig. 5.56. 24.0 23.0
r*
r,
22.0 21.0
ro  ~
24'=35~
20.0 19.0 fRe
18.0 17.0 16.0 15.0 14.0
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
r*
FIGURE 5.56 Fully developed friction factor for laminar flow in an annular sector duct [1].
FORCED CONVECTION, INTERNAL FLOW IN DUCTS
5.111
Schenkel [279] has developed the following approximate equation for f Re in an annular secular duct: 24 [ 0.63 (1  r*)][ 1 (lr*)] 2 (5.341) f R e = 1~ l + r . 1+~ l + r * This equation is valid for ~ > ~min(r*). The values of ~min for r* = 0, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, and 0.9 are 60 °, 50 °, 42 °, 35 °, 28.5 °, 22.5 °, 17.5°, 13°, 8.5 °, and 4 °, respectively. For ~ > ~)min(r*), the predictions of Eq. 5.341 are in excellent accord with the results presented in Fig. 5.56. BenAli et al. [281] and Soliman [286] have investigated fully developed flow and heat transfer in annular sector ducts with the 0), (~, and @ boundary conditions. The Nusselt numbers obtained by those investigators can be found in Table 5.61. Simultaneously developing flow in annular sector ducts for air (Pr = 0.7) has been analyzed by Renzoni and Prakash [287]. In their analysis, the outer curved wall is treated as adiabatic, and the ~ boundary condition is imposed on the inner curved wall as well as on the two straight walls of the sector. The fully developed friction factors, incremental pressure drop numbers, hydrodynamic entrance lengths, and thermal entrance lengths are presented in Table 5.62. The term L~y used in Table 5.62 is defined as the dimensionless axial distance at which fapp Re = 1.05f Re. The fully developed Nusselt numbers are represented by NU/d in order not to confuse the reader since the thermal boundary condition applied in Renzoni and Prakash [287] is different from those defined in the section.
StadiumShaped Ducts A stadiumshaped duct and a modified stadiumshaped duct are displayed in the insets of Fig. 5.57. Zarling [288] has obtained the f Re and Null1 for fully developed laminar flow in stadiumshaped ducts. Cheng and Jamil [289] have determined the f Re and Nu for the mod
24.0 23.0 22.0
\', \',
,
.........

21.0 20.0
•:'1
\ \\\',
i
12;.
 19.0 18.0 ./'Re
5.5
17.0
5.0
16.0 "
4.5


N
U
j
4.0
15.0 14.0
, 3.
~
!, 0.l
I ,. I , 0.2 0.3
1 ~ t, 0.4 0.5
l, 0.6
1 ~"T";"'rr13.0 0.7 0.8 0.9 1.0
a" =2M2b
FIGURE 5.57 Fully developed f Re and Num for stadiumshaped and modified stadiumshaped ducts [1].
5.112
CHAPTER FIVE
TABLE 5.61 Fully Developed Nusselt Numbers for Annular Sector Duct [281,286]
2~ = 5 °
2~ = 10 °
2~ = 15 °
2~ = 20 °
2~ = 30 °
2~ = 40 °
2~ = 50 °
2~) = 60 °
2~ = 90 °
2~ = 120 °
2~ = 150 °
2~ = 180 °
2~ = 210 °
2~ = 240 °
2~ = 270 °
2~ = 300 °
2~ = 330 °
2~ = 350 °
r* =
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Num NUll2 NUT NUll1 Num Nux Num Num NUT Num NUH2 NUT NUll1 NUH2 NUT NUHI NUH2 NUT Num NUH2 NUT NUn1 Num NUT Num NUH2 NUT Num Num NUT Num Num NUT
2.723 0.0326 1.707 2.896 0.1427 1.996 3.041 0.3145 2.208 3.163 0.6253 2.374 3.354 1.348 2.615 3.490 2.041 2.780 3.589 2.531 2.893 3.660 2.821 2.973 3.782 3.060 3.097 3.863 3.028 3.157 3.956 2.967 3.222 4.067 2.915 3.301 4.192 2.876 3.392 4.322 2.847 3.491 4.453 2.825 3.594 4.581 2.808 3.699 4.704 2.794 3.798 4.782 2.787 3.864
3.254 0.0618 2.045 3.387 0.2746 2.341 3.488 0.6250 2.541 3.564 1.083 2.684 3.656 2.000 2.868 3.697 2.603 2.966 3.711 2.901 3.015 3.715 3.025 3.038 3.761 3.060 3.086 3.893 3.003 3.180 4.079 2.953 3.322 4.284 2.916 3.480 4.490 2.888 3.641 4.685 2.868 3.807 4.867 2.853 3.961 5.034 2.842 4.107 5.186 2.833 4.241 5.279 2.827 4.327
3.850 0.1253 2.440 3.893 0.5298 2.717 3.903 1.135 2.876 3.888 1.760 2.971 3.821 2.602 3.045 3.746 2.942 3.046 3.690 3.053 3.032 3.685 3.079 3.029 3.815 3.037 3.128 4.072 2.982 3.326 4.360 2.941 3.553 4.639 2.912 3.782 4.894 2.892 3.999 5.120 2.878 4.198 5.320 2.867 4.379 5.496 2.860 4.542 5.652 2.853 4.688 5.745 2.849 4.776
4.460 0.2756 2.886 4.326 1.045 3.096 4.177 1.881 3.166 4.031 2.456 3.163 3.802 2.946 3.090 3.683 3.065 3.025 3.658 3.085 3.010 3.697 3.072 3.041 4.006 3.009 3.280 4.392 2.960 3.590 4.757 2.927 3.892 5.077 2.905 4.169 5.351 2.891 4.415 5.580 2.882 4.631 5.777 2.875 4.818 5.946 2.870 4.982 6.093 2.865 5.124 6.179 2.861 5.208
4.976 0.6420 3.355 4.554 1.849 3.405 4.207 2.575 3.316 3.951 2.887 3.194 3.686 3.065 3.029 3.640 3.086 2.997 3.712 3.070 3.053 3.844 3.046 3.157 4.351 2.979 3.562 4.836 2.937 3.966 5.239 2.913 4.322 5.565 2.899 4.621 5.835 2.890 4.874 6.045 2.882 5.080 6.225 2.879 5.263 6.377 2.875 5.413 6.510 2.871 5.540 6.583 2.870 5.620
5.243 1.426 3.773 4.471 2.603 3.523 3.997 2.964 3.249 3.747 3.051 3.077 3.627 3.088 2.988 3.743 3.066 3.079 3.948 3.035 3.241 4.183 3.005 3.426 4.855 2.954 3.987 5.383 2.916 4.457 5.778 2.901 4.828 6.080 2.891 5.118 6.334 2.886 5.352 6.505 2.881 5.539 6.660 2.879 5.700 6.788 2.875 5.825 6.904 2.869 5.926 6.960 2.869 5.993
5.101 2.419 3.962 4.098 3.015 3.337 3.709 3.111 3.051 3.616 3.129 2.980 3.785 3.062 3.113 4.106 3.069 3.366 4.446 3.040 3.642 4.763 2.957 3.910 5.510 2.916 4.576 6.010 2.900 5.055 6.357 2.892 5.396 6.611 2.885 5.651 6.845 2.880 5.843 6.956 2.874 5.989 7.078 2.867 6.099 7.178 2.866 6.205 7.279 2.866 6.295 7.311 2.867 6.347
4.466 2.922 3.634 3.675 3.077 3.026 3.641 3.085 3.000 3.834 3.056 3.153 4.371 2.993 3.582 4.875 2.953 4.008 5.290 2.930 4.337 5.624 2.916 4.686 6.299 2.896 5.342 6.700 2.887 5.744 6.964 2.876 5.997 7.150 2.868 6.175 7.368 2.868 6.323 7.394 2.869 6.437 7.479 2.871 6.528 7.548 2.874 6.601 7.638 2.877 6.664 7.638 2.878 6.699
3.648 3.083 3.005 3.889 3.049 3.197 4.462 2.987 3.662 4.979 2.947 4.009 5.726 2.912 4.782 6.208 2.898 5.250 6.537 2.889 5.573 6.775 2.883 5.812 7.208 2.868 6.238 7.443 2.869 6.488 7.590 2.874 6.646 7.691 2.881 6.775 7.884 2.887 6.834 7.820 2.894 6.894 7.864 2.900 6.941 7.899 2.905 6.979 7.976 2.909 7.010 7.945 2.911 7.705
NUll1 NUH2 NUT NUHI NUll2 NUT NUll1 Num NUT Num NUll2 NUT NUll1 NUH2 NUT Num Num NUT NUll1 NUH2 NUT
FORCED CONVECTION, INTERNAL FLOW IN DUCTS
5.113
TABLE 5.62 Fully Developed Fluid Flow and Heat Transfer Characteristics of Annular Sector Ducts with Adiabatic OuterCurved Wall [287]
r*
fRe
K(oo)
0.2 0.5 0.8
15.65 16.01 14.21
1.77 1.32 1.42
0.2 0.5 0.8
15.35 14.90 15.03
1.64 1.37 1.33
0.2 0.5 0.8
14.73 14.29 17.58
1.46 1.42 1.07
NUfd
L Sth
3.433 4.372 3.340
0.1530 0.0924 0.0898
3.493 3.933 3.113
0.1320 0.0838 0.1090
3.461 3.235 3.327
0.1070 0.0972 0.1230
t+hy
2~ = 15° 0.0775 0.0500 0.0529 2~ = 22.5° 0.0703 0.0516 0.0476 2~ = 45° 0.0574 0.0529 0.0303
ified stadiumshaped ducts. The f R e and Num for fully developed laminar flow in rectangular ducts are also included for the purpose of comparison.
MoonShaped Ducts A moonshaped duct is depicted in Fig. 5.58. Shah and London [1] have determined the fully developed f Re and the velocity profile for moonshaped ducts. These follow:
s S"
    
Z
cl (2acos0) u = ~ (r a  b z) 1  ~ r
FIGURE 5.58
¢1D2 fRe   ~ 2u.,
A moonshaped duct.
where
Dh = 2a[ (2  (x*2)~ J +( o~*)~ + 2sin 2~
Cla2 (1~13~'4+ 2CX. 2  1 ) ¢  8/'30~.3 sin ¢ + (0~'2  ~ ) sin 2 ¢  1/12 sin 4~ U m 
4
(2  (x'z), + sin 2 ,
(5.342)
(5.343)
(5.344)
(5.345)
In the preceding equations, o~*  b/a and cl = gdp/dx.
Corrugated Ducts Three corrugated ducts are schematically shown in the insets of Fig. 5.59. Hu and Chang [265] have analyzed the f Re for fully developed laminar flow in circumferentially corrugated circular ducts with n sinusoidal corrugations over the circumference as shown in Fig. 5.59, inset a, for e* = ~ a = 0.06. The perimeter and hydraulic diameter of these ducts must be evaluated numerically. However, their free flow area Ac is given by Ac = no~2(1 + 0.5e2). The f R e , Null1, and Null2 values determined by Hu and Chang [265] for circumferentially corrugated circular ducts with sinusoidal corrugations are presented in Table 5.63 as functions
5.114
CHAPTER FIVE 16
1211
/
c
10 e* = ~a
fRe
= 0.06
E
'r 2 ! 0
sin
(a) I 20
I 40
60'
(b) I 60
I 80
(c)
I 100
I 120
I 140
i 160
i 180
24,, dell
FIGURE 5.59 Fully developed friction factors for circumferentially corrugated circular ducts [2]. of n and e*, which are defined in Fig. 5.59. A n g l e 2¢ in Fig. 5.59 is related to n simply as 2~ = 360°/n. Schenkel [279] has d e t e r m i n e d the fully d e v e l o p e d friction factors for circular ducts with semicircular corrugations, as that shown in Fig. 5.59, inset b. For this kind of duct,
Ac = r~a2 sin ¢
]
sin ¢ + cos ¢~ ,
sin0
P = n2a ~
TABLE 5.63 Fully Developed Friction Factors and Nusselt Numbers for Circumferentially Corrugated Circular Ducts With Sinusoidal Corrugations [1] e*
fRe
Num
Num
Dh/2a
0.02 0.04 0.06 0.08 0.10 0.12
15.990 15.962 15.915 15.850 15.765 15.678
4.356 4.334 4.297 4.244 4.176 4.090
4.357 4.335 4.299 4.246 4.177 4.089
0.9986 0.9944 0.9874 0.9776 0.9650 0.9501
12
0.02 0.04 0.06 0.08 0.10
15.952 15.806 15.559 15.200 14.711
4.340 4.267 4.142 3.962 3.723
4.340 4.267 4.140 3.956 3.709
0.9966 0.9863 0.9689 0.9439 0.9107
16
0.02 0.04 0.06 0.08
15.887 15.542 14.943 14.051
4.316 4.168 3.912 3.540
4.316 4.167 3.906 3.527
0.9938 0.9747 0.9418 0.8934
24
0.02 0.04 0.06
15.679 14.671 12.872
4.245 3.875 3.231
4.245 3.870 3.219
0.9856 0.9402 0.8583
n
(5.346)
FORCED CONVECTION,INTERNALFLOW IN DUCTS
5.115
The radius of the semicircular corrugation is a sin ~. The f Re values for ducts with semicircular corrugations can be determined using the following expression given by Schenkel [279]: f R e = 6.4537 + 0.8350(I)  3.6909 x 102~2 + 8.6674 x 104t~3  1.0588 x 10st~4 + 6.2094 × 108(I)5  1.3261 x 104(~6 (5.347) where ~ is in degrees. Equation 5.347 is valid for 0 < 2~ < 180 °. When 2~ = 180, this geometry reduces to a circular duct. The prediction of f Re = 16 was obtained from Eq. 5.347 for circular ducts. Schenkel [279] has also determined the fully developed friction factors for laminar flow in circular ducts having triangular corrugations with an angle of 60 ° , as shown in Fig. 5.59, inset c. For this type of duct, the cross section of the fluid flow area Ac and wetted perimeter P can be calculated as follows:
Ac = rra 2 cos ~ + Vr3 sin ~
P = 4rra sin___¢_¢
(5.348)
The f Re values for ducts with triangular corrugations can be obtained with the following expression [279]: f R e = 3.8952 + 0.3692~  3.2483 x 103(~2  3.3187 x 10st~3 + 4.5962 x 10Tt~4
(5.349)
where ~ is in degrees. Equation 5.349 is valid for 0 < 2¢ < 120 °. A comparison of f Re for these three types of corrugated ducts with e* = 0.06 is displayed in Fig. 5.59.
Parallel Plate Ducts W i t h S p a n w i s e Periodic C o r r u g a t i o n s at O n e Wall
Two types of corrugations (triangular and rectangular) in parallel plate ducts are displayed in the insets of Figs. 5.60 and 5.61, respectively. Sparrow and Charmchi [290] have obtained the solutions for fully developed laminar flow in these ducts. The flow in the duct is considered to be perpendicular to the plane of the paper. Both ducts are assumed to be infinite in the span
22.5
4.5 
20.0
fRe
4.0
NUHI
3.0 15.0
Null1
fRe
2.0 20*
~
I0.0
1.o
5.0
~o 0
0.1
0.2
0.3
0.4
0.5
a/b FIGURE 5.60 Fully developed friction factors and Nusselt numbers for flat ducts with spanwiseperiodic triangular corrugations at one wall [290].
5.116
CHAPTERFIVE 40 c/d
35
'....
30
113 112 2/3
25
b/d ffi" 1
24
f Re 20 15 b/d= 1/5 b/d = 5
10
0
0
0.2
0.4
0.6
0.8
1.0
a/b
FIGURE 5.61 Fully developed friction factors and Nusselt numbers for flat ducts with spanwiseperiodicrectangular corrugations at one wall [291]. wise direction; therefore, the end effects due to the short bounding walls are neglected. The corrugated wall is subjected to the ~ thermal boundary condition, while the flat wall is considered to be adiabatic. The crosssectional area and perimeter of a flat duct with spanwise triangular corrugation can be found by: A c  n ( b 2  a 2) tan ¢,
sin ¢ P = 2 n ( b  a) 11 ++ cos
(5.350)
where n represents the number of triangular corrugations and 2¢ is the angle of the top vertex of the triangle. The fully developed f Re and Null1 values obtained by Sparrow and Charmchi [290] are shown in Fig. 5.60, which is taken from Shah and Bhatti [2]. If a/b = 0, the duct with triangular corrugations reduces to an array of isosceles triangles. The f R e and Null1 values from Fig. 5.60 agree well with the values obtained from the corresponding figures in the section concerning triangular ducts. Fully developed laminar flow and heat transfer in a parallel plate duct with spanwiseperiodic rectangular corrugations at one wall have been investigated by Sparrow and Chukaev [291]. The end effect is also ignored in their analysis. The fully developed f Re is shown in Fig. 5.61, which is based on the results reported by Sparrow and Chukaev [291] and the extension by Shah and Bhatti [2]. The heat transfer characteristics for the three pairs of geometric parameters can be found in Sparrow and Chukaev [291].
C u s p e d Ducts A c u s p e d duct, also referred to as a s t a r  s h a p e d duct, such as the one shown in Fig. 5.62, is made up of concave circular arcs. The fully developed f R e , Null1, and Nut, in laminar flow are given in Table 5.64, in which n is the number of the concave circular arcs in the cusped ducts. The values f o r f R e , Nut, and Null1 are taken from Shah and London [1], Dong et al. [292], and
FORCED CONVECTION, INTERNAL FLOW IN DUCTS
5.117
TABLE 5.64 Fully Developed fRe, Nux, and Num for Laminar Flow in Cusped Ducts n
fRe
Nux
NUll1
3 4 5 6 8
6.503 6.606 6.634 6.639 6.629
0.92 1.09 1.23 
1.352 
Dong and Ebadian [293]. An analysis of thermally developing laminar flow in cusped ducts can be found in Dong et al. [292].
Cardioid Ducts
FIGURE 5.62 A cusped duct with four concave walls.
Y.
A cardioid duct is shown in Fig. 5.63. Fully developed laminar flow and heat transfer under the (~ boundary condition have been analyzed by Tyagi [294]. The f R e and NUn1 values derived from this analysis are 5.675 and 4.208, respectively. The Nusselt number for the @ thermal boundary condition was found to be 4.097 [1].
Unusual Singly Connected Ducts ~
z
For the fully developed friction factors for laminar flow in unusual singly connected ducts, interested readers are encouraged to consult Shah and Bhatti [2].
Z r=2a(l+cosS) FIGURE 5.63 A cardioid duct.
OTHER DOUBLY CONNECTED DUCTS ,*2b o 2hi
[,
f
26o 2a o
r* 2hi 2bo
Fully developed laminar flow and heat transfer in several doubly connected ducts are discussed in the following sections.
Confocal Elliptical Ducts
FIGURE 5.64 A confocal elliptical duct.
A confocal elliptical duct is shown in Fig. 5.64. According to the analysis by Topakoglu and Arnas [295], the friction factor for fully developed laminar flow in confocal elliptical ducts can be computed by 256A 3 f R e = rClooPZ(ao+ bo)4
where
1
(mS)
_
I o o =  ~ ( 1  ~ 4) 1+~  2 m 4 1 (o2
1
1 + 0.)2 + 4 In co
(5.351)
(
(1  (o2)2 1 
m4 2 ~J
(5.352)
5.118
CHAPTER FIVE
Zc
(m4)
(ao + bo) 2 = 4 (1  032) 1 + ~
P
E
(5.353)
(m2 ]
 2 (1 + m2)E1 + 1 + ~ 03Eo~
(5.354)
03 = ( ai + bio b = °t*r* + [ 1  +°t*2(1(x*
(5.355)
ao + bo
1 (x* / 1/2 m=
1+o~*]
(x* 
'
bo ao'
r* 
bi bo
(5.356)
E1 and Eo, are the complete elliptical integrals of the second kind. These are evaluated using the a r g u m e n t s l  bo/ao 2 2 and 1  bi2]ai,2 respectively. In addition, b~/ag is related to 03 and m by m e a n s of the following: bi _ 1  (m2/032)
ai
(5.357)
1 + (m2/032)
The fully developed Nusselt n u m b e r s Null1 d e t e r m i n e d from the analysis of Topakoglu and A r n a s [295], together with the f R e calculated from Eq. 5.351, are displayed in Table 5.65.
Regular Polygonal Ducts With Centered Circular Cores The product of fully developed friction factor and Reynolds n u m b e r in laminar flow f R e obtained by R a t k o w s k y and Epstein [296] for polygonal ducts with centered circular cores (see the inset in Fig. 5.65) are shown in Fig. 5.65. The fully developed Num obtained by C h e n g and Jamil [297] are given in Fig. 5.66. It can be observed that as n ~ o% the value of f Re approaches 6.222 for a* = 1 (annular duct); f R e approaches 16 for o~* = 0 (circular duct).
Circular Ducts With Centered Regular Polygonal Cores The product of fully developed friction factors and the Reynolds n u m b e r f Re obtained by H a g e n and R a t k o w s k y [298] for laminar flow in circular ducts with centered regular polygoTABLE 5.65
The f R e and Num for fully Developed Laminar Flow in Confocal Elliptical Ducts [1]
(x* = 0.2 r* 0.02 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0.95 0.98
0.4
0.6
f Re
Num
f Re
NUll1
f Re
19.419 19.452 19.478 19.495 19.507 19.516 19.525 19.534 19.544 19.555 19.561 19.565
5.1237 5.1252 5.1230 5.1185 5.1130 5.1072 5.1016 5.0965 5.0921 5.0885 5.0788 m
19.468 19.622 19.759 19.871 19.973 20.072 20.171 20.268 20.365 20.460 20.506 20.534
5.1231 20.291 5.1395 20.622 5.1479 20.965 5.1541 21.201 5.1626 21.404 5.1751 21.585 5.1922 21.749 5.2137 21.896 5.2390 22.029 5.2676 22.148 5.2836 22.203 22.234
0.8
0.9
0.95
NUHI
f Re
Num
f Re
Num
f Re
Num
5.4782 5.5534 5.6162 5.6770 5.7441 5.7441 5.8966 5.9779 6.0597 6.1404 6.1801 
21.766 22.388 22.750 22.974 23.135 23.257 23.257 23.429 23.490 23.539 23.560 23.572
6.5083 6.7384 6.8973 7.0218 7.1325 7.1325 7.3224 7.4023 7.4724 7.5336 7.5606 
22.436 23.151 23.454 23.610 23.708 23.773 23.819 23.851 23.874 23.890 23.896 23.900
7.1933 7.5273 7.6945 7.7961 7.8696 7.8696 7.9699 8.0046 8.0320 8.0536 8.0621 m
22.620 23.366 23.643 23.777 23.855 23.903 23.934 23.953 23.966 23.973 23.975 23.976
7.4100 7.7940 7.9574 8.0427 8.0955 8.0955 8.1546 8.1711 8.1825 8.1901 8.1928
24
,''~ ~
~~ ~ ~~
~
~~
,
~~.~,~ ~y,.,,,~ . !,,~.. ._~. ~ ,. .~,,~,~', .
~
20
t
 ....~.....
~
,0 )' f Re
"
,. !,
'\
,,L
.
\
.
,o
::TI,
8
6 0.2
0
0.4
0.6
0.8
1.0
(Is F I G U R E 5.65 Fully developed friction factors for regular polygonal ducts with centered circular cores and circular ducts with centered rectangular polygonal cores [2].
:,,
6
~
.f"
.
'"
i/
//
I"
. ,i/ , 0

0.1
" ~..~.
"
l
a
ro

~..J ..~4 ~~\,, '= ~, ~'=. 0.2
0.3
~.:~ 0.4
0.5
FIGURE 5.66 Fully developed Nusselt numbers for regular polygonal ducts with centered circular cores and circular ducts with centered rectangular polygonal cores [2].
5.119
5.120
CHAPTERFIVE nal cores (see inset in Fig. 5.65) is shown in Fig. 5.65. Corresponding fully developed Null1 obtained by Cheng and Jamil [289] are depicted in Fig. 5.66. The f Re and NUll1 for concentric circular annular ducts are shown in Figs. 5.65 and 5.66 for the purpose of comparison.
FIGURE 5.67 An isosceles triangular duct with an inscribed circular core.
Isosceles Triangular Ducts With Inscribed Circular Cores
An isosceles triangular duct with an inscribed circular core is shown in Fig. 5.67. The f Re obtained for fully developed laminar flow in such a duct by Bowen [299] can be expressed in terms of ~, as follows: f R e = 12.0000  0.1605~ + 4.2883 x 103t~2  1.0566 x 104t~3 + 1.6251 x 106t~4  1.04821 x 108~5 (5.358) where ~ is in degrees.
Elliptical Ducts With Centered Circular Cores For elliptical ducts with centered circular cores, fully developed laminar flow has been analyzed by Shivakumar [300]. The f Re values are given in Table 5.66, in which o~* denotes the ratio of the length of the minor axis to the length of the major axis of the ellipse and r* is the ratio of the diameter of the circular core to the length of the minor axis.
TABLE 5.66 Fully Developed Friction Factors for Elliptical Ducts With Centered Circular Cores [300] fRe o~*
r* = 0.5
0.5 0.7 0.9
19.321 21.694 23.519
0.6 ~ 23.435
0.7 19.402 23.159
0.95
16.816
CONCLUDING REMARKS This chapter discusses forced convection in various ducts. The formulas, correlations, equations, tables, and figures included in this chapter are given for the purpose of practical calculations. However, the following effects are not considered: a detailed discussion of heat source and dissipation effects, nonnewtonian fluids, varying thermal property effects, porous wall ducts, unsteadystate effects, rotating ducts, combined radiation, and convection. The interested reader can consult Kays and Perkins [263] and Kakaq, Shah, and Aung [301] for further information regarding these effects.
NOMENCLATURE Ac
flow crosssectional area, m 2
a
radius of a circular duct, m; halflength of major axis of an elliptic duct, m; halflength of the width of a rectangular duct, m
FORCED CONVECTION, INTERNAL FLOW IN DUCTS
heDh/k
Bi
Biot number
Br
Brinkmann number for the 03 boundary condition, = ktUZm/k(Tw,m Te) Brinkmann number for the (~ boundary condition, ktu2/q"Dh
Br' b C
G D De De*
Dg Dh Dl E(m)
e* F
f lapp
f~ f~ c~( ) Gr Gr' Gz
=
5.121
=
halfspacing of a parallel plate duct, m; coil spacing, m; half length of minor axis of an elliptic duct, m; half length of height of a rectangular duct, m constant specific heat of the fluid at constant pressure, J/(kg.K) diameter of a circular cylinder, m Dean number = Re k/a/R modified Dean number = Re V'Dh/R general length hydraulic diameter of the duct = 4Ac/P, m laminar equivalent diameter, m complete elliptic integral of the second kind with argument m, which is defined by Eq. 2.252 eccentricity of the eccentric annular duct = e/(ro  ri); amplitude of the circular duct with sinusoidal corrugation = e/a a multiplicative factor entering various expressions circumferentially averaged fully developed friction factor = xw/(pu2/2) apparent Fanning friction factor = Ap*/(2x/Dh) friction factor for curved ducts = Xw/(puam/2) friction factor for straight ducts eigenfunctions Grashof number = ~ga3AT/v 2 modified Grashof number = ~ga4q"/kv
®
Graetz number = mcp/kL = p/(4DhX*) uniform wall heat flux boundary conditions
@
thermal boundary condition referring to uniform axial wall heat flux with uniform peripheral wall temperature thermal boundary condition referring to axially and circumferentially uniform wall heat flux
He
H~ H~ h
he
Ji( ) K
r(x)
conductive thermal boundary condition thermal boundary condition referring to exponential wall heat flux helical coil number length of the fin on the major axis in an elliptic duct, m length of the fin on the minor axis in an elliptic duct, m convective heat transfer coefficient, W/(m2.K) convective heat transfer coefficient for the duct exterior, W/(m2.K) Bessel functions of the first kind and orders 0 or 1 corresponding to i = 0 or 1 wall conductivity parameter = ks/kw8~
k l
incremental pressure drop number, defined by Eq. 5.5 thermal conductivity, W/(m.K) length of fin, m
l*
relative length of fins =//a
.,
~
o
g
~
8
~'~
B
..~ ~ ~ ,,
8
=:~
~
o~
~" p,
~
~~~
=
•
•
~
o
"~*
.,~
 
~'~
*
T,
*
"
.
*"
~
** ~ ~
~
~ ~
~I
~,
~0 0
~_   ~
""
~
~
~
~
~~
•
r~
~,,~°
o
0 o
~""
0
=
~
= = ~ = 
~,
=
IZL
~,
i..1o
~
8 ~ "
~*
o
o
~
~
~°
..=
,~"
~
~
E
~
,,
=
~
~
,.~
=
~
0
,Z.,'"
I
=. ,..~
~.
9.
~
~_~..~. ~
o
o
.
=
g ~
rD 0
~.
~
~
bo
" ~
o
0
0
~
~" ~ .~
., ~ ~ , ~
g
=
~
=
"~
=
~" = ' Z
~
E
~
II
~
,,
E"
~~
>
FORCED CONVECTION, INTERNAL FLOW IN DUCTS
5.123
dimensionless parameter for eccentric annular duct; thermal energy source function, rate of thermal energy generated per unit volume of the fluid, W / m
3
duct dimension, m
S
SD2/k(Tw, m Te) for 03 boundary condition;
S*
thermal energy source number,
Sk
= SDh/q" for @ boundary conditions Stark number = ewGT3Dh/k
T
fluid temperature, K
T, Tm Tw
fluid bulk mean temperature, K wall temperature at the inside duct periphery, K
Tw, m
circumferentially averaged wall temperature, K
T*w,max
dimensionless maximum wall temperature
Z~w,min
dimensionless minimum wall temperature
®
uniform wall temperature boundary condition
=
ambient fluid temperature, K,
convection boundary condition
@
radiative boundary condition
u
fluid velocity, fluid axial velocity in x direction, m/s
Um
fluid mean axial velocity, m/s
Umax
fluid maximum axial velocity for fully developed flow, rn/s
Ut
turbulent friction or shear velocity = V~Xw/p,m/s
U+
wall coordinate = u/u,, dimensionless
I/
fluid velocity component in y or r direction, m/s
W
fluid velocity component in the z or 0 direction, m/s
X
axial (streamwise) coordinate in the Cartesian or cylindrical coordinate system, m
X+
dimensionless axial coordinate for the hydrodynamic entrance region, = x,/Dh Re
X*
dimensionless axial coordinate for the thermal entrance region, = X/Dh Pe
XL
twist ratio
y
Cartesian coordinate across the flow cross section, m; distance measured from the duct wall, m
y
+
wall coordinate
=
yut/v
Y
distance of the centroid of the duct cross section measured from the base, m
Ymax
normal distance from the base to the point where umax occurs in the duct cross section, m Cartesian coordinates across the flow cross section, m; distance from the apex of a triangle, m
Greek Symbols t~
fluid thermal diffusivity = k/pcp, m2/s
tx*
aspect ratio of a rectangular channel = 2b/2a; ratio of the minor axis to the major axis of an elliptic duct, 2b/2a
13
coefficient of thermal expansion, 1/K
~n
eigenvalues
5.124
CHAPTER FIVE
r( )
gamma function
Y
dimensionless parameter defined by Eq. 5.24; ratio of heat fluxes at two walls of a parallel plate duct
8
hydrodynamic boundary layer thickness, m; thickness of a twisted tape, m
8w
duct wall thickness, m
E
distance between centers of two circles of an eccentric annular duct, m; amplitude of a circular duct with sinusoidal corrugations, m; roughness of duct wall, m
~w O
emissivity of the duct wall material; eddy diffusivity, m2/s
Om
dimensionless fluid bulk mean temperature = (Tin Tw)/(Te Tw)
dimensionless fluid temperature for the boundary condition of axially constant wall heat flux, = ( T  Te)/q,'Dflk dimensionless fluid temperature for a doubly connected duct, defined in Shah and London [1] dimensionless circumferentially averaged wall temperature (l = i for the inner wall, l = o for outer wall) for the fundamental boundary condition of kind k when the inner or outer wall ( j = i or o) is heated or cooled; dimensionless fluid bulk mean temperature if ! = m fluid bulk mean temperature for the fundamental boundary condition of kind k when the inner wall ( j = i) or outer wall ( j = o) is heated or cooled
o~
influence coefficients derived from the fundamental solutions of the second kind, = (o (2) 0(2)'~/[0(2) 0(2)` \ " m o ~ ".'io l ' \ v i i  mi )
o~
influence coefficients derived from the fundamental solutions of the second kind, = (o (2) o" o(2> (2) O(2)'~ \vmii )/(0oo.... l fluid dynamic viscosity coefficient, Pas
v
fluid kinematic viscosity coefficient = la/p, m2/s
P
fluid density, kg/m 3 StefanBoltzmann constant = 5.6697 × 108 W/(m2.K 4) wall shear stress, Pa
(k)
lj
dimensionless heat flux at a point in the flow field for the jth wall of a doubly connected duct, defined in Ref. 1 ,~(k) = q~,Dh/ dimensionless wall heat flow defined in a manner similar to ,,tj
k ( T j  Te) for k = 1, 3; = q"/q~ for k = 2, 4 (ff~Jm, T
dimensionless mean wall heat flux for boundary condition of axially constant wall temperature, = q"Dh/k( 7",,  T~)
~x,T
dimensionless local wall heat flux for boundary condition of axially constant wall temperature, = q.~'Dh/k( T~  Te)
07>, O(o'>
dimensionless heat flux at a point in the flow field for the inner or outer wall of a concentric or eccentric annular duct apex angle or halfapex angle of a duct; angle of tube curvature
d~
coefficient, defined by Eq. 5.250
Subscripts bc
thermal boundary condition center, centroid, or curved finned duct initial value at the entrance of the duct or where the heat transfer begins
FORCED CONVECTION, INTERNAL FLOW IN DUCTS
/ fd
fluid
finless
finless duct
H
® boundary condition
5.125
fully developed flow
H1
boundary condition
H2
@ boundary condition
H4
boundary condition
H5
boundary condition
hy
hydrodynamic
i
inner surface of a doubly connected duct
in
inlet
J
heated wall of a doubly connected duct, = i or o
l
laminar flow
m
mean
max
maximum
min
minimum
o
outer surface of a doubly connected duct
P
peripheral value
S
smooth, straight duct
slug
slug flow
T
03 boundary condition
T3
@ boundary condition
T4
@ boundary conditions
t
turbulent
th
thermal
X
an arbitrary section along the duct length; a local value as opposed to a m e a n value; axial
w
wall or fluid at the wall
oo
fully developed value at x =
oo
REFERENCES 1. R. K. Shah, and A. L. London, "Laminar Flow Forced Convection in Ducts," Supplement 1 to Advances in Heat Transfer, eds. T. E Irvine and J. P. Hartnett, Academic Press, New York, 1978. 2. R. K. Shah, and M. S. Bhatti, "Laminar Convection Heat Transfer in Ducts," Handbook of SinglePhase Convective Heat Transfer, eds. S. Kakaq, R. K. Shah, and W. Aung, WileyInterscience, John Wiley & Sons, New York, 1987. 3. M. S. Bhatti, "Fully Developed Temperature Distribution in a Circular Tube with Uniform Wall Temperature," unpublished paper, OwensCorning Fiberglass Corporation, Granville, Ohio, 1985. 4. M. L. Michelsen, and J. Villadsen, "The Graetz Problem with Axial Heat Conduction," Int. J. Heat Mass Transfer, (17): 13911402, 1974. 5. J. W. Ou, and K. C. Cheng, "Viscous Dissipation Effects on Thermal Entrance Heat Transfer in Laminar and Turbulent Pipe Flows with Uniform Wall Temperature," AIAA, paper no. 74743 or A S M E paper no. 74HT50, 1974.
5.126
CHAPTER FIVE
6. V. P. Tyagi, "Laminar Forced Convection of a Dissipative Fluid in a Channel," J. Heat Transfer, (88): 161169, 1966. 7. H. J. Hickman, "An Asymptotic Study of the NusseltGraetz Problem, Part 1: Large x Behavior," J. Heat Transfer, (96): 354358, 1974. 8. Y. S. Kadaner, Y. P. Rassadkin, and E. L. Spektor, "Heat Transfer in Laminar Liquid Flow through a Pipe Cooled by Radiation," Heat TransferSov. Res., (3/5): 182188, 1971. 9. S. Piva, "An Analytical Approach to Fully Developed Heating of Laminar Flows in Circular Pipes," Int. Comm. Heat Mass Transfer, (22/6): 815824, 1995. 10. R.W. Hornbeck, "Laminar Flow in the Entrance Region of a Pipe," Appl. Sci. Res., (A13): 224232, 1964. 11. R.Y. Chen, "Flow in the Entrance Region at Low Reynolds Numbers," J. Fluids Eng., (95): 153158, 1973. 12. H. L. Weissberg, "End Correction for Slow Viscous Flow through Long Tubes," Phys. Fluids, (5): 10331036, 1962. 13. J. H. Linehan, and S. R. Hirsch, "Entrance Correction for Creeping Flow in Short Tubes," J. Fluids Eng., (99): 778779, 1977. 14. L. Graetz, "Lilger die W~irmeleitungs F~ihigkeit von Fltissigkeiten (On the Thermal Conductivity of Liquids)," part 1,Ann Phys. Chem., (18): 7994, 1883; part 2,Ann. Phys. Chem., (25): 337357,1885. 15. W. Nusselt, "Die Abh~ingigkeit der W~irmtibergangszahl vonder Rohrl~inge (The Dependence of the HeatTransfer Coefficient on the Tube Length)", VDIZ, (54): 11541158, 1910. 16. J. Newman, "The Graetz Problem," The Fundamental Principles of Current Distribution and Mass Transfer in Electrochemical Cells, ed. A. J. Bard, vol. 6, Dekker, New York, pp. 187352, 1973. 17. G. M. Brown, "Heat or Mass Transfer in a Fluid in Laminar Flow in a Circular or Flat Conduit," AIChE J.," (6): 179183, 1960. 18. B. K. Larkin, "HighOrder Eigenfunctions of the Graetz Problem," AIChE J., (7): 530, 1961. 19. A. Lrvrque, "Les Lois de la Transmission de Chaleur par Convection," Ann. Mines, Mem., ser. 12, (13): 201299, 305362, 381415, 1928. 20. H. Hausen, "Dartellung des W~irmeiaberganges in Rohren durch verallgemeinerte Potenzbeziehyngen," VDIZ., Suppl. "Verfahrenstechnik," (4): 9198, 1943. 21. C. Laohakul, C. Y. Chan, K. Y. Look, and C. W. Tan, "On Approximate Solutions of the Graetz Problem with Axial Conduction," Int. J. Heat Mass Transfer, (28): 541545, 1985. 22. M. A. Ebadian, and H. Y. Zhang, "An Exact Solution of Extended Graetz Problem with Axial Heat Conduction," Int. J. Heat Mass Transfer, (32): 17091717, 1989. 23. T. V. Nguyen, "Laminar Heat Transfer for Thermally Developing Flow in Ducts," Int. J. Heat Mass Transfer, (35): 17331741, 1992. 24. D. K. Hennecke, "Heat Transfer by HagenPoiseuille Flow in the Thermal Development Region with Axial Conduction," WtirmeStoffiibertrag., (1): 177184, 1968. 25. R. Siegel, E. M. Sparrow, and T. M. Hallman, "Steady Laminar Heat Transfer in a Circular Tube with Prescribed Wall Heat Flux," Appl. Sci. Res., (A7): 386392, 1958. 26. C.J. Hsu, "Heat Transfer in a Round Tube with Sinusoidal Wall Heat Flux Distribution," AIChE J., (11): 690695, 1965. 27. H. C. Brinkmann, "Heat Effects in Capillary Flow," Appl., Sci. Res., (A2): 120124, 1951. 28. J. W. Ou, and K. C. Cheng, "Viscous Dissipation Effects on Thermal Entrance Region Heat Transfer in Pipes with Uniform Wall Heat Flux," Appl. Sci. Res., (28): 289301, 1973. 29. T. Basu, and D. N. Roy, "Laminar Heat Transfer in a Tube with Viscous Dissipation," Int. J. Heat Mass Transfer, (28): 699701, 1985. 30. C. J. Hsu, "Exact Solution to EntryRegion Laminar Heat Transfer with Axial Conduction and Boundary Conditions of the Third Kind," Chem. Eng. Sci., (23): 457468, 1968. 31. T. E Lin, K. H. Hawks, and W. Leidenfrost, "Analysis of Viscous Dissipation Effect on Thermal Entrance Heat Transfer in Laminar Pipe Flows with Convective Boundary Conditions," Wi~rmeund Stoff~bertragung, (17): 97105, 1983.
FORCED CONVECTION, INTERNAL FLOW IN DUCTS
5.12"/
32. M. K. Jensen, "Simultaneously Developing Laminar Flow in an Isothermal Circular Tube," Int. Comm. Heat Mass Transfer, (16): 811820, 1989. 33. H. H. AIAIi, and M. S. Selim, "Analysis of Laminar Flow Forced Convection Heat Transfer with Uniform Heating in the Entrance Region of a Circular Tube," Can. J. Chem. Eng., (70): 11011107, 1992. 34. M. S. Bhatti, "Limiting Laminar Heat Transfer in Circular and Flat Ducts by Analogy with Transient Heat Conduction Problems," unpublished paper, OwensCorning Fiberglass Corporation, Granville, Ohio, 1985. 35. G. Pagliarini, "Steady Laminar Heat Transfer in the Entry Region of Circular Tubes with Axial Diffusion of Heat and Momentum," Int. J. Heat Mass Transfer, (32/6): 10371052, 1989. 36. A. Barletta, "On Forced Convection in a Circular Duct with Slug Flow and Viscous Dissipation," Int. Comm. Heat Mass Transfer, (23): 6978, 1996. 37. V. Javeri, "Simultaneous Development of the Laminar Velocity and Temperature Fields in a Circular Duct for the Temperature Boundary Condition of the Third Kind," Int. J. Heat Mass Transfer, (19): 943949, 1976. 38. G. S. Barozzi, and G. Pagliarini, "A Method to Solve Conjugate Heat Transfer Problems: The Case of Fully Developed Laminar Flow in a Pipe," J. Heat Transfer, (107): 7783, 1985. 39. J. C. Kuo, and T. E Lin, "Steady Conjugate Heat Transfer in Fully Developed Laminar Pipe Flows," J. of Thermophysics and Heat Transfer, (2/3): 281283, 1988. 40. G. Pagliarini, "Conjugate Heat Transfer for Simultaneously Developing Laminar Flow in a Circular Tube," J. of Heat Transfer, (113): 763766, 1991. 41. W. Pfenniger, "Experiments with Laminar Flow in the Inlet Length of a Tube at High Reynolds Numbers With and Without Boundary Layer Suction," technical report, Northrop Aircraft Inc., Hawthorne, California, 1952. 42. L. Prandtl, "l~lber den Reibungswiderstand Strrmender Luft," Ergebnisse der Aerodynamics Versuchstalt zu GOttingen, edition 3, pp. 15, 1927. 43. J. Nikuradse, "Gesetzm/assigkeiten der turbulenten StrOmung in glatten Rohren," Forsch. Arb. Ing. Wes., (356), 1932; English translation, NASA TT FIO, 359, 1966. 44. L. Prandtl, "Neuere Ergebnisse der Turbulenzforschung," VDIZ., (77/5): 105114, 1933; English translation NACA TM 7320, 1933. 45. M. S. Bhatti, and R. K. Shah, "Turbulent and Transition Flow Convective Heat Transfer in Ducts," Handbook of SinglePhase Convective Heat Transfer, eds. S. Kaka~, R. K. Shah, and W. Aung, WileyInterscience, New York, 1987. 46. T. von K~irm~in, "Mechanische ,/i,hnlichkeit und Turbulenz," Nachr. Ges. Wiss GOttingen, Math. Phys. Klasse, (5): 5876, 1930; English translation, NACA TM 611, 1931. 47. C. Wang, "On the Velocity Distribution of Turbulent Flow in Pipes and Channels of Constant Cross Section," J. Appl. Mech., (68): A85A90, 1946. 48. H. Darcy, "Recherches Exprrimentales Relatives Aux Mouvements de l'Eau Dans Tuyauz," Mem. Pr~s. Acad. des Sci. Inst. Frances, (15): 141, 1858. 49. H. Blasius, "Das ,/i~hnlichkeitsgesetz bei Reibungsvorg/ingen in Fltissigkeiten," Forschg. Arb. Ing.Wes., (131): Berlin, 1913. 50. W. H. McAdams, Heat Transmission, 3d ed., McGrawHill, New York, 1954. 51. T. B. Drew, E. C. Koo, and W. H. McAdams, "The Friction Factor for Clean Round Pipes," Trans. AIChE, (28): 5672, 1932. 52. L. Prandtl, Fiihrrer durch die StOmungslehre, Vieweg, Braunschweig, p. 359, 1944. 53. T. von K~irm~in, "Turbulence and Skin Friction," J. Aerosp. Sci., (7): 120, 1934. 54. C. E Colebrook, "Turbulent Flow in Pipes with Particular Reference to the Transition Region between the Smooth and Rough Pipes Laws," J. Inst. Civil Eng., (11): 133156, 1939. 55. G. K. Filonenko, "Hydraulic Resistance in Pipes (in Russia), Teplonergetika, (1/4): 4044, 1954. 56. R. Techo, R. R. Tickner, and R. E. James, "An Accurate Equation for the Computation of the Friction Factor for Smooth Pipes from the Reynolds Number," J. Appl. Mech. (32): 443, 1965.
5.128
CHAPTER FIVE
57. H. Schlichting, Boundary Layer Theory, 7th ed., McGrawHill, New York, 1979. 58. L. E Moody, "Friction Factors for Pipe Flow," Trans. ASME, (66): 671684, 1944. 59. M. R. Rao, "Forced Convection Heat Transfer and Fluid Friction in Fully Developed Turbulent Flow in Smooth and Rough Tubes," Indian ASI, pp. 3.83.23, 1989. 60. J. Nikuradse, "Strrmungsgesetze in rauhen Rohren," Forsch. Arb. Ing.Wes., (361), 1933; English translation, NACA TM 1292. 61. D. J. Wood, "An Explicit Friction Factor Relationship," Civ. Eng., (36): 6061, 1966. 62. P. K. Swamee, and A. K. Jain, "Explicit Equation for PipeFlow Problems," J. Hydraulic Div. ASCE, (102): 657664, 1976. 63. S.W. Churchill, "Friction Factor Equation Spans All Fluid Flow Regimes," Chem. Eng., 9192, 1977. 64. N. H. Chen, "An Explicit Equation for Friction Factor in Pipe," Ind. Eng. Chem. Fund., (18): 296297, 1979. 65. G. E Round, "An Explicit Approximation for the Friction FactorReynolds Number Relation for Rough and Smooth Pipes," Can. J. Chem. Eng., (58): 122123, 1980. 66. D.J. Zigrang, and N. D. Sylvester, "Explicit Approximations to the Solution of Colebrook's Friction Factor Equation," AIChE J., (28): 514515, 1982. 67. S. E. Haaland, "Simple and Explicit Formulas for the Friction Factor in Turbulent Pipe Flow," J. Fluids Eng., (105): 8990, 1983. 68. T. K. Serghides, "Estimate Friction Factor Accurately," Chem. Eng., (91): 6364, 1984. 69. V. Gnielinski, "New Equations for Heat and Mass Transfer in Turbulent Pipe and Channel Flow," Int. Chem. Eng., (16): 359368, 1976. 70. P. W. Dittus, and L. M. K. Boelter, "Heat Transfer in Automobile Radiators of the Tubular Type," Univ. Calif. Pub. Eng., (2/13): 443461,1930; reprinted in Int. Comm. Heat Mass Transfer, (12): 322, 1985. 71. A. E Colburn, "A Method of Correlating Forced Convection Heat Transfer Data and a Comparison With Fluid Friction," Trans. AIChE, (19): 174210, 1933; reprinted in Int. J. Heat Mass Transfer, (7): 13591384, 1964. 72. T. von K~irm~in, "The Analogy Between Fluid Friction and Heat Transfer," Trans. ASME, (61): 705710, 1939. 73. R. E. Drexel, and W. H. McAdams, "Heat Transfer Coefficients for Air Flowing in Round Tubes, and Around Finned Cylinders," NACA A R R No. 4F28; also Wartime Report W108, 1945. 74. W. L. Friend, and A. B. Metzner, "Turbulent Heat Transfer inside Tubes and the Analogy among Heat, Mass, and Momentum Transfer," AIChE J., (4): 393402, 1958. 75. B. S. Petukhov, and V. V. Kirillov, "The Problem of Heat Exchange in the Turbulent Flow of Liquids in Tubes," (in Russian) Teploenergetika, (4/4): 6368, 1958; see also B. S. Petukhov and V. N. Popov, "Theoretical Calculation of Heat Exchange in Turbulent Flow in Tubes of an Incompressible Fluid with Variable Physical Properties," High Temp., (1/1): 6983, 1963. 76. H. Hausen, "Neue Gleichungen ftir die W~irmetibertragung bei freier oder erzwungener Stromung," Allg. Warmetchn., (9): 7579, 1959. 77. R. L. Webb, "A Critical Evaluation of Analytical Solutions and Reynolds Analogy Equations for Turbulent Heat and Mass Transfer in Short Tubes," Wiirmeund strffiibertrag., (4): 197204, 1971. 78. E. N. Sieder, and G. E. Tate, "Heat Transfer and Pressure Drop of Liquids in Tubes," Ind. Eng. Chem, (28): 14291436, 1936. 79. O. C. Sandall, O. T. Hanna, and P. R. Mazet, "A New Theoretical Formula for Turbulent Heat and Mass Transfer with Gases or Liquids in Tube Flow," Can. J. Chem. Eng., (58): 443447, 1980. 80. R. H. Notter, and C. A. Sleicher, "A Solution to the Turbulent Graetz Problem III. Fully Developed and Entry Region Heat Transfer Rates," Chem. Eng. Sci., (27): 20732093, 1972. 81. R. C. Martinelli, "Heat Transfer to Molten Metals," Trans. ASME, (69): 947959, 1947. 82. W. Nunner, "W~irmetibergang und Druckabfall in Rauhen Rohren," VDIForschungsheft 445, ser. B, (22): 539, 1956; English translation, (786), Atomic Energy Research Establishment, Harwell, United Kingdom.
FORCED CONVECTION, INTERNAL FLOW IN DUCTS
5.129
83. D. E Dipprey, and R. H. Sabersky, "Heat and Momentum Transfer in Smooth and Rough Tubes at Various Prandtl Numbers," Int. J. Heat Mass Transfer, (6): 329353, 1963. 84. R. A. Gowen, and J. W. Smith, "Turbulent Heat Transfer from Smooth and Rough Surfaces," Int. J. Heat Mass Transfer, (11): 16571673, 1968. 85. Y. Kawase, and J. J. Ulbrecht, "Turbulent Heat and Mass Transfer in Dilute Polymer Solutions," Chem. Eng. Sci., (37): 10391046, 1982. 86. Y. Kawase, and A. De, "Turbulent Heat and Mass Transfer in Newtonian and Dilute Polymer Solutions Flowing through Rough Tubes," Int. J. Heat Mass Transfer, (27): 140142, 1984. 87. W. Zhiqing, "Study on Correction Coefficients of Laminar and Turbulent Entrance Region Effect in Round Pipe," Appl. Math. Mech., (3/3): 433446, 1982. 88. A. R. Barbin, and J. B. Jones, "Turbulent Flow in the Inlet Region of a Smooth Pipe," J. Basic Eng., (85): 2934, 1963. 89. H. Latzko, "Der W~irmeiibergang and einen turbulenten Fliissigkeitsoder Gasstrom," Z. Angew. Math. Mech., (1/4): 268290, 1921; English translation, NACA TM 1068, 1944. 90. C. A. Sleicher, and M. Tribus, "Heat Transfer in a Pipe with Turbulent Flow and Arbitrary WallTemperature Distribution," 1956 Heat Transfer and Fluid Mechanics Institute, Stanford University Press, Stanford, pp. 5978, 1956. 91. H. L. Becker, "Heat Transfer in Turbulent Tube Flow," Appl. Sci. Res., ser. A, (6): 147191, 1956. 92. R. G. Deissler, "Analysis of Turbulent Heat Transfer and Flow in the Entrance Regions of Smooth Passages," NACA TN 3016, 1953. 93. E. M. Sparrow, T. M. Hallman, and R. Siegel, "Turbulent Heat Transfer in the Thermal Entrance Region of a Pipe with Uniform Heat Flux," Appl. Sci. Res., ser. A, (7): 3752, 1957. 94. J. A. Malina, and E. M. Sparrow, "VariableProperty, ConstantProperty, and Entrance Region Heat Transfer Results for Turbulent Flow of Water and Oil in a Circular Tube," Chem. Eng. Sci., (19): 953962, 1964. 95. M. A1Arabi, "Turbulent Heat Transfer in the Entrance Region of a Tube," Heat Transfer Eng., (3): 7683, 1982. 96. C.J. Chen, and J. S. Chiou, "Laminar and Turbulent Heat Transfer in the Pipe Entrance Region for Liquid Metals," Int. J. Heat Mass Transfer, (24): 11791189, 1981. 97. L. G. Genin, E. V. Kudryavtseva, Y. A. Pakhotin, and V. G. Svindov, "Temperature Fields and Heat Transfer for a Turbulent Flow of Liquid Metal on an Initial Thermal Section," Teplofiz. Vysokikh Temp., (16/6): 12431249, 1978. 98. L. M. K. Boelter, G. Young, and H. W. Iverson, "An Investigation of Aircraft Heaters XXVIImDistribution of Heat Transfer Rate in the Entrance Section of a Circular Tube," NACA TN 1451, 1948. 99. A. E Mills, "Experimental Investigation of Turbulent Heat Transfer in the Entrance Region of a Circular Conduit," J. Mech. Eng. Sci., (4): 6377, 1962. 100. S. W. Churchill, "Comprehensive Correlating Equations for Heat, Mass and Momentum Transfer in Fully Developed Flow in Smooth Tubes," Ind. Eng. Chem. Found., (16/1): 109116, 1977. 101. V. E. Kaupas, P. S. Poskas, and J. V. Vilemas, "Heat Transfer to a TransitionRange Gas Flow in a Pipe at High Heat Fluxes (2. Heat Transfer in Laminar to Turbulent Flow Transition)," Heat TransferSov. Res., (21/3): 340351, 1989. 102. N. M. Natarajan, and S. M. Lakshmanan, "Laminar Flow through Annuli: Analytical Method of Pressure Drop," Indian Chem. Eng., (5/3): 5053, 1973. 103. R. K. Shah, "A Correlation for Laminar Hydrodynamic Entry Length Solutions for Circular and Noncircular Ducts," J. Fluids Eng., (100): 177179, 1978. 104. S. Kakaq, and O. Yiicel, Laminar Flow Heat Transfer in an Annulus with Simultaneous Development of Velocity and Temperature Fields, Technical and Scientific Council of Turkey, TUBITAK, ISITEK No. 19, Ankara, Turkey, 1974. 105. N. A. V. Piercy, M. S. Hooper, and H. E Winny, "Viscous Flow through Pipes with Cores," London Edinburgh Dublin, Philos. Mag. J. Sci., (15): 647676, 1933. 106. W. Tiedt, "Berechnung des laniinaren und turbulenten Reibungswiderstandes konzentrischer und exzentrischer Ringspalte," part I, Chem.Ztg. Chem. Appar., (90): 813821, 1966: part II, Chem.Ztg. Chem. Appar., (91): 1725, 1967.
5.130
CHAPTER FIVE 107. E. Becker, "Strrmungsvorg~inge in ringfrrmigen Spalten und ihre Beziehung zum Poiseuilleschen Gesetz," Forsch. Geb. Ingenieurwes., VDI, (48): 1907. 108. K. C. Cheng, and G. J. Hwang, "Laminar Forced Convection in Eccentric Annuli," AIChE J., (14): 510512, 1968. 109. R. W. Hanks, "The LaminarTurbulent Transition for Flow in Pipes, Concentric Annuli, and Parallel Plates," AIChE J., (9): 4548, 1963. 110. J. G. Knudsen, and D. L. Katz, Fluid Dynamics and Heat Transfer, Robert E. Kneger, Huntingdon, N.Y., pp. 191193, 1979. 111. W. M. Kays, and E. Y. Leung, "Heat Transfer in Annular Passages: Hydrodynamically Developed Turbulent Flow with Arbitrarily Prescribed Heat Flux," Int. J. Heat Mass Transfer, (6): 537557, 1963. 112. O. C. Jones Jr., and J. C. M. Leung, "An Improvement in the Calculation of Turbulent Friction in Smooth Concentric Annuli," J. Fluids Eng., (103): 615623, 1981. 113. O. E. Dwyer, "Eddy Transport in Liquid Metal Heat Transfer," AIChE J., (9): 261268, 1963. 114. R. R. Rothfus, C. C. Monrad, K. G. Sikchi, and W. J. Heideger, "Isothermal Skin Friction in Flow through Annular Sections," Ind. Eng. Chem., (47): 913918, 1955. 115. R. M. Olson, and E. M. Sparrow, "Measurements of Turbulent Flow Development in Tubes and Annuli with Square or Rounded Entrances," AIChE J., (9): 766770, 1963. 116. T. H. Okiishi, and G. K. Serovy, "An Experimental Study of the Turbulent Flow Boundary Layer Development in Smooth Annuli," J. Basic Eng., (89): 823836, 1967. 117. A. Quarmby, and R. K. Anand, "Turbulent Heat Transfer in the Thermal Entrance Region of Concentric Annuli with Uniform Wall Heat Flux," Int. J. Heat Mass Transfer, (13): 395411, 1970. 118. A. Roberts and H. Barrow, "Turbulent Heat Transfer in the Thermal Entrance Region of an Internally Heated Annulus," Proc. Inst. Mech. Engrs., (182/3H): 268276, 1967. 119. V. K. Jonsson, and E. M. Sparrow, "Experiments on Turbulent Flow Phenomena in Eccentric Annular Ducts," J. Fluid Mech., (25): 6568, 1966. 120. V. K. Jonsson, "Experimental Studies of Turbulent Flow Phenomena in Eccentric Annuli," Ph.D. thesis, University of Minnesota, Minneapolis, 1965. 121. Y. Lee, and H. Barrow, "Turbulent Flow and Heat Transfer in Concentric and Eccentric Annuli," Proc. Thermodyn. and Fluid Mech. Convention, Institute of Mechanical Engineers, London, paper no. 12, 1964. 122. R. G. Deissler, and M. E Taylor, "Analysis of Fully Developed Turbulent Heat Transfer in an Annulus of Various Eccentricities," NACA TN 3451, 1955. 123. W. S. Yu, and O. E. Dwyer, "Heat Transfer to Liquid Metals Flowing Turbulently in Eccentric AnnuliI," Nucl. Sci. Eng., (24): 105117, 1966. 124. W. Ricker, J. H. T. Wade, and W. Wilson, "On the Velocity Fields in Eccentric Annuli," ASME, paper 68WA/FE35, 1968. 125. R. L. Judd, and J. H. T. Wade, "Forced Convection Heat Transfer in Eccentric Annular Passages," Heat Transfer and Fluid Mechanics Institute, Stanford University, Stanford, pp. 272288, 1963. 126. E. Y. Leung, W. M. Kays, and W. C. Reynolds, "Heat Transfer with Turbulent Flow in Concentric and Eccentric Annuli With Constant and Variable Heat Flux," TR AHT4, Mech. Eng. Dept., Stanford University, Stanford, 1962. 127. K. C. Cheng, and R. S. Wu, "Viscous Dissipation Effects on Convective Instability and Heat Transfer in Plane Poiseuille Flow Heated from Below," Appl. Sci. Res., (32), 327346, 1976. 128. J. W. Ou, and K. C. Cheng, "Effects of Pressure Work and Viscous Dissipation on Graetz Problem for Gas Flow in ParallelPlate Channels," WiirmeStoff~bertrag., (6): 191198, 1973. 129. S. Pahor, and J. Strand, "A Note on Heat Transfer in Laminar Flow through a Gap," Appl. Sci Res., (A10): 8184, 1961. 130. C. C. Grosjean, S. Pahor, and J. Strand, "Heat Transfer in Laminar Flow through a Gap," Appl. Sci. Res., (All): 292294, 1963. 131. W. Nusselt, "Der W~irmeaustausch am Berieselungsktihler," VDIZ., (67): 206210, 1923.
FORCED CONVECTION, INTERNAL FLOW IN DUCTS
5.131
132. R. D. Cess, and E. C. Shaffer, "Heat Transfer to Laminar Flow between Parallel Plates with a Prescribed Wall Heat Flux," Appl. Sci. Res., (A8): 339344, 1959. 133. E. M. Sparrow, and S. H. Lin, "Turbulent Heat Transfer in a ParallelPlate Channel," Int. J. Heat Mass Transfer, (6): 248249, 1963. 134. A. S. Jones, "TwoDimensional Adiabatic Forced Convection at Low Prclet Number," Appl. Sci. Res., (125): 337348, 1972. 135. C. J. Hsu, "An Exact Analysis of Low Prclet Number Thermal Entry Region Heat Transfer in Transversely Nonuniform Velocity Fields," A I C h E J., (17): 732740, 1971. 136. C. L. Hwang, and L. T. Fan, "Finite Difference Analysis of Forced Convection Heat Transfer in Entrance Region of a Flat Rectangular Duct," Appl. Sci. Res., (A13): 401422, 1964. 137. K. Stephan, "W~irmetibergand und druckabfall bei nicht ausgebildeter Laminarstrrmung in Rohren und in ebenen Spalten," ChemlngTech., (31): 773778, 1959. 138. M. S. Bhatti, and C. W. Savery, "Heat Transfer in the Entrance Region of a Straight Channel; Laminar Flow With Uniform Wall Temperature," J. Heat Transfer, (100): 539542, 1978. 139. R. Das, and A. K. Mohanty, "Forced Convection Heat Transfer in the Entrance Region of a Parallel Plate Channel," Int. J. Heat Mass Transfer, (26): 14031405, 1983. 140. H. S. Heaton, W. C. Reynolds, and W. M. Kays, "Heat Transfer in Annular Passages: Simultaneous Development of Velocity and Temperature Fields in Laminar Flow," Int. J. Heat Mass Transfer, (7): 763781, 1964. 141. T. V. Nguyen, "Low Reynolds Number Simultaneously Developing Flows in the Entrance Region of Parallel Plates," Int. J. Heat Mass Transfer, (34): 12191225, 1991. 142. T. V. Nguyen, and I. L. MacLaineCross, "Simultaneously Developing Laminar Flow, Forced Convection in the Entrance Region of Parallel Plates," J. Heat Transfer, (113): 837842, 1991. 143. G. S. Beavers, E. M. Sparrow, and R. A. Magnuson, "Experiments on the Breakdown of Laminar Flow in a ParallelPlate Channel," Int. J. Heat Mass Transfer, (13): 809815, 1970. 144. P. Hrycak, and R. Andrushkiw, "Calculation of Critical Reynolds Number in Round Pipes and Infinite Channels and Heat Transfer in Transition Regions," Heat Transfer 1974, (II): 183187, 1974. 145. G. S. Beavers, E. M. Sparrow, and J. R. Lloyd, "Low Reynolds Number Flow in Large Aspect Ratio Rectangular Ducts," J. Basic Eng., (93): 296299, 1971. 146. R. B. Dean, "Reynolds Number Dependence of Skin Friction and Other Bulk Flow Variables in TwoDimensional Rectangular Duct Flow," J. Fluids Eng., (100): 215223, 1978. 147. S. Kakaq, and S. Paykoc, "Analysis of Turbulent Forced Convection Heat Transfer between Parallel Plates," J. Pure Appl. Sci., (1/1): 2747, 1968. 148. L. Duchatelle, and L. Vautrey, "Determination des Coefficients de Convection d'un Alliage NaK en Ecoulement Turbulent Entre Plaques Planes Paralleles," Int. J. Heat Mass Transfer, (7): 10171031, 1964. 149. M. Sakakibara, and K. Endo, "Analysis of Heat Transfer for Turbulent Flow between Parallel Plates," Int. Chem. Eng., (18): 728733, 1976. 150. A. A. Shibani, and M. N. Ozisik, "A Solution to Heat Transfer in Turbulent Flow between Parallel Plates," Int. J. Heat Mass Transfer, (20): 565573, 1977. 151. A. P. Hatton, and A. Quarmby, "The Effect of Axially Varying and Unsymmetrical Boundary Conditions on Heat Transfer with Turbulent Flow between Parallel Plates," Int. J. Heat Mass Transfer, (6): 903914, 1963. 152. A. P. Hatton, A. Quarmby, and I. Grundy, "Further Calculations on the Heat Transfer with Turbulent Flow between Parallel Plates," Int. J. Heat Mass Transfer, (7): 817823, 1964. 153. M. Sakakibara, "Analysis of Heat Transfer in the Entrance Region with Fully Developed Turbulent Flow between Parallel PlatesThe Case of Uniform Wall Heat Flux," Mem. Fac. of Eng. Fukui Univ., (30/2): 107120, 1982. 154. M. N. Ozisik, R. M. Cotta, and W. S. Kim, "Heat Transfer in Turbulent Forced Convection between ParallelPlates," Can. J. Chem. Eng., (67): 771776, 1989. 155. S. M. Marco, and L. S. Han, "A Note on Limiting Laminar Nusselt Number in Ducts with Constant Temperature Gradient by Analogy to ThinPlate Theory," Trans. ASME, (77): 625630, 1955.
5.132
CHAPTER FIVE 156. H. E E Purday, Streamline Flow, Constable, London, 1949; same as An Introduction to the Mechanics of Viscous Flow, Dover, New York, 1949. 157. N. M. Natarajan, and S. M. Lakshmanan, "Laminar Flow in Rectangular Ducts: Prediction of Velocity Profiles and Friction Factor," Indian J. Technol., (10): 435438, 1972. 158. R. M. Curr, D. Sharma, and D. G. Tatchell, "Numerical Prediction of Some ThreeDimensional Boundary Layers in Ducts," Comput. Methods Appl. Mech. Eng., (1): 143158, 1972. 159. M. Tachibana, and Y. Iemoto, "Steady Laminar Flow in the Inlet Region of Rectangular Ducts," Bull. JSME, (24/193): 11511158, 1981. 160. P. Wibulswas, "Laminar Flow Heat Transfer in NonCircular Ducts," Ph.D. thesis, London University, London, 1966. 161. J. B. Aparecido, and R. M. Cotta, "Thermally Developing Laminar Flow inside Rectangular Ducts," Int. J. Heat Mass Transfer, (33): 341347, 1990. 162. A. R. Chandrupatla, and V. M. K. Sastri, "Laminar Forced Convection Heat Transfer of a NonNewtonian Fluid in a Square Duct," Int. J. Heat Mass Transfer, (20): 13151324, 1977. 163. A. R. Chandrupatla, and V. M. K. Sastri, "Laminar Flow and Heat Transfer to a NonNewtonian Fluid in an Entrance Region of a Square Duct with Prescribed Constant Axial Wall Heat Flux," Numer. Heat Transfer, (1): 243254, 1978. 164. S. J. Davies, and C. M. White, "An Experimental Study of the Flow of Water in Pipes of Rectangular Section," Proc. Roy. Soc., (All9): 92107, 1928. 165. J. Allen, and N. D. Grunberg, "The Resistance to the Flow of Water along Smooth Rectangular Passages and the Effect of a Slight Convergence or Divergence of the Boundaries," Philos. Mag., Ser. (7): 490502, 1937. 166. R. J. Cornish, "Flow in a Pipe of Rectangular Cross Section," Proc. Roy. Soc. London, (A120): 691700, 1928. 167. J. P. Hartnett, and C. Y. Koh, and S. T. McComas, "A Comparison of Predicted and Measured Friction Factors for Turbulent Flow through Rectangular Ducts," J. Heat Transfer, (84): 8288, 1962. 168. O. C. Jones Jr., "An Improvement in the Calculation of Turbulent Friction in Rectangular Ducts," J. Fluids Eng., (98): 173181, 1976. 169. J. E Hartnet, and T. E Irvine Jr., "Nusselt Values for Estimating Liquid Metal Heat Transfer in Noncircular Ducts," AIChE J., (3): 313317, 1957. 170. A. HajiSheikh, M. Mashena, and M. J. HajiSheikh, "Heat Transfer Coefficient in Ducts with Constant Wall Temperature," J. Heat Transfer, (105): 878883, 1983. 171. K. C. Cheng, "Laminar Forced Convection in Regular Polygonal Ducts with Uniform Peripheral Heat Flux," J. Heat Transfer, (91): 156157, 1969. 172. R. K. Shah, "Laminar Flow Friction and Forced Convection Heat Transfer in Ducts of Arbitrary Geometry," Int. J. Heat Mass Transfer, (18): 849862, 1975. 173. V. K. Migay, "Hydraulic Resistance of Triangular Channels in Laminar Flow (in Russia)," Izv. Vyssh. Uchebn. Zared. Energ., (6/5): 122124, 1963. 174. E. M. Sparrow, and A. HajiSheikh, "Laminar Heat Transfer and Pressure Drop in Isosceles Triangular, Right Triangular, and Circular Sector Ducts," J. Heat Transfer, (87): 426427, 1965. 175. M. Iqbal, A. K. Khatry, and B. D. Aggarwala, "On the Second Fundamental Problem of Combined Free and Forced Convection through Vertical NonCircular Ducts," Appl. Sci. Res., (26): 183208, 1972. 176. C. A. C. Altemani, and E. M. Sparrow, "Turbulent Heat Transfer and Fluid Flow in an Unsymmetrically Heated Triangular Duct," J. Heat Transfer, (102): 590597, 1980. 177. E C. Bandopadhayay, and C. M. Ambrose, "A Generalized Length Dimension for Noncircular Ducts," Lett. Heat Mass Transfer, (7): 323328, 1980. 178. J. Nikuradse, "Untersuchungen uber Turbulent Stromung in nicht kreisformigen Rohren," Ing.Arch., (1): 306332, 1930. 179. L. Schiller, "Uber den Str6mungswiderstand von Rohren Verschiedenen Querschnitts und Rauhigkeitsgrades," Z. Angew. Malh. Mech., (3): 213, 1923.
FORCED CONVECTION, INTERNAL FLOW IN DUCTS
5.133
180. D. A. Campbell, and H. C. Perkins, "Variable Property Turbulent Heat and Momentum Transfer for Air in a Vertical Rounded Corner Triangular Duct," Int. J. Heat Mass Transfer, (11): 10031012, 1968. 181. M. S. Bhatti, "Laminar Flow in the Entrance Region of Elliptical Ducts," J. Fluids Eng., (105): 290296, 1983. 182. L. N. Tao, "On Some Laminar ForcedConvection Problems," J. Heat Transfer, (83): 466472, 1961. 183. N. T. Dunwoody, "Thermal Results for Forced Convection through Elliptical Ducts," J. Appl. Mech., (29): 165170, 1962. 184. S. M. Richardson, "Leveque Solution for Flow in an Elliptical Duct," Letters in Heat and Mass Transfer, (7): 353362, 1980. 185. R. P. Someswara, N. C. Ramacharyulu, and V. V. G. Krishnamurty, "Laminar Forced Convection in Elliptical Ducts," Appl. Sci. Res., (21): 185193, 1969. 186. H. Barrow, and A. Roberts, "Flow and Heat Transfer in Elliptic Ducts," Heat Transfer 1970, paper no. FC 4.1, Versailles, 1970. 187. D. Cain, and J. Duffy, "An Experimental Investigation on Turbulent Flow in Elliptical Ducts," Int. J. Mech. Sci., (13): 451459, 1971. 188. W. R. Dean, "Note on the Motion of a Fluid in a Curved Pipe," Philos. Mag., ser. 7, (4): 208223, 1927. 189. W. R. Dean, "The Streamline Motion of Fluid in a Curved Pipe," Philos. Mag., ser. 7, (5/30): 673695, 1928. 190. Y. Mori, and W. Nakayama, "Study on Forced Convective Heat Transfer in Curved Pipes (lst Report, Laminar Region)," Int. J. Heat Mass Transfer, (8): 6782, 1965. 191. M. Adler, "Flow in a Curved Tube," Z. Angew. Math. Mech., (14): 257265, 1934. 192. S. V. Patankar, V. S. Pratap, and D. B. Spalding, "Prediction of Laminar Flow and Heat Transfer in Helically Coiled Pipes," J. Fluid Mech., (62/3): 539551, 1974. 193. P. S. Srinivasan, S. S. Nandapurkar, and S. S. Holland, "Friction Factors for Coils," Trans. Inst. Chem. Eng., (48): T156T161, 1970. 194. R. L. Manlapaz, and S. W. Churchill, "Fully Developed Laminar Flow in a Helically Coiled Tube of Finite Pitch," Chem. Eng. Commun., (7): 5778, 1980. 195. R. K. Shah, and S. D. Joshi, "Convective Heat Transfer in Curved Ducts," Handbook of Single Phase Convective Heat Transfer, eds. S. Kakaq, R. K. Shah, and W. Aung, Wiley Interscience, John Wiley & Sons, New York, 1987. 196. Y. Mori, and W. Nakayama, "Study on Forced Convective Heat Transfer in Curved Pipes (3rd Report, Theoretical Analysis under the Condition of Uniform Wall Temperature and Practical Formulae)," Int. J. Heat Mass Transfer, (10): 681695, 1967. 197. J. M. Tarbell, and M. R. Samuels, "Momentum and Heat Transfer in Helical Coils," Chem. Eng., j . m Lausanne (Netherlands), (5): 117127, 1973. 198. N. A. Dravid, K. A. Smith, E. W. Merrill, and P. L. T. Brian, "Effect of Secondary Fluid Motion on Laminar Flow Heat Transfer in Helically Coiled Tubes," A I C h E J., (17): 11141122, 1971. 199. M. Akiyama, and K. C. Cheng, "Laminar Forced Convection Heat Transfer in Curved Pipes with Uniform Wall Temperature," Int. J. Heat Mass Transfer, (15): 14261431, 1972. 200. C. E. Kalb, and J. D. Seader, "Fully Developed ViscousFlow Heat Transfer in Curved Circular Tubes with Uniform Wall Temperature," A I C h E J., (20): 340346, 1974. 201. R. L. Manlapaz, and S. W. Churchill, "Fully Developed Laminar Convection from a Helical Coil," Chem. Eng. Commun., (9): 185200, 1981. 202. V. Kubair, and N. R. Kuldor, "Heat Transfer to Newtonian Fluids in Spiral Coils at Constant Tube Wall Temperature in Laminar Flow," Indian Journal Tech., (3): 144146, 1965. 203. V. Kubair, and N. R. Kuldor, "Heat Transfer to Newtonian Fluids in Coiled Pipes in Laminar Flow," Int. J. Heat Mass Transfer, (9): 6375, 1966. 204. C. E. Kalb, and J. D. Seader, "Heat Mass Transfer Phenomena for Viscous Flow in Curved Circular Tubes," Int. J. Heat Mass Transfer, (15): 801817, 1972.
5.134
CHAPTER FIVE 205. N. J. Rabadi, J. C. E Chow, and H. A. Simon, "An Efficient Numerical Procedure for the Solution of Laminar Flow and Heat Transfer in Coiled Tubes," Numer. Heat Transfer, (2): 279289, 1979. 206. G. Yang, Z. E Dong, and M. A. Ebadian, "Convective Heat Transfer in a Helicoidal Pipe Heat Exchanger," J. Heat Transfer, (115): 796800, 1993. 207. G. Yang, Z. E Dong, and M. A. Ebadian, "Laminar Forced Convection in a Helicoidal Pipe with Finite Pitch," Int. J. Heat Mass Transfer, (38): 853862, 1995. 208. D. S. Austen, and H. M. Soliman, "Laminar Flow and Heat Transfer in Helically Coiled Tubes with Substantial Pitch," Experimental Thermal and Fluid Science, (1): 183194, 1988. 209. M. A. AbulHamayel, and K. J. Bell, "Heat Transfer in Helically Coiled Tubes with Laminar Flow," A S M E paper, no. 79WA/HT11, 1979. 210. C. X. Lin, E Zhang, and M. A. Ebadian, "Laminar Forced Convection in the Entrance Region of Helicoidal Pipes," Int. J. Heat and Mass Transfer In press. 211. Z. E Dong, and M. A. Ebadian, "Computer Simulation of Laminar and Turbulent Flow in Helicoidal Pipes," in Computer Simulations in Compact Heat Exchanges, B. Sunden and M. Faghri eds., Computational Mechanics Publications, Southampton, UK. In press. 212. S. Liu, and J. H. Masliyah, "Developing Convective Heat Transfer in Helicoidal Pipes with Finite Pitch," Int. J. Heat and Fluid Flow, (15/1): 6674, 1994. 213. G. W. Hogg, "The Effect of Secondary Flow on Point Heat Transfer Coefficients for Turbulent Flow inside Curved Tubes," Ph.D. thesis, University of Idaho, Moscow, ID, 1968. 214. H. Ito, "Friction Factors for Turbulent Flow in Curved Pipes," J. Basic Eng., (81): 123134, 1959. 215. B. E. Boyce, J. G. Coiller, and J. Levy, "Hold Up and Pressure Drop Measurements in the Two Phase Flow of Air Water Mixtures in Helical Coils," Cocurrent Gas Liquid Fluid, Plenum Press, London, pp. 203231, 1969. 216. G. E C. Rogers, and Y. R. Mayhew, "Heat Transfer and Pressure Loss in Helically Coiled Tubes with Turbulent Flow," Int. J. of Heat Mass Transfer, (7): 12071216, 1964. 217. S. V. Patankar, V. S. Pratap, and D. B. Spalding, "Prediction of Turbulent Flow in Curved Pipes," J. Fluid Mech., (67/3): 583595, 1975. 218. E. E Schmidt, "W~irmeiibergang und Druckverlust in Rohrschlangen," Chem. Ing. Tech., (39): 781789, 1967. 219. N. H. Pratt, "The Heat Transfer in a Reaction Tank Cooled by Means of a Coil," Trans. Inst. Chem. Eng., (25): 163180, 1947. 220. V. K. Orlov, and E A. Tselishchev, "Heat Exchange in a Spiral Coil with Turbulent Flow of Water," Thermal Eng., (translated from Teploenergetika), (11/12): 9799, 1964. 221. G. Yang, and M. A. Ebadian, "Turbulent Forced Convection in a Helicoidal Pipe with Substantial Pitch," Int. J. Heat Mass Transfer, (39): 20152022, 1996. 222. K. C. Cheng, R. C. Lin, and J. W. Ou, "Fully Developed Laminar Flow in Curved Rectangular Channels," Journal of Fluids Eng., (98): 4148, 1976. 223. K. C. Cheng, and M. Akiyama, "Laminar Forced Convection Heat Transfer in a Curved Rectangular Channel," Int. J. Heat Mass Transfer, (13): 471490, 1970. 224. Y. Mori, Y. Uchida, and T. Ukon, "Forced Convective Heat Transfer in a Curved Channel with a Square Cross Section," Int. J. Heat Mass Transfer, (14): 17871805, 1976. 225. J. A. Baylis, "Experiments on Laminar Flow in Curved Channels of Square Cross Section," J. Fluid Mech., (48/3): 417422, 1971. 226. B. Joseph, E. P. Smith, and R. J. Adler, "Numerical Treatment of Laminar Flow in a Helically Coiled Tube of Square Cross Section: Part lmStationary Helically Coiled Tubes," AIChE J., (21): 965979, 1975. 227. J. H. Masliyah, and K. Nanadakumar, "Fully Developed Laminar Flow in a Helical Tube of Finite Pitch," Chem. Eng. Commun., (29): 125138, 1984. 228. K. C. Cheng, R. C. Lin, and J. W. Ou, "Graetz Problem in Curved Square Channels," J. of Heat Transfer, (97): 244248, 1975. 229. K. C. Cheng, R. C. Lin, and J. W. Ou, "Graetz Problem in Curved Rectangular Channels with Convective Boundary ConditionmThe Effect of Secondary Flow on Liquid SolidificationFree Zone," Int. J. Heat Mass Transfer, (18): 996999, 1975.
FORCED CONVECTION, INTERNAL FLOW IN DUCTS
5.135
230. A. Butuzov, M. K. Bezrodnyy, and M. M. Pustovit, "Hydraulic Resistance and Heat Transfer in Forced Flow in Rectangular Coiled Tubes," Heat TransferSov. Res., (7/4): 8488, 1975. 231. V. Kadambi, "Heat Transfer and Pressure Drop in a Helically Coiled Rectangular Duct," A S M E paper no. 83WA/HT1, 1983. 232. R. C. Xin, A. Awwad, Z. E Dong, and M. A. Ebadian, "An Experimental Study of SinglePhase and Twophase Flow Pressure Drop in Annular Helicoidal Pipes," Int. J. of Heat and Fluid Flow, in press. 233. S. Garimella, D. E. Richards, and R. N. Christensen, "Experimental Investigation of Heat Transfer in Coiled Annular Ducts," J. of Heat Transfer, (110): 329336, 1988. 234 Z. E Dong, and M. A. Ebadian, "Numerical Analysis of Laminar Flow in Curved Elliptic Ducts," J. Fluids Eng., (113): 555562, 19~;1. 235. Z. E Dong, and M. A. Ebadian, "Thermal Developing Flow in a Curved Duct of Elliptic Cross Section," Numer. Heat Transfer, part A, (24): 197212, 1993. 236. Z. E Dong, and M. A. Ebadian, "Effects of Buoyancy on Laminar Flow in Curved Elliptic Duct," J. Heat Transfer, (114): 936943, 1992. 237. K. Rehme, "Convective Heat Transfer over Rod Bundles," Handbook of Single Phase Convective Heat Transfer, eds. S. Kakaq, R. K. Shah, and W. Aung, WileyInterscience, John Wiley & Sons, New York, 1987. 238. V. Ramachandra, "The Numerical Prediction of Flow and Heat Transfer in Rod Bundle Geometries," Ph.D. thesis, Imperial College of Science and Technology, London, 1979. 239. H. Rosenberg, "Numerical Solution of the Velocity Profile in Axial Laminar Flow through a Bank of Touching Rods in a Triangular Array," Trans. Am. Nucl. Soc., (1): 5557, 1958. 240. E. M. Sparrow, and A. L. Loeffler Jr., "Longitudinal Laminar Flow between Cylinders Arranged in Regular Array," AIChE J., (5): 325330, 1959. 241. R. A. Axford, "TwoDimensional Multiregion Analysis of Temperature Fields and Heat Fluxes in Tube Bundles with Internal Solid Nuclear Heat Sources," LA3167, Los Alamos Scientific Laboratory, Los Alamos, New Mexico, 1964. 242. R. A. Axford, "TwoDimensional Multiregional Analysis of Temperature Fields in Reactor Tube Bundles," Nucl. Eng. Design, (6): 2542, 1967. 243. E S. Shih, "Laminar Flow in Axisymmetric Conduits by a Rational Approach," Can. J. Chem. Eng., (45): 285294, 1967. 244. K. Rehme, "LaminarstrOmung in Stabbiindeln," ChemieIngenieurTechnik, (43): 962966, 1971. 245. K. Rehme, "Laminarstrrmung in Stabbtindeln," Reaktortagung 1971, Deutsches Atomforum, Bonn, Germany, pp. 130133, 1971. 246. K. Johannsen, "Longitudinal Flow over Tube Bundles," Low Reynolds Number Flow Heat Exchangers, eds. S. Kakaq, R. K. Shah, and A. E. Bergles, Hemisphere, New York, pp. 229273, 1983. 247. J. Mahik, J. Hejna, and J. Schmid, "Pressure Losses and Heat Transfer in NonCircular Channels with Hydraulically Smooth Walls," Int. J. Heal Mass Transfer, (18): 139149, 1975. 248. M. D. Mikhailov, "Finite Element Analysis of Turbulent Heat Transfer in Rod Bundles," Turbulent Forced Convection in Channels and Bundles, eds. S. Kakaq, and D. B. Spalding, (1): 250277, 1979. 249. V. I. Subbotin, M. K. Ibragimov et al., Hydrodynamics and Heat Transfer in Nuclear Power Systems, Atomizdat, Moscow, 1975. 250. O. E. Dwyer, and H. C. Berry, "Laminar Flow Heat Transfer for InLine Flow through Unbarred Rod Bundles," Nucl. Sci. Eng., (42): 8188, 1970. 251. K. Rehme, "Simple Method of Predicting Friction Factors of Turbulent Flow in NonCircular Channels," Int. J. Heat Mass Transfer, (10): 933950, 1973. 252. S. K. Cheng and N. E. Todreas, "Hydrodynamic Models and Correlations for Bare and WireWrapped Hexagonal Rod BundlesDBundle Friction Factors, Subchannel Friction Factors and Mixing Parameters," Nuclear Engineering and Design, (92): 227251, 1986. 253. E. M. Sparrow, A. L. Loeffler Jr., and H. A. Hubbard, "Heat Transfer to Longitudinal Laminar Flow between Cylinders," Trans. ASME, J. Heat Transfer, (83): 415422, 1961. 254. C. J. Hsu, "Laminar and Slug Flow Heat Transfer Characteristics of Fuel Rods Adjacent to Fuel Subassembly Walls," Nucl. Sci. Eng., (49): 398404, 1972.
5.136
CHAPTER FIVE
255. O. Miyatake, and H. Iwashita, "LaminarFlow Heat Transfer to a Fluid Flowing Axially Between Cylinders with a Uniform Surface Temperature," Int. J. Heat Mass Transfer, (33): 417425, 1990. 256. O. Miyatake, and H. Iwashita, "LaminarFlow Heat Transfer to a Fluid Flowing Axially Between Cylinders with a Uniform Wall Heat Flux," Int. J. Heat Mass Transfer, (34): 322327, 1991. 257. R. Meyder, "Solving the Conservation Equations in Fuel Rod Bundles Exposed to Parallel Flow by Means of CurvilinearOrthogonal Coordinates," J. Comp. Physics, (17): 5367, 1975. 258. J. H. Kim, "Heat Transfer in Longitudinal Laminar Flow along Circular Cylinders in Square Array," Fluid Flow and Heat Transfer over Rod or Tube Bundles, eds. S. C. Yao and P. A. Pfund, ASME, New York, pp. 155161, 1979. 259. J. Ohnemus, Wtirmeubergang und Druck verlust in einem Zenbtralkanal enies Stabbniidels in quadratischer Anordnung, Diplomarbeit, Inst. Ftir Neutrrnenphysik und Reaktortechnik, Kernforschungszentrum Karlsruhe, 1982. 260. B. CJ. Chen, T. H. Chien, W. T. Sha, and J. H. Kim, "Solution of Flow in an Infinite Square Array of Circular Tubes by Using Boundary Fitted Coordinate Systems," Numerical Grid Generation, ed. J. E Thompson, Elsevier, New York, pp. 619632, 1982. 261. R. G. Deissler, and M. E Taylor, "Analysis of Turbulent Flow and Heat Transfer in Noncircular Passages," N A S A TR31, 1959. 262. R. G. Deissler, and M. E Taylor, "Analysis of Axial Turbulent Flow and Heat Transfer through Banks of Rods or Tubes," Reactor Heat Transfer Conf., New York, TID 75299, part 1, pp. 416461, 1956. 263. W. M. Kays, and H. C. Perkins, "Forced Convection, Internal Flow in Ducts," Handbook of Heat Transfer, McGrawHill, New York, 1985. 264. M. W. Maresca, and O. E. Dwyer, "Heat Transfer to Mercury Flowing in Line through a Bundle of Circular Rods," J. Heat Transfer, (86): 180185, 1964. 265. M. H. Hu, and Y. P. Chang, "Optimization of Finned Tubes for Heat Transfer in Laminar Flow," J. Heat Transfer, (95): 332338, 1973; for numerical results, see M. H. Hu, "Flow and Thermal Analysis for Mechanically Enhanced Heat Transfer Tubes," Ph.D. thesis, State University of New York at Buffalo, 1973. 266. C. Prakash, and Y. Liu, "Analysis of Laminar Flow and Heat Transfer in the Entrance Region of an Internally Finned Circular Duct," J. Heat Transfer, (107): 8491, 1985. 267. J. H. Masliyah, and K. Nandakumar, "Heat Transfer in Internally Finned Tubes," J. Heat Transfer, (98): 257261, 1976. 268. M. K. Gangal, and B. D. Aggarwala, "Combined Free and Forced Convection in Laminar Internally Finned Square Ducts," Z. Angew. Math. Phys., (28): 8596, 1977. 269. B. D. Aggarwala, and M. K. Gangal, "Heat Transfer in Rectangular Ducts with Fins from Opposite Walls," Z. Angew. Math. Phys., (56): 253266, 1976. 270. M. K. Gangal, "Some Problems in Channel Flow," Ph.D. thesis, University of Calgary, Calgary, 1974. 271. K. Nandakumar, and J. H. Masliyah, "Fully Developed Viscous Flow in Internally Finned Tubes," Chem. Eng. J., (10): 113120, 1975. 272. R. M. Manglik, and A. E. Bergles, "Heat Transfer and Pressure Drop Correlations for TwistedTape Inserts in Isothermal Tubes: Part ILaminar Flows," J. Heat Transfer, (115): 881889, 1993. 273. R. M. Manglik, and A. E. Bergles, "Heat Transfer and Pressure Drop Correlations for TwistedTape Inserts in Isothermal Tubes: Part IITransition and Turbulent Flows," J. Heat Transfer, (115): 890896, 1993. 274. Z. E Dong, and M. A. Ebadian, "Analysis of Combined Natural and Forced Convection in Vertical Semicircular Ducts with Radial Internal Fins," Numer. Heat Transfer, part A, (27): 359372, 1995. 275. Z. E Dong, and M. A. Ebadian, "A Numerical Analysis of Thermally Developing Flow in an Elliptic Duct with Fins," Int. J. Heat Fluid Flow, (12): 166172, 1991. 276. B. Farhanieh, and B. Sunden, "Three Dimensional Laminar Flow and Heat Transfer in the Entrance Region of Trapezoidal Ducts," Int. J. Numerical Methods in Fluids, (13): 537556, 1991. 277. C. Chiranjivi, and P. S. Rao, "Laminar and Turbulent Convection Heat Transfer in a Symmetric Trapezoidal Channel," Indian Journal of Technology, (9): 416420, 1971.
FORCED CONVECTION, INTERNAL FLOW IN DUCTS
5.137
278. N. Nakamura, S. Hiraoka, and I. Yamada, "Flow and Heat Transfer of Laminar Forced Convection in Arbitrary Polygonal Ducts," Heat TransferJpn. Res., (2/4): 5663, 1974. 279. G. Schenkel, Laminar DurchstrOmte Profilkantile; Ersatzradien und Widerstansbeiwerte, FortschrittBeriche der VDI Zeitschriften, Reihe: Stromungstechnik, vol. 7, no. 62, 1981. 280. E. R. G. Eckert, and T. E Irvine Jr., "Flow in Corners of Passages with Noncircular Cross Sections," Trans. ASME, (78): 709718, 1956. 281. T. M. BenAli, H. M. Soliman, and E. K. Zariffeh, "Further Results for Laminar Heat Transfer in Annular Section and Circular Sector Ducts," J. Heat Transfer, (111): 109001093, 1989. 282. H. M. Soliman, A. A. Menis, and A. C. Trupp, "Laminar Flow in the Entrance Region of Circular Sector Ducts," J. Appl. Mech., (49): 640642, 1982. 283. E. M. Sparrow, and A. HajiSheikh, "Flow and Heat Transfer in Ducts of Arbitrary Shape with Arbitrary Thermal Boundary Conditions," J. Heat Transfer, (88): 351358, 1966. Discussion by C. E Neville, J. Heat Transfer, (91): 588589, 1969. 284. S. W. Hong, and A. E. Bergles, "Augmentation of Laminar Flow Heat Transfer in Tubes by Means of TwistedTape Inserts," tech. rep. HTL5, ISUEMIAmes 75011, Eng. Res. Inst., Iowa State University, Ames, 1974. 285. E. M. Sparrow, T. S. Chen, and V. K. Jonsson, "Laminar Flow and Pressure Drop in Internally Finned Annular Ducts," Int. J. Heat Mass Transfer, (7): 583585, 1964. 286. H. M. Soliman, "Laminar Heat Transfer in Annular Sector Ducts," J. Heat Transfer, (109): 247249, 1987. 287. E Renzoni, and C. Prakash, "Analysis of Laminar Flow and Heat Transfer in the Entrance Region of an Internally Finned Concentric Circular Annular Ducts," J. Heat Transfer, (109): 532538, 1987. 288. J. P. Zarling, "Application of SchwarzNeumann Technique to Fully Developed Laminar Heat Transfer in Noncircular Ducts," J. Heat Transfer, (99): 332335, 1977. 289. K. C. Cheng, and M. Jamil, "Laminar Flow and Heat Transfer in Circular Ducts with Diametrically Opposite Flat Sides and Ducts of Multiply Connected Cross Sections," Can. J. Chem. Eng., (48): 333334, 1970. 290. E. M. Sparrow, and M. Charmchi, "Heat Transfer and Fluid Flow Characteristics of SpanwisePeriodic Corrugated Ducts," Int. J. Heat Mass Transfer, (23): 471481, 1980. 291. E. M. Sparrow, and A. Chukaev, "ForcedConvection Heat Transfer in a Duct Having Rectangular Protuberances," Numer. Heat Transfer, (3): 149167, 1980. 292. Z. E Dong, M. A. Ebadian, and A. Campo, "Numerical Analysis of Convective Heat Transfer in the Entrance Region of Cusped Ducts," Numer. Heat Transfer, part A, (20): 459472, 1991. 293. Z. E Dong, and M. A. Ebadian, "Mixed Convection in the Cusped Duct," J. Heat Transfer, (116): 250253, 1994. 294. V. P. Tyagi, "A General NonCircular Duct Convective HeatTransfer Problem for Liquids and Gases," Int. J. Heat Mass Transfer, (9): 13211340, 1966. 295. H. C. Topakoglu, and O. A. Arnas, "Convective Heat Transfer for Steady Laminar Flow between Two Confocal Elliptical Pipes with Longitudinal Uniform Wall Temperature Gradient," Int. J. Heat Mass Transfer, (17): 14871498, 1974. 296. D. A. Ratkowsky, and N. Epstein, "Laminar Flow in Regular Polygonal Ducts with Circular Centered Cores," Can. J. Chem. Eng., (46): 2226, 1968. 297. K. C. Cheng, and M. Jamil, "Laminar Flow and Heat Transfer in Ducts of Multiply Connected Cross Sections," ASME, paper no. 67HT6, 1967. 298. S. L. Hagen, and D. A. Ratkowsky, "Laminar Flow in Cylindrical Ducts Having Regular Polygonal Shaped Cores," Can. J. Chem. Eng., (46): 387388, 1968. 299. B. D. Bowen, "Laminar Flow in UnusualShaped Ducts," B.S. thesis, University of British Columbia, Vancouver, 1967. 300. E N. Shivakumar, "Viscous Flow in Pipes Whose Cross Sections are Doubly Connected Regions," Appi. Sci. Res., (27): 355365, 1973. 301. S. Kakaq, R. K. Shah, and W. Aung, Handbook of SinglePhase Convective Heat Transfer, WileyInterscience, John Wiley & Sons, New York, 1987.
CHAPTER 6
FORCED CONVECTION, EXTERNAL FLOWS M. W. Rubesin Retired, Ames Research CentermNASA
M. Inouye Retired, Ames Research CentermNASA
R G. Parikh Boeing Commercial Airplane Group
INTRODUCTION In the current era of large electronic computers, many complex problems in convection are being solved precisely by numerical solution of equations expressing basic principles. Keen insight into the fine points of such problems, however, requires extensive parametric studies that consume computer time, and therefore the numerical approach is usually applied only to a few examples. Further, these numerical programs may occupy so much computer storage space that their use as subroutines within generalized systems studies becomes impractical. Thus there still exists a need for general formulas and data correlations that can be used in preliminary design, in systems studies where convection is only one of many inputs, in creative design where inventiveness is based on understanding the influences of the variables of a problem, and in verifying computer codes for convective heat transfer to complex bodies. This chapter provides many of these tools for the case of forced convection over simple bodies. Specifically, theoretical equations and correlations of data are presented for evaluating the local rate of heat transfer between the surface of a body and an encompassing fluid at different temperatures and in relative motion. Forced convection requires either that the fluid be pumped past the body, as for a model in a wind tunnel, or the body be propelled through the fluid, as an aircraft in the atmosphere. The methods presented apply equally to either situation when velocities are expressed relative to the body. Gravity forces are usually negligible under these conditions. Further, the contents of this chapter are confined to those conditions where the fluid behaves as a continuum. The evaluation of forced convection to bodies has become a major problem in many aspects of modern technology. A few examples of applications include the following: thermally deicing aircraft surfaces; turbine blade cooling; furnace tube bundles; and protecting highperformance aircraft, missile nose cones, and reentry bodies from intense aerodynamic heating. Formulas for evaluating convective heat transfer rates are generally established through a combination of theoretical analysis and experimentation. Analysis is almost universally based on boundary 6.1
6.2
CHAPTER SIX
layer theorythe mathematical solution of conservation equations of individual species, overall mass, momentum, and energy that are applicable to the thin region of fluid adjacent to the surfaces of bodies where the effects of shear, heat conduction, and species diffusion are controlling. Separated flows are not considered here. The experimentation involves the measurement of solid and fluid temperatures and of heat flux in a multitude of ways.
DEFINITION OF TERMS The local convective heat flux from a point on a body is often expressed through Newton's law of cooling, generalized as
q'~ =
peUe
St
(iw  ie,eff)
(6.1)
Enthalpy is used as the measure of the thermal driving potential to broaden the application of Eq. 6.1 to thermally perfect gases with temperaturedependent specific heats where
i= f cp(T) dT
(6.2)
With bodies having constant surface temperatures, at low speeds ie,eff = ie = cpTe
(6.3)
At high speeds where frictional heating takes place, ie,eff = ie + r(O)(u2/2) = iaw
(6.4)
where r(0) is the recovery factor, having approximate values for air of 0.85 and 0.9 for laminar and turbulent flow, respectively. For bodies with nonuniform surface temperature distributions, ie,eff depends not only on the conditions at the boundary layer edge but also on the distribution of the surface temperature upstream of the location being considered. For constant fluid properties, Eq. 6.1 correlates both theoretical and experimental results for a wide range of flow and temperature conditions through the single parameter Stanton number St. As will be seen in subsequent sections, this correlation is even useful when St is dependent on ie or ie,eff and the equation is nonlinear. The Reynolds analogy, defined as the ratio of the Stanton number to the local skin friction coefficient St/(cr/2 ) is a function of the Prandtl number and is extremely useful for estimating heat transfer. Pressure drop can be used to predict heat transfer in pipes, and the skin friction can be used to predict Stanton number for external flows. When mass transfer of a foreign gas occurs at a surface, an equation similar to Eq. 6.1 is employed to define the local mass flux of the species i of a binary mixture as j i w  peUeCmi(giw gie)
The Reynolds analogy can be extended to express
(6.5)
Cmi/(Q/2) as a function of the Lewis number.
TWODIMENSIONAL LAMINAR BOUNDARY LAYER Uniform FreeStream Conditions The most studied configuration for forced convection has been the "flat plate," a surface at constant pressure with a sharp leading edge. The simplicity of this configuration so facilitates
FORCED CONVECTION, EXTERNAL FLOWS
6.3
the solution of the boundary layer equations, even for a variety of surface boundary conditions, that the bulk of heuristic theoretical boundary layer research is identified with the flat plate. These results are useful because much that is learned can be extended to more realistic body shapes using computer codes and applied directly to platelike surfaces (e.g., supersonic aircraft wings or fins having wedge cross sections and attached shock waves).
Uniform Surface Temperature Governing DifferentialEquations. For a fluid as general as a gas in chemical equilibrium, the boundary layer equations for laminar flow over a flat plate are: (pu) + q7 (pv)= 0 ay pu ~x + pv ~
Ol
Ol
(6.6)
= 0y l.t
0 [ B OI
(
(6.7)
1 )0(u2/2)]
PU~x+PVffffy=~y ~rT~y +B 1   ~ r r
0y
(6.8)
with the boundary conditions x=0 x>0
y>0; y)oo; y=0;
U .~ U e
I = le l> Ie
u=0
v=0
U  U e
(6.9)
I = iw = constant
or
0i
ay 
0
for
I = i,w
The leading edge of the plate is located at x = 0. The surface boundary conditions at y = 0 reflect the assumed conditions of zero mass transfer, a prescribed uniform temperature including the case of zero heat flux and an implied condition of a smooth surface. A stream function ~ defined as
OV pU = pe ~y
pV =pe OV 0X
(6.10)
immediately satisfies Eq. 6.6. F l u i d W i t h C o n s t a n t Properties. When the density and viscosity in Eqs. 6.6 and 6.7 are constant, the velocity field is independent of the temperature field. Blasius [1] collapsed the partial differential equations (Eqs. 6.6 and 6.7) to a single ordinary differential equation by transforming the coordinate system from x and y to ~ and 11, defined as " X
"I] = # y U e / ' l ) e X
(6.11)
The stream function defined by Eq. (6.10) is expressed as V = m(;)f(q) where
m(~)
VUel)e~
and the velocity components are expressed in terms of the similarity variable 1"1 u u~=f'(rl)
v
1 [rlf'(rl)  f(rl)]
ue  2
W~ex e
(6.12)
6.4
C H A P T E R SIX
where f(rl) is the solution of the ordinary differential equation (6.13)
f'" + a//eff'= 0 with the boundary conditions 11=0
f=0,f'=0
11 ~ oo
f' ~ 1
(6.14)
The u velocity profile is shown in Fig. 6.1. The velocity ratio reaches a value of 0.99 at a boundary layer thickness of 8 X
5  ~ ~VRe/~x
(6.15)
/
J
i
f
tm
11
jJ i
v
/ " 0
~2
0.4
0.6
a8
1.0
u/us
FIGURE 6.1 Similarvelocity profile in the laminar boundary layer on a fiat plateconstant fluid properties.
The local skin friction coefficient is ci_
2
f'(0)
_ 0.332
~//peUeX/l.te
(6.16)
W/~exe
The average skin friction coefficient for the length of the plate up to x is defined as
dy l fxo cf 2 x _ ~dx
0.664 V/Rexe
(6.17)
Equation 6.17 indicates that on a fiat plate the average skin friction coefficient is equal to twice the local skin friction coefficient at the trailing edge. Experimental verification of the Blasius theory has been hindered by the difficulty in reproducing the ideal fiat plate boundary conditions in the laboratory. Whenever uniform pressure was attained and the effects of a real leading edge were accounted for, however, it was found that the preceding calculated results were always verified to within the accuracy of the experiment. Pohlhausen [2] utilized the Blasius coordinate system and velocity distribution to evaluate the convective heating processes within the constantproperty boundary layer on a flat plate. He solved two problems:
FORCED CONVECTION, EXTERNAL FLOWS
6.5
1. The convective heat transfer rate to a plate with uniform surface temperature for fluid speeds sufficiently low to make viscous dissipation negligible 2. The temperature attained by an insulated plate (zero surface heat transfer) when exposed to a highspeed stream where viscous dissipation is important The latter is the plate thermometer or adiabatic wall problem. Eckert and Drewitz [3] showed that the general problem of heat transfer to a uniformsurfacetemperature plate in constantproperty highspeed flow is merely the superposition of the two Pohlhausen solutions. For a uniformsurfacetemperature plate in a lowspeed flow (U 2 0
y>0; y=0; y~oo;
T=Te T=Tw T ~ Te
When Eq. 6.18 is transformed to the independent variables (Eq. 6.11) and the new normalized dependent variable TTe Y0(rl) = Tw Te
(6.19)
is introduced, the ordinary homogeneous differential equation that results is Yg' + ½Pr fYg = 0
(6.20)
where fis the Blasius stream function. The transformed boundary conditions are 1"1=0 g o = l } rl+~ Yo+O
(6.21)
Solutions for Pr = 0.5 and 1.0 are shown in Fig. 6.2 as solid curves. The abscissa of this figure is the thermal boundary layer thickness parameter rlH, consisting of the Blasius boundary layer similarity parameter multiplied by Pr 1/3.The close agreement of the two solid curves suggests for Pr near unity that the thermal boundary layer thickness where Y0 = 0.01 is inversely proportional to approximately Pr 1/3or ~ir
5Pr 1/3
8
V'Rexe
x
m
x
Pr 1/3
(6.22)
Thus, fluids with Pr less than unity have thermal boundary layers that are thick relative to their flow boundary layers. Conversely, fluids with Pr greater than unity have relatively thin thermal boundary layers. This latter condition suggests a particularly simple solution of Eq. 6.20 for very large Pr [4] because the temperature variations occur where the velocity distribution is still linear in 1"1 (see Fig. 6.1 for 11< 2.0). The linear velocity condition in Eq. 6.20 permits expressing Y0 explicitly in terms of 11//. The solution for this case of large Pr is shown as a dashed line in Fig. 6.2 and agrees quite well with the calculations based on the more exact velocity distributions for Pr near unity. This agreement indicates that Eq. 6.22 is applicable over a large range of Pr from values characteristic of gases to those for heavy oils. The local Stanton number found by Pohlhausen is represented very well by St= a form consistent with the parameter 1"1..
0.332Pr 2/3 ~Xe
(6.23)
6.6
CHAPTER SIX 1.0
0.8
I
0.6
\
v
Pr
i!
).9
0.4
0.5 0.2
2
3
4
5
6
r}H prl/3yVue/(VeX )
FIGURE 6.2 Temperature distributions in the laminar boundary layer on a flat plate at uniform temperatureconstant property, lowspeed flow.
The modified Reynolds analogy from Eqs. 6.16 and 6.23 is St = ~ Pr 2/3
(6.24)
The excellent agreement of this formula with the precise numerical results of Pohlhausen [2] over a large range of Pr is shown graphically in Fig. 6.4 (solid curves are the numerical results). The dashed line, labeled 1.02 P r 2/3, results from the analysis employing a linear velocity distribution throughout the boundary layer [5]. Equation 6.24 has been shown to be consistent with experimental results through a successive series of data correlations dating back to Colburn [6]. The average Stanton number up to station x is S~ = _1 x
St dx = 2St
(6.25)
For an insulated plate in highspeed flow with constant properties, Eq. 6.8 combined with Eq. 6.7 reduces to
pu ~x + pv ~ 
Pr /)yY+  Ce kaY]
(6.26)
with boundary conditions x=0 x>0
y>0; T  Te y >oo; T>Te aT y = 0; ~y  0
(6.27)
FORCED CONVECTION, EXTERNAL FLOWS
6.7
When the independent variables are transformed to the Blasius variables and a new dependent variable
TTe
(6.28)
r(]])  Re212Ce
is introduced into Eq. 6.26, the ordinary inhomogeneous equation that results is r" + ½Pr f r' = 2Pr f,,2
(6.29)
rl=0 r'=~} 11>oo r >
(6.30)
with boundary conditions
The solutions of this problem are indicated in Fig. 6.3. The temperature distributions shown are based on calculations employing exact velocity distributions for Pr near unity and a linear velocity distribution for very large Pr. These temperatures have been normalized by the temperature rise at the surface, and the abscissa is the rlH utilized in Fig. 6.2.
m.0,

x. ,...
~
08
N\Xx k
N\
0.6
i t,....., ..... "
0
0.4
0.5~
~
"
,%
0.2
0
' x
I
2
3
4 nil:
w ,., __., . _ . . . . .
5
6
7
8
9
Pr'/3Y'~/ue/(VeX)
FIGURE 6.3 Temperature profiles in the laminar boundary layer on an insulated plateconstantproperty, highspeed flow.
Figure 6.3 shows less correlation for different Pr than was exhibited for the uniformsurfacetemperature case in Fig. 6.2. The implication here is that the thermal boundary layer produced by viscous dissipation grows at a rate different from Pr 1/3 for all but the very large values of Pr. For Pr near unity, the growth factor is closer to Pr °28. The adiabatic wall temperature (recovery temperature) is given by 2 Ue Taw = Te + r(0) 2Cp
(6.31)
Figure 6.4 shows the dependence of the recovery factor r(0) on Pr as given by Refs. 2 and 5 (solid line). In the region 0.5 < Pr < 2, the formula r(0) = er 1/2
(6.32)
6.8
CHAPTER SIX
I00
1.92 P r l / 3  ~ , , . ~,. ~ "~ ~" r.,~ S
I0
I
,.,.,..c',
pr,,2
f .
p r  Z 3_./
._ % ,
OJ o
""~'A
O0
/
I
IO
100
1000
F I G U R E 6.4 Influence of Prandtl number on the recovery factor and modified Reynolds analogy for a laminar boundary layer on a flat plate.
represents the calculated values to within 1 percent. For Pr = 7, Eq. 6.32 yields results high by 5.4 percent. The dashed line labeled A in Fig. 6.4 represents an extrapolation of the exact numerical results for Pr < 15 to approach asymptotically the limiting value r(0) = 1.92Pr ~'3
(6.33)
resulting explicitly when a linear velocity distribution exists throughout the thermal boundary layer. For a uniformsurfacetemperature plate in highspeed flow, the temperature distribution within the boundary layer is expressed by a superposition of the two Pohlhausen solutions [3]. This is permissible because the energy equation (Eq. 6.8) with constant properties is linear in temperature. The general solution of the energy equation is the sum of the general solution of the homogeneous equation (Eq. 6.18) and a particular solution of the inhomogeneous equation (Eq. 6.26): . e~
T  Te = (Tw Taw)I1o(11)+ ~
r(rl)
(6.34)
Y0(rl) and r(rl) (Eqs. 6.19 and 6.28) are indicated in Figs. 6.2 and 6.3, respectively. The local heat flux is expressed as q " = Pe/peUeCp St (Tw  Taw)
(6.35)
The appropriate Stanton number is again represented by Eq. 6.23, Taw is given by Eq. 6.31, and r(0) is given by Eq. 6.32 or Fig. 6.4. Liquids With Variable Viscosity. When the temperature difference between a liquid and a surface becomes significant, it is necessary to consider the temperature dependence of the viscosity across the boundary layer. Calculations of convective heating were made [7] for a liquid whose viscosity varies as ~w
T+T~
FORCED CONVECTION,EXTERNALFLOWS
6.9
where b and Tc are constants. The boundary layer equations are solved through a transformation of independent variables identical in form with Eq. 6.11, but with the kinematic viscosity v evaluated at the surface temperature. The resulting transformed momentum equation is
f") ' (~ww
+yl f f , , =0
and the energy equation, where viscous dissipation has been neglected, is identical with Eq. 6.20, but with Pr evaluated at the wall temperature. In Ref. 7 the form of the solution requires a choice of the constant b (b = 3 in most of the examples) but avoids the necessity of predetermining an explicit value of the constant To. The skin friction and heat transfer are expressed directly in terms of the viscosity ratio across the boundary layer ~w/~te and the Prandtl number at the surface. Figure 6.5 shows the velocity distributions in a boundary layer of a liquid with Prw = 100 (e.g., sulfuric acid at room temperature). For this Prandtl number, the thermal boundary layer penetration into the liquid is much less than the flow boundary layer, and the regions where viscosity variations occur are confined close to the surface. The curve corresponding to }.tw/l.te= 1 is the Blasius solution (see Fig. 6.1). The curve labeled ~w/l.te = 0.23 corresponds to a heated surface where the low viscosity near the surface requires steeper velocity gradients to maintain a continuity of shear with the outer portion of the boundary layer. The heated freestream cases reveal the opposite effects. In general, the outer portions of the flow boundary layers act similarly to the velocity distribution of the Blasius case except for their being displaced in or out by the effects that have taken place in the thermal boundary layer.
0.8
0,6 o
OA
0.2
/
¢///
/
/ 2
3
4
5
6
y ~/Ue/(VEX)
FIGURE 6.5 Velocityprofiles in the laminar boundary layer of a liquid, Prw = 100 [7]. The temperature profiles for different Prw and ~w/~l, e a r e indicated in Fig. 6.6. Note that the curves for Prw = 10 apply equally well for greater Prandtl numbers because of the use of the thermal boundary layer thickness parameter as the abscissa (see Fig. 6.2). The effects of the viscosity variation across the boundary layer on the surface shear stress and heat flux are shown in Fig. 6.7. The shear stress is normalized by the value obtained from the Bla
6.10
CHAPTER SIX
1.0~1
0"81: ~ ~
P% >10 I
!
°.6
ii
oii°'°
0.4 I ,
i
0.2
0
I
2
3
4
5
6
Pr1/a y Vue/vwx FIGURE 6.6 Temperature profiles in the laminar boundary layer of a liquid [7].
sius solution with the same freestream properties. The heat flux is normalized by the Pohlhausen value with the viscosity and Prandtl number evaluated at the wall temperature. Note that at the higher Prandtl numbers the wall shear becomes less dependent on the fluid properties. Ideal Gases at High Temperatures. The speeds of m o d e m military aircraft, missiles, or reentry bodies are so high that the resulting recovery temperatures are several times to orders of
2r
::L o v x o 0
v~
= •
,
~~c~z~
s preceded by an unheated portion of length s, and St (x, 0) is the Stanton number on a plate with a uniform temperature over its entire length. The first term in parentheses in the enthalpy potential arises from the difference between the leadingedge enthalpy of the plate and the recovery enthalpy. The integral term accounts for the portions where continuous surface enthalpy variations occur. The last term sums over a k number of discontinuous jumps in surface enthalpy that may occur downstream of the leading edge. The terms iw(sT) and iw(S;) represent the surface enthalpy just upstream and downstream of sj where the ]th jump in enthalpy occurs. The effect of a stepwise discontinuity in surface temperature on a flat plate can be expressed as
t,x s, E (s/3411,3
St (x, 0) = 1 
(6.66)
This closedform equation was obtained through similarity solutions of the energy equation by investigators who assumed that the velocity profile in the boundary layer is linear in rio for the case of constant Ce and Prr or in y for the case of constant fluid properties [24, 25]. Note that the right side does not contain terms involving the fluid properties, a direct consequence of Ce and Prr being assumed constant throughout the boundary layer. Again, an intuitive approach to include property variations is to use the local surface enthalpy in the reference enthalpy technique for evaluating St (x, 0). The stepwise discontinuoussurfacetemperature solutions are used solely to define the functional form of the enthalpy potential appropriate to an arbitrary surface temperature. A plot of Eq. 6.66 is given in Fig. 6.12 (13p= 0 for a flat plate). The preceding theory has been verified by several experiments. For example, in Ref. 26, local heat transfer rates were measured in the presence of ramplike temperature distributions that began downstream of the leading edge (see inset in Fig. 6.13). The data shown in Fig. 6.13 agree with the theory (solid line) to within +10 percent, the estimated accuracy of the data. The dotdashed line in the figure represents the use of a local temperature potential in estimating the heat flux and yields large errors for this particular form of the surface temperature distribution. Had the leading edge been raised above the recovery temperature, the error in neglecting the variablesurfacetemperature effects would have diminished. In general, if continuous variations in the surface temperature or enthalpy are large compared to the overall driving potential, the variablesurfacetemperature methods must be utilized. For discontinuous surface temperatures, much smaller variations are important. Stepwise and Arbitrary Heat Flux Distribution. It is often necessary to evaluate the surface temperatures resulting from a prescribed heat flux distribution. The superposition of solutions yields the surface enthalpy distribution as
iw(X)  iaw= Surface With Mass Transfer.
peUe
0.207 fox q~'(x') dx" St (x, 0)x [1 (Xt]X)3'4] 213 
(6.67)

An effective method of alleviating the intense convective heating of surfaces exposed to veryhighenthalpy streams is by means of mass transfer cooling systems. The coolant is introduced, usually in gaseous form, into the hot boundary
6.20
CHAPTERSIX 2.4
r
2.2
2.0
o
/ 
1.8
.,,.... 03 tn
O~ 0.5~ ~.6
1.4
"
1.0
0
0.2
/
0.4
0.8
06
1.0
s/x
F I G U R E 6.12 Effect of a stepwise surface temperature discontinuity on the local Stanton numberNEq. 6.66 for a flat plate (13p 0 ) and Eq. 6.124 for a surface with a pressure gradient (13p> 0)
._
,6
1
I
1
I
.
,. X o~. 0
0.o~. 0.8 x
~o., •
O.6
'%
O.4
0.2
0
0.1
0.2
O.B
0.4
0.5
0.6
0.7
08
0.9
i.o
Xo/X
F I G U R E 6.13 Comparison of data and theory for a flat plate with a delayed ramp surface temperature distribution [26].
FORCED CONVECTION,EXTERNALFLOWS
6.21
layer through the surface being protected. Mass transfer cooling is particularly applicable to leading edges of wings, fins, and nose tips of aircraft; turbine blades; and reentry capsules and missile nose cones. The types of systems include transpiration, ablation, and film cooling. A transpiration cooling system is characterized by a porous surface material that remains intact while the coolant is being pumped through the pores toward the hot boundary layer. The coolant may be a gas or a liquid that changes phase within the porous surface or after it emerges from the surface. This system operates best when pore sizes are so small that the coolant leaves at the surface temperature of the porous solid. These systems are complicated in that they require coolant storage vessels, pumps, controls, distribution ducts, and filters to avoid pore clogging. It is also difficult to fabricate strong, aerodynamically smooth porous surfaces. Another drawback of these systems is that they are unstable because a clogged pore resulting from local overheating seals off the flow of coolant and causes local failure. The advantages of transpiration cooling systems are their versatility in the choice of coolant and coolant distribution, the reusability of the system, and the retention of intact aerodynamically contoured surfaces. An ablation cooling system is one where the gas entering the boundary layer has been generated by the thermal destruction of a sacrificial solid thermally protecting an underlying structure. The simplest ablation mechanism is the sublimation of a homogeneous material. More complex ablation involves the thermal degradation of composite materials such as reinforced plastics where a heatabsorbing pyrolysis occurs below the surface. The gaseous products of pyrolysis pass through a carbonaceous char, gaining additional sensible heat or chemical heat through endothermic reactions, and then pass into the boundary layer to absorb additional heat. Ablation cooling has the enormous advantage of being selfcontrolled and requiring no active elements. The disadvantages are that surface dimensions are usually altered, part of the char is eroded mechanically by shear forces rather than through heatabsorbing phase change, and often the heavy gaseous products are not as effective in blocking the incoming convective heat as light gases. Furthermore, ablation systems are generally not reusable. For shorttime applications, dimensional stability has been achieved in ablation systems by employing porous refractory metal surfaces that have been impregnated with lowermeltingtemperature metals that absorb heat by melting, vaporizing, and transpiring through the porous refractory matrix. A film cooling system is one where a surface is protected by a film of coolant introduced into the boundary layer from either a finite length of porous surface or a slot upstream. A liquid can be used as the coolant to absorb heat by vaporization as it is drawn along the surface by the main stream gas. A severe limitation on such systems is the requirement that gravity or inertial forces act in a direction that will keep the liquid film stable and against the surface. In addition, a film cooling system requires all the active elements of a transpiration cooling system. Its main advantage over the latter is in the simpler construction of limited areas of porous surfaces or slots and in localized ducting. The boundary layer behavior over a continuously transpirationcooled surface and an ablationcooled surface is generally the same, differing only as a result of the specific chemical identity of the coolant. The effects of a porous surface when the pore size is below some threshold dimension that is a small fraction of the local boundary layer thickness, and of the flow of liquid char over ablating surfaces, do not appreciably alter the behavior of the boundary layer and can be neglected in design considerations. Thus, boundary layer theory with continuous mass injection is applicable to both types of cooling systems. Further, results of experiments involving transpiration systems can be utilized in the prediction of the behavior of ablation systems. In film cooling systems, because of the discontinuities formed by slots or limited porous regions, the boundarylayer profiles at various stations along the surface are dissimilar so that prediction methods are quite complex and rely on experimental data or rather complicated numerical analyses. Uniform Surface Temperature, Air as Coolant. When the boundary layer and coolant gases are the same, the equations controlling boundary layer behavior are Eqs. 6.66.8. The mass injection at the surface simply alters the boundary conditions (Eq. 6.9) at the wall to be
6.22
C H A P T E R SIX
x>0, y=0
u = 0 , V=Vw(X) ~)i ~v = 0
I = constant = iw or i,,,, i.e.,
,
where
f(0) = 
t
2pwV_______~Wpe ue ~PeUeX~j~e .
(6.68)
As boundary layer similarity requires f(0) to be independent of x, pwVw must be proportional to x 1/2. A simple heat balance on an element of a porous surface with a transpired gas, or of a subliming surface, reveals that this distribution of gaseous injection is uniquely compatible with the requirements of a constant surface temperature. Thus, Pw = constant, and Vw  x 1~2. This mass injection distribution has practical importance because the porous surface can operate at its maximum allowable temperature everywhere, thereby minimizing the coolant required. The boundary layer equations, together with these boundary conditions, were solved in a series of similarity theories such as those described in the section on the twodimensional laminar boundary layer, beginning with solutions for constant properties and progressing to ideal gases with variable properties. A rather complete bibliography of these theories and corroborating experiments is given in Ref. 27. The assumption of constant C e ~P/~ePe proved equally useful in this problem as with the impervious plate in extending constantproperty solutions to high Mach number cases where air still behaves as an ideal gas. The results shown here are predominantly from Refs. 2830. It is found in these analyses that blowing, i.e., negative values of f(0), thickens both the flow and thermal boundary layers. In addition, the velocity profiles take on an S shape characteristic of boundary layers approaching separation (see Fig. 6.19). Separation, (3u/by)w = 0, occurs when f(0) =  1 . 2 3 8 [28]. These Sshaped profiles are less stable, thereby decreasing transition Reynolds numbers with increased blowing rates. Near the surface, blowing reduces the velocity and temperature gradients and the corresponding shear and heat transfer rates, respectively. The heat transfer does not drop as rapidly as the shear and, consequently, Reynolds analogy becomes dependent on the blowing rate. The recovery factor, r(0), also depends on the blowing rate when Pr does not equal unity. For the case of a slender body where r(0)u2/2 >> I , ,  le, the reduction in the heat flux qw is dependent on the product of the reduction in Stanton number and the recovery factor. This is shown in Fig. 6.14 by the line labeled AirAir.  
I . O

0.8
L
0.6
:g
.
0.4
_
•
I ~/20Air '~HeAir
02
,.
~ ,~
Air
0
O. I
0.2
0.3
0.4 0.5 PwV,,/(,%ue Sto)
0.6
0.7
0.8
FIGURE 6.14 Reductionof the local heat flux by surface mass transfer of a foreign gas, Tw Te < r(O)U2e/2Cp,[31].
FORCED CONVECTION,EXTERNALFLOWS
6.23
Uniform Surface Temperature, Foreign Gas as Coolant. The effectiveness of air injection in reducing convective heat flux stimulated investigations into the use of other coolants. With the introduction of a foreign species into the boundary layer, the boundary layer equations reduce to the continuity equation (Eq. 6.6), the momentum equation (Eq. 6.7), the energy equation pu~fxx+PV~yy=by
~
ffyy(ili2)(1Le)~y j+
g 1~r F
by
and the diffusion equation bK1 /)K1 b ( g L e bK1) Pu~x +Pv by  by PrF by
(6.70)
Here, PrF/Le = Sc is the Schmidt number, K1 is the coolant mass concentration, and ia and i2 are the coolant and air enthalpies, respectively. These equations are the bases for the bulk of the analyses involving foreign gas transpiration. The boundary conditions for Eqs. 6.69 and 6.70 are x=0, y>0 K~=0 x > O, y ) oo K1) O y =0 K1 = Klw The value of Klw is dependent on the blowing rate, and some hypothesis is required to establish this dependency. Most authors have assumed a zero net mass transfer of air into the surface. This requires that the air carried away from the surface by the mass motion normal to the surface be balanced by the diffusion of air toward the surface. Since/(1 = 1  K2 and J1 = J2, this balance of air motion can be expressed directly in terms of the coolant properties as Pw~)12 ( bgl I _
]lw
(6.71)
Equation 6.71 is the required relationship between the blowing rate of the coolant and its concentration at the wall. When the diffusion flux is expressed in terms of the mass transfer coefficient, the concentration of the coolant at the wall is given by 1 K~w = 1 + (peUe/PwVw)Cml
(6.72)
Note that the mass transfer coefficient is defined differently in Ref. 31. Although the total heat flux at the surface in a binary gas is composed of the sum of a conduction term and a diffusion term, the results of analyses are expressed solely in terms of the heat conduction term. The reason is that this term is equal to the heat gained by the coolant in passing from its reservoir to the surface in contact with the boundary layer. This simple heat balance is
\~y ]w= pwVw(ilw ilc)
(6.73)
where the subscript c represents the initial coolant condition in the reservoir. Further, the recovery factor is defined in terms of the surface temperature that results in zero heat conduction at the surface for a given mass transfer rate. This adiabatic condition is achieved in experiments by setting Tic = Tw for a given mass flow rate so that the coolant gains no heat. In terms of the Stanton number, the heat balance indicated by Eq. 6.73 can be expressed as
q" _ qr~ o
pwVw ilw ilc peUeSt0 iZ~wo i2~
(6.74)
6.24
CHAPTER SIX
The subscript 0 in this equation signifies zero mass transfer at the surface (f(0) = 0). The subscript 2 indicates that the enthalpy potential contributing to the heat flux by conduction alone is dependent only on the specific heat of the air and not that of the coolant. The effect of mass transfer on heat flux from slender bodies is shown in Fig. 6.14 for a variety of coolant gases. The corresponding mass transfer coefficients are shown in Fig. 6.15.
1.0
'
'
IAir 08
xXN
....
0.6 ~
~
~
~"~
J 0.4 ~
~
l
ill
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
pwvw / (peueSt0 )
FIGURE 6.15 Reduction of the local mass transfer coefficient by surface mass transfer of a foreign gas [31]. (Reprinted by permission of Pergamon Press.)
An important result from these figures is that lighter gases are more effective in reducing the transport coefficients and surface heat flux. For a range of calculations including Mach numbers as high as 12 and surface temperatures from freestream (392°R; 218 K) to recovery temperatures, the maximum departure from these correlation lines of individual solutions is +15 percent for q" and +_25 percent for Cmi. T h e bulk of the discrete numerical results lie within about onethird of these bandwidths. The maximum spread of results is obtained with the mass transfer of hydrogen, and the spread is smaller with helium and much smaller with the heavier gases. The differences in results from different calculations with the light gases are due primarily to the use of different gas properties [32]. Figures 6.14 and 6.15 are particularly useful for obtaining the mass transfer rate required in a transpiration cooling system. Usually the following quantities are specified: the coolant and its reservoir conditions, the porous surface material and its maximum allowable surface temperature, and the inviscid flow conditions outside the boundary layer. For cases where the difference between the surface temperature and the boundary layer edge temperature is small compared to the temperature rise by frictional heating, Fig. 6.14 can be used directly. For the specified conditions, the factor (ilw  i l c)/ (i ~wo  i2w) can be readily established from the thermodynamic properties of the coolant and air and i2~w0= iaw. This factor is used as the slope of a straight line that represents Eq. 6.74 and passes through the origin of Fig. 6.14. The intercept of this line with the appropriate heat blockage curve on the figure is the operating condition required to yield the specified surface temperature. The abscissa of this point yields the required local mass transfer rate. The behavior of a coolant composed of mixtures of He, Ar, Xe, and N2 injected into a nitrogen boundary layer has been analyzed in Ref. 33. Examination of the results reveals that
FORCED CONVECTION, EXTERNAL FLOWS
6.25
at a given mass transfer rate the skin friction coefficient is correlated for different coolants within +_5 percent using the mean molecular weight of the coolant. Thus, for a coolant composed of a mixture of n gases, the average molecular weight of the coolant at reservoir conditions defined as 1
M,a,, = Y.7 (K,c/M,)
(6.75)
can be used to interpolate between the curves of Figs. 6.14 and 6.15. Uniform Surface Injection. Although a mass transfer distribution yielding a uniform surface temperature is most efficient, it is much easier to construct a porous surface with a uniform mass transfer distribution. Libby and Chen [34] have considered the effects of uniform foreign gas injection on the temperature distribution of a porous flat plate. For these conditions, however, boundary layer similarity does not hold. Libby and Chen extended the work of Iglisch [35] and Lew and Fanucci [36], where direct numerical solutions of the partial differential equations were employed. An example of the nonuniform surface enthalpy and coolant concentrations resulting from these calculations is shown in Fig. 6.16.
tO, ..&
'
v
u ._
/ J
0.8
._e
/
I
x
0.6
/_
.._u t
o
0.4
.....,
i
0.2
0
/
/
I/ 0.1
0.2
Pr F = Le = I
All Di} equal
0.3
0.4
0.,5
0.6
0.7
pwvwl(PeUe)"~peue x I/u.e FIGURE 6.16 Surface temperature and coolant concentration distribution along a plate with uniform foreign gas injection, Le = Pr = Ce= 1, all binary diffusion coefficients equal [34]. (Reprinted by permission from The Physics of Fluids.)
Film Cooling With Upstream Transpiration. Film cooling systems provide protection to a surface exposed to highenthalpy streams by injecting a coolant into the hot boundary layer upstream of the surface. Injection can take place through local porous sections or slots at various angles to the surface. The coolant may be the same gas as in the boundary layer, a foreign gas, or a liquid. In the upstream porous section, more coolant is provided than is required to maintain safe surface temperatures. The excess coolant is entrained in the boundary layer close to the surface and is carried downstream, providing an insulating layer between the hot freestream gas and the surface. This layer is dissipated by laminar mixing while flowing with the boundary layer in such a way that protection is afforded over a limited distance. Because of the discontinuous nature of the surface injection, the boundary layer downstream of the
6.26
CHAPTER SIX
discontinuity is far from similar. Numerical solutions of higherorder integral equations [37] or of finite difference forms of the partial differential equations [38] have been used for evaluating the local convection to the surface. Examples of thermal protection offered by an upstream transpiration system are indicated in Fig. 6.17 (see Ref. 37). The figure shows the temperature rise that occurs on an insulated surface downstream of a transpirationcooled section for two amounts of blowing. Also indicated are the corresponding surface temperatures for the case where the upstream section was cooled internally, i.e., f(0) = 0. The quantity TwL is the upstream surface temperature in either case. The differences between the curves labeled f(0) = 0.5 and f(0) = 1 and the f(0) curve show how the presence of the transpired coolant within the downstream boundary layer distorts the temperature profiles so as to afford greater downstream protection depending on the amount of blowing.
1.0
O8
f( 0 . ~
~ '
o.5
~
~

~
~ /
0.6

"5 '~, 0.4
//;/
/
r/
0.2
0
I
2
3
4
5
x/L
F I G U R E 6.17 Effect of upstream transpiration cooling on the temperature distribution of the impervious portion of a fiat plate.
Cone in Supersonic Flow. The preceding solutions for a flat plate may be applied to a cone in supersonic flow through the Mangler transformation [39], which in its most general form relates the boundary layer flow over an arbitrary axisymmetric body to an equivalent flow over a twodimensional body. This transformation is contained in Eq. 6.89, which results in transformed axisymmetric momentum and energy equations equivalent to the twodimensional equations (Eqs. 6.95 and 6.97). Hence, solutions of these equations are applicable to either a twodimensional or an axisymmetric flow, the differences being contained solely in the coordinate transformations. For the case of an arbitrary pressure distribution, it is just as convenient to solve the axisymmetric problem directly. When the solution for the equivalent twodimensional flow already exists, however, as for a flat plate, then the results for the corresponding axisymmetric problem can be obtained by direct comparison. This correspondence exists for a cone in supersonic flow when the surface pressure is constant. Solutions of Eqs. 6.95 and 6.97 for a flat plate expressed in terms of ~ and rl may then be applied to a cone. Illustrative examples are presented in the following subsections.
FORCED CONVECTION,EXTERNALFLOWS
6.27
Uniform Surface Temperature. Transformations (see Eq. 6.89) for a flat plate (k = 0) or a cone (k = 1) become
~ PwgwUe(sinOc) 2a2 k l L
2k ]x2k+I + (6.76)
1"1=
~/(2k + 1)Ue foe
2pwg.,x
P dy
with Pw, gw, and Ueequal to constants, ~ = 0, and r = x sin 0c. The transformed momentum and energy equations (Eqs. 6.95 and 6.97) are essentially the same as Eqs. 6.37 and 6.38 for a flat l~late. For the same wall and boundary layer edge conditions, then, the solutions for f(rl) and I (rl) are the same for a flat plate or a cone. These results may be expressed in terms of physical variables as 2pwla~ fo~l Y = ~ (2k ~" 1)/./e p an
(6.77)
gw(3U~ =peUe2~(2k+ 1)ge g~Pw f"(O) Xw= \ by ]w 2peUeX ktePe
(6.78)
• • kwle / ( 2 k + l ) k t e kLwPwP(o) q(~"= k {()T~  \ ~Y ]w ~l.wCpw4 2peUeX ktePe
(6.79)
For a given value of x and the same flow properties, the boundary layer on a cone (k = 1) is thinner by a factor of l/X/3, and the surface shear stress and heat transfer are larger by a factor of V3. The local skin friction and heat transfer coefficients are related similarly: (Cf)cone
_.
(C/)flat plate
Stcone = X/~
(6.80)
Stflat plate
The local and average coefficients for a cone are related as follows: (~I) "~f cone 
(St)
4
(6.81)
"~ cone " "~
These relationships may be used to obtain cone flow results from the fiatplate results of the section on uniform freestream conditions. Realgas solutions for air obtained in this manner are given in Ref. 17. Nonuniform Surface Temperature. Transformations (Eq. 6.76) are applicable to flows with nonuniform surface temperature provided a linear viscosity law is assumed (gp = constant). The fiatplate results given previously for constant Prr may be applied to a cone with the requirement that the surface boundary conditions be the same in terms of ~. For a flat plate, ~  x, and for a cone, ~  x 3. Therefore, the flatplate results must be modified in such a way that lengths in the x direction are replaced by x 3 to obtain the cone results. For example, Eq. 6.66, which expresses the effect of a stepwise surface temperature for a fiat plate, becomes for a cone
lst( ,s)
St (x, 0) lcone= [ 1 
(xS_)9/411/3
(6.82)
Similar expressions may be derived for the effect of an arbitrary heat flux distribution. Mass Transfer, Uniform Surface Temperature. The uniformsurfacetemperature results above may be extended to include mass transfer. Similarity requires that f(0) be constant,
6.28
CHAPTER SIX
which determines the blowing distribution along the surface. The normal velocity component from Eq. 6.99 is
Vw / ( 2 k + 1)~tw Ue~/ 2pwUeX f(O)
(6.83)
For a cone, Vw X1/2as for a flat plate, but is larger by the factor ~/3 for given values of x and f(0) (thus, the blowing parameter PwVw/peUeSt0 remains unchanged). With this difference, the fiatplate results may be applied to a cone according to the equations above. For a nonsimilar blowing distribution, for example, Vw= constant, Eq. 6.83 is not applicable. Solutions to this problem may be found in Ref. 40. 
Surface With Streamwise Pressure Gradient
Gas With Uniform Elemental Composition in Chemical Equilibrium. Except for a few configurations such as a fiat plate and wedges or cones in supersonic flow, the pressure varies over the body surface as determined by inviscid flow theory. The influence of pressure gradients on forced convection in laminar boundary layers is presented in this section. Axisymmetric bodies at zero angle of attack and yawed cylinders of infinite length will be treated together to illustrate their close relationship (see Fig. 6.18). Because boundary layer theory requires negligible pressure gradients across the boundary layer, the techniques presented here apply only to those bodies whose local surface radius of curvature (1, in Fig. 6.18) is large compared with the boundary layer thickness, thereby minimizing centrifugal forces. Governing Differential Equations. In the absence of foreign gas injection, the boundary layer can be considered to have uniform elemental composition and be in chemical equilibrium, and is governed by the following equations: 3 (purk) +
ax Ou ~x + ov ay 
a
7y
dx +~y
pu ~x + pv igy  igy lt
pu ~x + pv ~ y = ffyy
(pvr k) = 0
(6.84)
~t
(6.85) (for k = 0 only)
ffyy+ g 1 
~y
(6.86)
2
where k = 1 for the axisymmetric body and k = 0 for the yawed cylinder.
V
U
X
Voo Uoo
(a) k= 1
i
/
(b)k:O
FIGURE 6.18 Sketch of coordinates employed for related twodimensional flows. (a) axisymmetric body; (b) yawed cylinder of infinite length.
(6.87)
FORCED CONVECTION, EXTERNAL FLOWS
6.29
The boundary conditions are x = O, y > 0
u = Ue(O) I=Ie
for
W=We=V=sinA x > O, y ) oo
k=0
u ) Ue(X) I~Ie
for
W~We=V=sinA
y=0
>
k=0
(6.88)
u=w=0 v = 0 for an impervious surface or vw(x) with transpiration I=iw(X)
for
q"¢0
bI by0
or
for
I=iaw(X)
The yaw angle A is the complement of the angle between the freestream direction and the cylinder axis (see Fig. 6.18). Transformation o f Variables and Equations. The extensions of transformations Eq. 6.36 to include the effects of pressure gradients are 1 fox
{ r ~ 2k
; = ~t~r pw~twUe~~}
Ue(F/L)k fox
Ti = ~trV/2(;_~ )
p
dx = ;(x)
(6.89)
dy
where Br is a reference viscosity and_ L is a reference length introduced to make ~ and 11 dimensionless [16, 41]. The function ~(~) is yet to be determined and is a key element in the extension of similarity solutions to flows where the inviscid boundary conditions do not permit boundary layer similarity. (Note the change in symbols employed here from those of Refs. 16 and 41.) The dependent variables are defined as u
fn  Ue(X)

I
I  le
w
~  We
(for k = 0 only)
(6.90)
Additional parameters that enter the equations are i t = 7 = I  (1  ts)~ 2  (ts  te)f~
(6.91)
le
2
We
where
ts = 1  ~
and
te = 1 
21e 2 U 2 d W e
2/e
(6.92)
(6.93)
The pressure gradient parameter is defined as 2 ( ~  ~) t, due ~p=Ue te d~ Both te and I]e are functions of x through their dependence on Ue(X).
(6.94)
6.30
CHAPTER SiX
The transformed partial differential equations of momentum and energy are
+ 2(~ (Cw~)~ + a   ~
f~
= 2(~  ~ ) ( f ~ 
~)(fnfq;f;fnq)
f~)
(6.95) (6.96)
  )fln{Cw(k1)[(tste)(f2)rldt'(lts)(W2)Tl]lr I + 2(~
~)(f,a[;f;La)
(6.97)
The boundary conditions for zero mass transfer are rl=0
f(¢. o)= y,(;. o)= ~(~. o)= o [(~, O) = iw(~) = i~(~)
and rl~oo

In(~,0)=0
for q : , 0 (6.98)
pp
forqw =0
fn(~, oo) = ~(~, oo) = [(~, oo) ~ 1
With mass transfer, f(~, 0) depends on
vw(x) as follows:
Ue(X )  ~l,r \ L ]
L
v~
(6.99)
+
Similar Solutions Similarity Criteria and Reduced Equations.
The partial differential equations (Eqs. 6.956.97) are not amenable to solution except by numerical methods utilizing highspeed computers. Considerable simplifications can result, as in the case of the flat plate, if these equations are reduced to ordinary differential equations through the similarity concept where the dependent variables f, ~, and [ are assumed functions of rl alone. Equations 6.956.97 become, for ~ = constant
(c~f,,),+Tf,,=~ '~~+ ~,
~
~
(6.100)
(CwW)' +fW= 0 (~rrf')'+ff '
(6.101)
l)[(t,t~)(f'2)'+(lt,)(~2) "]
{ C ~ ( ~ rr 
_
(6.102)
Ip
Consistent with the similarity assumption, none of the terms that appear in these equations or in_the related boundary conditions can be dependent on x or ~; that is, ~p, t~, and Iw, as well as ~, must be constant. The parameter t~, defined by Eq. 6.93, however, violates this requirement when Uevaries with x. The similarity assumption is also violated by the terms that contain t~ explicitly in Eqs. 6.100 and 6.102 and by the gas properties Cw, Prr, pe/p, which can be expressed in terms of t, which in turn depends on t~ through Eq. 6.91. Consequently, exact similar solutions are not possible under general stagnation region flow conditions. Exact similar solutions are possible for stagnation regions where Ue = 0 and t~ is a constant and equal to unity for an axisymmetric body and to t, for a yawed cylinder. The terms involving t~ drop out of Eqs. 6.1006.102, and similarity occurs for constant iw and ~p.
FORCED CONVECTION, EXTERNAL FLOWS
6.31
For similar flows, the pressure gradient parameter expressed as follows must be constant: { r ~2k
2 is
fox
/
~ P  Ue te dx
r ~2k
(6.103)
pw~l,wtle~~)
In a stagnation region, the fluid properties are nearly constant, and Ue X; also r  x. Thus, 13p= 1A for an axisymmetric body, and 13p= 1 for a yawed cylinder. The skin friction coefficient and Stanton number are defined under the conditions of similarity on axisymmetric and twodimensional bodies as follows. The components of the shear stress in the xi direction are given by
gwPwtsUedue () ui "r'wi=
dx ~
~pte
(6.104)
w
where the subscript i = 1 or 3 represents the x or z direction, respectively. The skin friction coefficient is defined as Cfn m
T,vn
2
Cfl~pwUeX_
with
2
(6.105)
pwUeUie ~tsxdue
~1.w
(6.106)
~ptel, le dx f~vt
Cf3 /pwUeX ~ ts X dUe~P 2 v
~tw
(6..107)
~deUe dx
Alternative forms of these equations that are sometimes more convenient are Cf1 .t / pwUeXeff
2 V
ktw
1  V ~ f'w
Cf34/ pwUeXeff 2 V
gw
(6.108)
1
 ~/2 ~'w
(6.109)
F ~2k
foXpw~l.wUe~)
where
xeff=
dx
{ r ~2k
(6.110)
pw~l,wUe~~) The corresponding Stanton number expressions for a surface at constant temperature are
st~pwUeX ~ is x due 1 ~l'w
St

pwUeXeff ~tw
["
~pte Ue dx Prow f ~  I o w
(6.111)
w
1 1 I"  V ~ Pr~w L  [ ~ w
(6.112)
Uniform Surface Temperature, Ideal Gas With ViscosityDensity Product and Prandtl Number Equal to Unity. For the case of an ideal gas with Cw = Pr = 1, similarity is possible away from the stagnation region of a body. Equations 6.100 and 6.102 for an axisymmetric body or a cylinder normal to the free stream reduce to
6.32
CHAPTER SIX
when
Ue
f,,, +ff,,= ~jp(f,2_ I )
(6.113)
[" + fl'= 0
(6.114)
satisfies Ue
~e
(6.115)
 A~f~d2
These equations are equivalent to those solved in Ref. 42 for a uniform surface temperature. Examples of the extensive solutions in Refs. 16 and 42 are presented in this section. Figure 6.19 shows the velocity distribution in the boundary layer of an axisymmetric body or an unswept cylinder for a cold wall (Iw = 0). (Note that the value of 11 used here is a factor of V ~ smaller than the one employed in sections on the flat plate.) An accelerating free stream (13p > 0, Ue increasing) reduces the flow boundary layer thickness and increases the velocity gradient rather uniformly through the boundary layer. A decelerating free stream (13p< 0) thickens the flow boundary layer rather severely and causes the velocity distribution to acquire an S shape. Eventually, the boundary layer will separate; that is, f g = 0. For the boundary conditions of Fig. 6.19, two solutions are possible for negative values of 13pnear separation. It is argued in Ref. 42 that true similarity with negative 13p cannot occur physically because Eq. 6.115 would require Ue ~ oo as ~ ~ 0. Thus, similar solutions with negative 13pcan only be approached after a period of nonsimilar flow, and depending on the conditions, one or the other of the similar solutions for a given 13pcan be attained. In Ref. 42, experimental evidence from Ref. 43 for turbulent flow is cited for the possible existence of doublevalued flowfield behavior. The velocity profiles shown are characteristic of those for either a cooled surface or one at the total enthalpy of the fluid. For a heated surface and 13p> 0, it is possible for the velocity ratio f ' in the outer portion of the boundary layer to attain a value greater than unity before approaching unity at the outer edge. The physical reason for this is that the acceleration of lowerdensity fluid by the favorable pressure gradient exceeds the retardation by the viscous forces.
~
1.0
y
01
0.6
/
0.4
i
~
" 0.3884
0.2
0
I
2
3
4
5
6
{
FIGURE 6.19 Similarvelocity distributions for body with surface pressure gradient 13p defined by Eq. 6.103, iw = 0, ?s= 1 [42].
FORCED CONVECTION, EXTERNAL FLOWS
6.33
The total enthalpy distribution in terms of the velocity is shown in Fig. 6.20 for lw = 0 and ts = 1. A pressure gradient can cause significant departures from the Crocco relationship (Eq. 6.40, represented by the straight line labeled 13p= 0). The latter is often used for approximate calculations even when pressure gradients are present. I'0
"
"o.36
0.3884 ~
0
0.2
0.4
0 .6
o.3657
Ol 8
i'
F I G U R E 6.20 Enthalpy and velocity relationship within similar boundary layers on a body with pressure gradient 13pdefined by Eq. 6.103, iw = 0, ?s = 1 [42].
Figure 6.21 shows the wall shear parameter f " required to evaluate the local skin friction coefficients by Eq. 6.106 or 6.108. These curves apply for the case where ts = 1. The doublevalued nature of f " for a cooled surface (I w= 0) for 13pnear separation ( f " = 0) is evident. Generally, f'w' is more sensitive to variations in 13pfor a hot surface. In fact, for cold wall conditions (lw = 0), the variation of f " with 13pfor [3p > 0 is quite modest. Also, a cooled surface tends to retard separation; that is, f " = 0 at a smaller value of ~p.
j
2.5
2o
L2
I.(3
0.5
05
,,
/
J
,.
.
0
0.,~
1.0
1.5
2.0
Bp
2.5
3.0
3.5
4.0
R G U R E 6.21 Effect of pressure gradient on the skin friction parameter f " for various wall temperature levels, ?s= 1 [42].
6.34
CHAPTER SIX
0.8
0.7 Iw=2
0.6 o
t
0.5
r~ 0.4
0.3
0.2 0.5
/
/ 0
0.5
IO
1.5
2.0
2.5
3.0
3.5
4.0
tip
FIGURE 6.22 Effect of pressure gradient on the heat transfer parameter [" for various wall temperature levels, ?s= 1, Pr = 1 [42]. The heat transfer parameter required in Eq. 6.111 or 6.112 to calculate Stanton number is shown in Fig. 6.22 for ts = 1. It should be noted that because of the similarity of Eqs. 6.101 and 6.102 for Pr = 1, I' W~  lw  Iaw w
(6.116)
Hence, the ordinate in Fig. 6.22 can also be used in conjunction with Eq. 6.107 or 6.109 to calculate the cross flow skin friction coefficient for cases of very small yaw angles (ts  1). Note that_ law is equal to unity because the solution of Eq. 6.102 with Pr = 1 and an insulated surface is I = 1. Although the trends exhibited in Figs. 6.21 and 6.22 are generally similar, it must be cautioned that such large variations in the Reynolds analogy factor occur that the latter is no longer a useful concept. The heat transfer parameter for a cooled surface shows a rather small variation with [3p for [3p > 1A, a fact first utilized in Ref. 44 to obtain relatively simple expressions for the local heat flux to blunt bodies in hypersonic flow. Cylinder Normal to the Free Stream, Fluid With Constant Properties. For constant fluid properties, Eqs. 6.100 and 6.102 reduce to
f'" + i f " = [3p(f,2 _ 1) [" + Pr f [ ' = 2Pr ¢ f ' [ ;
(6.117) (6.118)
For a cylinder normal to the free stream, the inviscid velocity distribution is given by
with
ue=Ax m
(6.119)
[3p 12m +m
(6.120)
The term on the right side of Eq. 6.118 has been added to account for a nonuniform surface temperature.
FORCED CONVECTION, EXTERNAL FLOWS
6.35
Similar solutions for Prandtl numbers other than unity may be obtained from Eqs. 6.117 and 6.118 or their equivalent. A major simplification is the independence of the momentum equation (Eq. 6.117), from the energy equation_(Eq. 6.118), which makes findependent of [. Also, the linear form of the energy equation in I permits handling arbitrary surface temperature distributions as in the case of the fiat plate. (See the section on the twodimensional laminar boundary layer.) Solutions of the momentum equation (Eq. 6.117) [45] yield velocity distributions generally similar to those of Fig. 6.19, and the skin friction parameter f~ shown by the line labeled 1 in Fig. 6.21. The skin friction coefficient is given by
cI / p e U ~ _ 2 V
f~,"
~l,e
(6.121)
V / 2  ~p
For a uniform surface temperature, solutions of the energy equation (Eq. 6.118) [46, 47] can be expressed as St (x, O)
9eUeX
Pr"
I"
(6.122)
where the heat transfer parameter in parentheses on the right is given by the line labeled 1 in Fig. 6.22. The exponent a for the Prandtl number is given in Table 6.3 as a function of [3p. TABLE 6.3 Exponent of Pr in Eq. 6.122 ~e a
0.199 0.746
0 0.673
1 0.645
1.6 0.633
In Ref. 46, an ingenious set of transformations is employed to evaluate the recovery factor away from the stagnation line. The results for 13p= 1 show a significant departure ( 10 percent for Pr = 0.7) from r(0) = Pr 1/2. These values, however, do not agree with calculations performed in Ref. 49. Perhaps the discrepancy is due to the evaluation of r(0) in Ref. 46 by taking the derivative of a function. Slight errors in the function itself could easily account for a 10 percent error in the derivative. For accuracies of r(0)within a few percent [48], it is recommended that r(0) = Pr 1,2
(6.123)
be employed for all ]]p. For a stepwise and arbitrary surface temperature distribution, the local heat flux distribution is given by Eq. 6.65. The term St (x, 0) represents the Stanton number for the same flow conditions but with a uniform surface temperature. Equation 6.65 was derived formally with the assumption that iaw is uniform along the surface; however, small continuous variations in law or Taw are permissible. The value of i,w at x = 0 is used in the first term on the right side of Eq. 6.65, and the local value of i,w is employed within the integral. Although Eq. 6.65 appears to be reciprocal in variations of iw and i,w, this is not the case. Changes in the same direction of both i~ and i,w do not necessarily cancel, because a change in i,w takes effect gradually downstream. The kernel function St (x, s)/[St (x, 0)] in Eq. 6.65 represents the behavior of the heat transfer coefficient after a jump in wall temperature at x = s. This function was obtained in Refs. 24 and 25 by solving the energy equation with the assumption of a linear velocity profile and is given by St (x, O) = 1 
(6.124)
6.36
CHAPTER SIX
3 a* = ~ 2 ( 2  13p)
1 b* = 3
(6.125)
The analogous function for a flat plate is given by Eq. 6.66. These results were improved upon in Ref. 49, where a powerlaw velocity profile was assumed, U/Ue = (y[ye) d*, with d* found to best fit Hartree's calculations [45] as listed in Table 6.4. The form of the kernel function (Eq. 6.124) is the same, but the exponents are changed to a* =
2+d* (1 + d * ) ( 2  I~p)
b* 
1 2 + d*
(6.126)
TABLE 6.4 Exponent d* for Velocity Function, u/ue = (y/ye) d* m ~p d*
0 0 0.88
1/9 0.2 0.86
1/3 0.5 0.80
1 1 0.76
4 1.6 0.66
Use of these values to calculate the heat flux distribution from Eqs. 6.65 and 6.124 yields excellent agreement with the exact solutions of Refs. 9, 47, and 50. Hence, for m > 0, values of a* and b* given by Eq. 6.126 are preferred to those given by Eq. 6.125. For m = 0, there is little difference between the kernel functions (Eq. 6.124) based on the two different sets of exponents. Values of the kernel function Eq. 6.124 are shown in Fig. 6.12 for 13p= 0.5 and 1. The effect of the upstream temperature jump decays more rapidly with increasing 13p. As 13p+ 2, the kernel function becomes unity for all s/x < 1, as is seen directly from the functional form of Eqs. 6.124 and 6.126. Axisymmetric Stagnation Point, HighSpeed Flow. The axisymmetric stagnation point has received attention from many investigators because of its importance in the assessment of the convective heat load of missile nose cones and atmospheric entry vehicles~ The speeds involved in these applications produce stagnation enthalpies where realgas behavior must be considered in the evaluation of the forced convection. Because of the very complex behavior of the physical properties of real gases, a characteristic common to all the studies is the reliance on numerical solutions followed by correlations of the results in terms of parameters involving the fluid properties. In addition to air, other gases have been treated because of the current interest in the exploration of the planetary neighbors of Earth. The contents of this section are confined to gases in chemical equilibrium with uniform elemental composition and to flows where boundary layer similarity occursnamely, the immediate vicinity of the stagnation pointmand where there is a uniform surface temperature. The effects of surface mass transfer of the same gas as exists in the free stream are also included. Equations 6.100 and 6.102 with ts = te = 1, t = I have been solved for real air in Ref. 16 and in Refs. 5155, with the latter references utilizing the concept of total properties kr, Cpr, Prr. The air properties of Ref. 56 were employed in all the studies except that of Ref. 55, which employed properties evaluated in Refs. 57 and 58, where careful consideration was given to the effect of dominant resonant charge exchange cross sections in establishing the thermal conductivity of ionized nitrogen. For speeds under 30,000 fi/s (9144 m/s), which represent relatively moderate entry conditions into the earth's atmosphere without appreciable ionization, the numerical results of Ref. 16 for 104 atm < Pst < 102 atm and 540°R (300 K) < Tw < 3100°R (1722 K) are correlated by Nuw
prO~ ReO.5 = 0.767
( ~ePe ) 0"43 ~twpw
(6.127)
FORCED CONVECTION,EXTERNALFLOWS
where
Nuw =
Rew =
6.37
q~cp~L (i~le)kw (due/dx)L2p~ law
At the stagnation point (where Ue= 0) the recovery enthalpy is identical to the stagnation enthalpy even for Prr ~ 1. Thus, from the definition of Nuw and Rew, the local heat flux at the stagnation point in airflow for the speed range up to 30,000 ft/s (9144 m/s) can be expressed as q'~' = (V~ewNUW)le(Iw1)pr~~/~wPwx/dx/dUe
(6.128)
or from Eq. 6.127 as q" = 0"767Pr~6
(laePe) ( P ) ~~x /due le(Iw1) 0 43 lawAwXO 07
(6.129)
Here, a negative value of q" represents heat flux toward the body. At speedsgreaterthan lO,O00ft/s (3048m/s), where/e = V~/2, Eq. 6.129 can be represented in much simpler form when a relatively cold surface temperature (below dissociation temperature) is assumed:
,,/rn
(~Tw
1.38 }°'°7~_~due 492 Tw/492 + 0.38 ~ U221([w 1)
q w ~ / ~ t = 121
(6.130a)
For the heat flux expressed in Btu/(s.ft2), the dimensionless velocity is expressed as U = V**/ (10,000 ft/s), Twis expressed in °R, and/'st is expressed in atm. In SI units, Eq. 6.130a becomes
,, /rn (~ Tw qw~/~ = 2382 273
1.38 / ° ' ° 7 / r n due T~/273 + 0.38] ~¢ V. ~ U221(iw 1)
(6.130b)
where U = V~/(3048 m/s), q" is in W/m% Twis expressed in K, and Pst is in N/m 2. Note that rn V~
m
1.54 / P s t  P• .! V.
v
(6.131)
P,t
For stagnationpointheating in gases other than air, correlations similar to Eq. 6.128 were obtained in Ref. 53 for speeds up to 30,000 ft/s (9144 m/s). The correlation is of the form NuE~=a*( laePe)B*
(6.132)
la~Pw with coefficients A* and B* given in Table 6.5 for the various gases considered. There is rather close agreement between Eqs. 6.127 and 6.132 for air. The heat flux is obtained from Eq. 6.128. TABLE 6.5
Coefficientsfor Eq. 6.132 [53]
Gas
A*
B*
Air N2 H2 CO2 A
0.718 0.645 0.675 0.649 0.515
0.475 0.398 0.358 0.332 0.110
6.38
CHAPTER SIX
For speeds above 30,000ft/s (9144 m/s), where the total enthalpy reaches values where ionization significantly lowers the viscosity ~te, the correlation for air (Eq. 6.127) from Ref. 16 begins to break down. Similar behavior was observed on pointed cones in Ref. 17. Similarly, it was found for argon in Ref. 53 that only the solutions for the lowest wall temperatures were correlated well by the property parameter ~l,ePe/~l, wp w. T h u s , extrapolation of the Nusselt number relations beyond the range of ktepe/ktwpwactually used in the correlations could yield significant errors. Two alternate approaches avoid this problem. In Ref. 52 the intermediate Nusselt number correlation was bypassed, and a correlation for air was achieved directly in terms of the heat flux for speeds up to 50,000ft/s (15,200re~s)in English units as follows: ,, / r, qw~/~
= 119
~/~r, ~x due U Z 1 9 ( l w
1)

(6.133)
(In SI units, the coefficient is 2342.) It will be noted that this equation is quite similar to Eq. 6.130. The implication of this similarity is that the large variations in ~te associated with ionization, which is not present in the range of velocities for which Eq. 6.130 is valid, can be ignored in the evaluation of heat flux. In fact, it is systematically demonstrated in Ref. 53 that the surface heat flux is quite insensitive to variations of the gas properties at the boundary layer edge and is controlled instead by the gradients of these properties near the surface. Apparently, correlations such as Eq. 6.127 or 6.132 result because in the speed range where they are applicable, the physical properties vary monotonically through the boundary layer, and their derivatives in the inner portion of the boundary layer are related to the ratio of properties across the boundary layer. For a variety of gases, the resulting heat flux expression, using strong shock relationships, is given by
,, ~
qw~ / ~ t = F0
,
r/~ due  ~
U22(i w 
1)
(6.134)
where the units are the same as in Eqs. 6.130a and 6.130b, and F0 is given in Table 6.6. From the value of F0 for air in Table 6.6 and the form of Eq. 6.134, it is apparent that the correlation of Ref. 53 yields results essentially identical to those of Eq. 6.133 taken from Ref. 52 and to Eq. 6.130 derived by directly extrapolating the lowerspeed range equation of Ref. 16. Thus, it is recommended that Eq. 6.134 with the coefficients of Table 6.6 be utilized to predict stagnationpoint heat flux at speeds greater than 10,000 ft/s (3048 m/s). At lower speeds, Eqs. 6.128 and 6.132 with the coefficients of Table 6.5 are appropriate. A comparison of these techniques with existing stagnationpoint measurements is shown in Fig. 6.23. It should be noted that convection predominated over radiation in these measurements despite speeds up to 50,000 ft/s (15,200 m/s) because of the small nose radii for the models. For body dimensions exceeding a few feet, shock layer radiation begins to compete with convection at speeds of about 25,000 ft/s (7620 m/s) and becomes the predominant heating mechanism at higher speeds. TABLE 6.6 CoefficientF0 Fo Gas (% volume) Air N2 CO 2
A 91% N210°/o CO2 50% N250% CO2 40% N210% CO250% A 65% CO235% A
s.ft2 \ atm ] 121 121 141 165 120 134 144 142
w( m
~7 N/m2 2382 2382 2775 3248 2362 2638 2834 2795
FORCED CONVECTION, EXTERNAL FLOWS
6.39
(Ie i w) x I0 6, J/kg 0
25 50 , ! , . I , 177~Air data, various sources
5000 ._. E
4000
1
I
75
I00 I00
b X
I
OJ
75 3.
4
= d 3000 ~,r
5o

~1
/7/~/~I ~J" ~/~/Eq"6134152]
'~ IO00Eq. 6.132
o
Io
I !
20 30 (Ieiw x 103 ,Btu/Ibm 1 ! 30 40 Voo x I0 3, f t l s (a)
1
I0
2O
(Ieiw) 25
x I0 6, d/kg 50
i
75
I00 I00 ~o X Od
75 ~.
a
Eq. 6.134
''
o.13
0 I
I0 i
0 I0
n,
¢~ ' ~ ~ ""!
o
o :~1
I
1 I I 1 E oRutowski and Chan [ 5 9 ] 4 0 0 0  nGruszczynski and Warren [ 6 0 ] zxYee et al. [ 61 ] I~; ~ 3 0 0 0  ONerem et el. [ 6 2 ]
I000
3
50
5000
2000
25 ~
5O
40
.,.,..
I~
V
~ Hoshizaki
2ooo
L.r
,~
"

~ ~_ ***~ "'Hoshizaki [ 52] ,
20 50 (Ieiw) x 10"3, Btu/Ibm
I
...... ,
i !
4O
5o
i
I
I
I
20
30
40
50
Voo x I0 s, ftls
50
o
:gl
(b)
FIGURE 6.23 Comparison of heat transfer rates predicted by Eq. 6.134 and the coefficients of Table 6.6 with data [53]. (a) air; (b) carbon dioxide.
Stagnation Line on a Cylinder in HighSpeed Flow. The stagnation line on a uniformtemperature cylinder of infinite length with its axis either normal to the free stream or swept back at angle A is characterized by boundary layer similarity solutions with 13p= 1. The solution in Ref. 16 of Eqs. 6.1006.102 for 13p= 1 yields the correlation
Nuw PrOw4 Reds = 0.57
( ~l,ePe)0"45 PwPw
(6.135)
for iw/le < ts (see Eq. 6.92 for definition of ts) and V**cos A < 29,000 ft/s (8840 m/s). Although Eq. 6.135 was established from calculations that employed realair properties, the resulting coefficient and exponents are consistent with those for lowspeed calculations for either constant properties or ideal gases. In terms of heat flux, Eq. 6.135 becomes
~l,e,..eX045[, /due (iw law) t~ w..wX005 p ) ~~
q~' = 0.57PrOw6 ( P )
(6.136)
6.40
CHAPTER SIX
where the recovery enthalpy
lawis given by
Iaw=le(1 
We2
Pr°w5)   ~  = l e  ( 1  PrTw) 0.5
V2 sin 2 A
(6.137)
and a negative q" represents heat flow into the body. An alternate form of Eq. 6.136, valid for V~ > 10,000 ft/s (3048 m/s) and having the same form as Eq. 6.130 for an axisymmetric stagnation point in airflow, is
[(Twill2
tt/Fn
qw ¥  ~ t = 87"3L\ 492 ]
1.38
]°°5/rndu e
Tw/492+0.38
~= ~x U22(Iw + 0.15 sin2 A  1)
with the following units: q~, Btu/(s.ft2); rn, ft; Pst, atm; In SI units, this equation is
,,/r, qw~/~
= 1718
Tw, °R; V=, ft/s; and
(6.138a)
U = VJ(10,000 ft/s).
[ ( T w ) 1/2 1.38 ]°.°Sr~__v_fdlg e ~ (Tw/273) + 0.38 ~ U22(/w + 0.15 sin 2 A  1)
(6.138b)
with q'w', W/m2; r,, m; Pst, N/m2; Tw, K; V=, m/s; and U = V=/(3048 m/s). The velocity gradient in Eq. 6.138 is obtained from Eq. 6.131, but with 1.54 changed to 1.43. In these equations, Pst and Pst are the inviscid flow conditions on the stagnation line of the swept cylinder. For an ideal gas in hypersonic flow, the inviscid flow relationships are particularly simple, and Eq. 6.138 shows that the heat flux is reduced with sweep by approximately cos 3/2A. Equation 6.138 may be extended to gases other than air by setting the quantity in brackets equal to unity and replacing the coefficient 87.3 by 0.72F0, where F0 is given in Table 6.6. Mass Transferin Stagnation Region. As on a flat plate, surface mass transfer is an effective means for alleviating convective heating in the stagnation regions of axisymmetric bodies and cylinders. The effect of surface mass transfer of a gas with the same elemental composition as the free stream will be treated initially. Consistent with the similarity requirements following Eqs. 6.1006.102, the surface mass transfer rate is given by
pwVw=f(O) ~gwPw[s ~p[e due dx
(6.139)
In the vicinity of the stagnation region and with a uniform surface temperature, the terms under the radical sign are constants. Thus, boundary layer similarity, that is, f(0) being independent of x, requires a uniform mass injection rate along the surface rather than one varying as x 1/2 as on a flat plate. A convenient correlation parameter, as in the case of the flat plate, is
Bm = ~ pwVw = pwvw(iwlaw) = Pete St0
q'w'0
f(O) PrTw V~p(Nuw/~/~ew)0
(6.140)
where the subscript 0 denotes zero surface mass transfer. The effect of surface mass transfer in reducing the Stanton number is indicated in Fig. 6.24. Note that for the stagnation point on a body of revolution or an unswept cylinder, the recovery enthalpy is equal to the total enthalpy. Hence, q" St so that Fig. 6.24 indicates the reduction of the heat flux as well as the Stanton number. The line representing the axisymmetric stagnation point correlates all the gases listed in Table 6.5 to within a few percent [54]. Although the curves for the cylinder are based on air calculations, the correlation for the various gases for the axisymmetric stagnation point implies that the cylinder results can be applied to other gases when the injected and fleestream gas are the same. Note that in the coordinate system of Fig. 6.24, the effect of sweep is quite small. Along the stagnation line of a swept cylinder, the recovery enthalpy is less than the stagnation enthalpy. Thus, the effect of surface mass transfer on the recovery factor, as shown in Fig. 6.25, should be considered in establishing the proper driving potential for the
FORCED CONVECTION, EXTERNAL FLOWS
1.0
0.8 ~Cylinder o
stognofion line, A ~,77 °
0.6 " ' ~ ~ ~ ' ~ _ ~  C y l . . . . _., . . ~ . ~ , , , ~ inder s/ognotion line, ,=0 Axisymrnetric stagnotion point~ ~ 0.4
,..
0.2
0.2
0.4
0.6
C).8
1.0
I2
1.4
1.6
pw vw/(peueSt o) F I G U R E 6.24
The reduction of Stanton number in stagnation regions by surface mass transfer.
0.86 ~
'
'
Pr r =0.7
0.84
~, : 1 w ~ / ( 2 I , l ,
0.82
~,o.~
?'\
.....
\ ' ~ .,~'~ ~
\ 0.74
.....
0
0.2
0.4
0.6 f(O)
0.8
1.0
 1.2
F I G U R E 6.25 The effect of surface mass transfer on the recovery factor on the stagnation line of a yawed cylinder of infinite length [48].
6.41
6.42
CHAPTER SIX
heat flux [48]. The heat flux for a swept cylinder normalized by its value with zero mass transfer is
q~ q~o
st [ 1
sin A'. 1
(6.141)
Sto 1  (1  to(O)) s~n2~ .  ~w
Because the term containing the recovery factor depends on sin 2 A, modification of the recovery factor to account for surface mass transfer need be made only for large sweep angles. The graphic procedure for using these figures in establishing the required mass flow rate to yield a prescribed surface temperature was described on page 6.24. The evaluation of the effectiveness of a transpiration cooling system utilizing a foreign gas is quite formidable and requires the use of complex computer codes.
Heat Transfer Over a Single Cylinder and Arrays of Cylinders in LowSpeed Cross Flow. A boundary layer subjected to a significant static pressure increase in the streamwise direction thickens considerably and may eventually separate from the surface of the body. For steady separation, the velocity gradient normal to the surface vanishes at the separation point, and downstream the flow direction is reversed. Although the skin friction vanishes, significant amounts of heat transfer can occur at the point of separation. Separation is commonly encountered in cross flow over blunt bodies such as circular cylinders. The flow pattern downstream of separation is very complex and is often accompanied by unsteadiness and vortex shedding. Theoretical treatment of such flows is still in an early stage of development [63], and heat transfer predictions must rely on experimental data. The following sections summarize the experimental data available on heat transfer rates from single cylinders and arrays of cylinders in cross flow. Single Cylinder The classic experiments on the average heat transfer rates from a cylinder in cross flow were performed by Hilpert [64] for a wide range of Reynolds numbers in air. Hilpert's results are shown in Fig. 6.26 as average Nusselt number versus the cross flow Reynolds number. In Ref. 66, Morgan made an extensive review of more recent heat transfer data obtained on a cylinder in cross flow. He found that the average Nusselt number could be correlated as (6.142)
Nua = (A + B Re,~*) Pr m•
103 
Item
Diameter
'Item
Diomeier
_/
5 + wire no. I 0 . 0 1 8 9 mm O pipe no. 8 ; 2 . 9 9 m m 25.0 mm 3 • x wire no. 2 0 . 0 2 4 5 mm A pipe no. 9 2 • wire no. 3 0 . 0 5 0 mm V pipe no, 10 4 4 0 mm 10z .v
5
V wire no. 4 wire no. 5 •
wire no. 6
0099 rnm 0 . 5 0 0 mm
~ pipe no. II El pipe no. 12
/
~r ,e, f
9 0 . 0 mm 150.0 mm
1.00 mm ..
2 10 5 3 2 I
y,.r~
., ._.rowI+ * " " F I,,,X4~" " k''+ X°4]
2 3 5
I0
2 3 5
102 2 3 .5
103
2 3 5
tO4
2 3
.5
I0 ~
2
3
,5
106
Red
FIGURE 6.26 Nusseltnumber for average heat transfer from circular cylinder in cross flow of air [64]. (Reprinted from Ref. 65 by permission of McGrawHill.)
FORCED CONVECTION, EXTERNAL FLOWS
6.43
where A, B, n*, and m* are constants. The value of the exponent m* found in various experiments lies between 0.3 and 0.4; Zukauskas [67] recommends a value of 0.37. The data were obtained mostly in air. Depending on the Reynolds number, the scatter in Nusselt numbers calculated from Eq. 6.142, using constants from various investigations, ranges from 10 percent to 29 percent. Other types of correlations were proposed and tested by Morgan and found to be less accurate than the above. Morgan attributed the scatter in the heat transfer data to three factors: (1) aspect ratio (length/diameter) of the cylinder, (2) freestream turbulence level, and (3) wind tunnel blockage effects. Corrections by Morgan [66] for the combined effects of freestream turbulence and wind tunnel blockage on Nusselt numbers measured in air are shown in Fig. 6.27. Here, d/dr is the ratio of the cylinder diameter to the wind tunnel height or diameter, 8 NUd is the increase in the Nusselt number over Hilpert's measurements (Fig. 6.26), and Tu is the intensity of the longitudinal turbulent fluctuations in the free stream. In addition to these corrections, Morgan also proposed the Nusselt number correlations shown in Table 6.7 that are applicable to an extremely wide range of Reynolds numbers in air. At the higher Reynolds numbers, these Nusselt numbers are quite consistent with those of Fig. 6.26. 1.0
I
1
1 I I i
I
i00' Tu


~
0.5
,

i

12%_
~
9 7
_,
,~ 0.1
/
I 0.01
li I I I I 0.0,5 0. I d/d'r
/
I
..
I
I 0.5
F I G U R E 6.27 Correction factors to Nusselt numbers for combined effects of wind tunnel blockage and freestream turbulence [66]. (Reprinted by permission of Academic Press.)
Local heat transfer rates from the surface of a cylinder in cross flow in air were measured by Schmidt and Wenner [68] and are shown in Fig. 6.28. The local Nusselt number is based on the local heat transfer coefficient and the cylinder diameter. Note that for subcritical Reynolds numbers (Red < 170,000), the local Nusselt number decreases initially along the surface from the forward stagnation point to a minimum at the separation point and subsequently reaches high values again in the separated portion of the flow on the back surface. For
6.44
CHAPTER SIX
Correlation of CrossFlow Forced Convection From Cylinders in Air [66]
T A B L E 6.7
NUd =
D2 Re,7'
Red From
To
D2
nl
104 4 x 103 9 x 102 1 35 5 X 103 5 X 104
4 X 103 9 X 102 1 35 5 × 103 5 X 104 2 x 105
0.437 0.565 0.800 0.795 0.583 0.148 0.0208
0.0895 0.136 0.280 0.384 0.471 0.633 0.814
Reynolds numbers above the critical value, transition from laminar to turbulent flow in the upstream attached boundary layer causes a dramatic increase in the local heat transfer, followed by a sharp decrease in the separated flow region. The local heat transfer distribution is extremely sensitive to the freestream turbulence intensity according to the measurements of Kestin and Maeder [69]; this is reflected in the correction terms shown in Fig. 6.27 for the average Nusselt number. Arrays of Cylinders. The heat transfer behavior of a tube in a bank differs considerably from that of a single tube immersed in a flow of infinite extent. The presence of adjacent tubes in an array and the turbulence and unsteadiness generated by upstream tubes generally tend to increase the overall heat transfer from a particular tube. After the flow has passed through several rows of tubes, however, the heat transfer from individual tubes becomes independent of their location and just a function of the Reynolds number with a parametric dependence on the array geometry. Average and local heat transfer data for tube banks have been summarized by Zukauskas [67].
NUd=
3 0 0 .__ Curve
oo
Red
I
59,800
2
101,300
3
170,000
4
257,600
5
426,000
/ 'front
FIGURE 6.28 Distribution of local heat transfer on the surface of a circular cylinder in cross flow in air [66]. (Reprinted from Ref. 65 by permission of McGrawHill.)
FORCED CONVECTION, EXTERNAL FLOWS
6.45
The overall heat transfer data for a tube in an infinite array are correlated by Kays and London [70] as: St = ch Pr 2/3 Re~ °'4
(6.143)
NUd = Ch P r 1/3 R e °'6
or
(6.144)
The Reynolds number in this correlation is based on the flow velocity at the minimum area section normal to the flow direction. A typical correlation for Ch is shown in Fig. 6.29 for the case of a staggered array. For effects of array geometry, Refs. 67 and 70 should be consulted.
0.50 I
y : 1.25
/
y : 1.00
0.40
= 0.75 ~..,...',
,.,....,.,.. ,...,m., /
u
" 0.30
0()0
/
©
0.20
~'td
Ot
0()0 I0t
0.10 1
1.0
1.5
20
I
1
I
1
215
~t F I G U R E 6.29 Correlation of coefficient in Eq. 6.143 or 6.144 for an infinite bank of staggered array, 300 < Red < 15,000 [70]. (Reprinted by permission of McGrawHill.)
For a tube located near the front of the bank, the overall or average heat transfer is lower than that predicted by correlations for infinite arrayr~ The necessary correction factor as a function of row number in an array as presented by Kays and London [70] is shown in Fig. 6.30. 1.00 J 0.9O
/
I'"
/
/
.~
I
0.80
0.704
5
6
8
I0
20 30 40 50 60 Row number in a tube array
80 I00
200
FIGURE 6.30 Correlation factor to account for rowtorow variation in heat transfer from a tube in a staggered or inline bank [70]. (Reprinted by permission of McGrawHill.)
6.46
CHAPTER SIX
TWODIMENSIONAL TURBULENT BOUNDARY LAYER Turbulence Transport Mechanisms and Modeling In turbulent flows, the transport of momentum, heat, and/or individual species within gradients of velocity, temperature, and concentration is caused predominantly by the chaotic motion of elements of fluid (eddies). This mixing process transports properties much more effectively than the molecular processes identified with viscosity, thermal conductivity, and diffusion. A rather complete description of these processes is given in Refs. 7173. Currently there exist computers with sufficient storage capacity and speed to allow computation of these timedependent motions for rather simple flows with finite difference meshes sufficiently fine to resolve the larger eddies of the motion. Even with such computations, however, it is necessary to model the effects of the eddies that are too small to be resolved. It is believed that since the transport of properties is governed by the larger eddies, the modeling process is less critical in these computations than where the entire turbulence is modeled. These "turbulence simulations" are still too costly for routine engineering computations and are used primarily to study the "physics" of particular turbulent flows. In fact, the results provide much more information than an engineer may ever want or need. In engineering computations, the turbulent transport of properties is usually treated in a statistical manner, where computations are concerned with the mean velocities, temperatures, and/or concentrations. This statistical approach, however, masks many of the actual physical processes in the dynamic flow field, which must be recovered by the modeling at some level of the turbulence statistics. This modeling was originally guided by the results from experiments, but currently this guidance can rely on "simulations" as well. The statistical turbulence models generally employed at present are based on timeaveraging, at a single point in space, the instantaneous dynamic equations representing the conservation of mass, momentum components, energy, and species concentrations. These equations, in their most general form, apply to compressible, viscous, heatconducting, and diffusing fluids. The statistical representation is initiated by expressing the dependent variables as the sum of a mean and fluctuating quantity, e.g.,
u=~+U ~
p=~+p' K~ =
p
Ki +
K;
Substitution of such a decomposition of the dependent variables into the basic conservation equations is then followed by timeaveraging the equations according to the following definition:
lfo'
f= im T I(r) dr It is seen that this definition of mean quantities eliminates terms that are linear in the fluctuating quantities. Moments of the fluctuating quantities that are retained in a boundary layer are p'v', (pv)'u', (pv)'w', (pv)'i', and (pv)'K~, which represent the turbulent flux of mass, momentum, heat, and species concentration in the direction normal to a surface. These quantities are added to their molecular counterparts. Details of these derivations can be found in several sources, e.g., Refs. 65, 71, 74, and 75. The evaluation of these statistical second moments is the goal of turbulence models. These models fall into two categories. First are models in which the turbulent fluxes are expressed in the same functional form as their laminar counterparts, but in which the molecular properties of viscosity, thermal conductivity, and diffusion coefficient are supplemented by corresponding eddy viscosities, conductivities, and diffusivities. The primary distinction is the recognition that the eddy coefficients are properties of the turbulent flow field, not the
FORCED CONVECTION,EXTERNALFLOWS
6.47
physical properties of the fluid. The second category of turbulence models includes those that express the turbulent moments in terms of partial differential field equations. When this is done, new moments (some of higher order, others involving pressure fluctuations, and still others involving space derivatives of the fluctuating quantities) appear in these equations. The number of moments grows faster than the number of the additional moment equations; thus, the set of equations cannot be closed. References 73 and 76 demonstrate this closure problem in detail. The classical turbulence models express the eddy viscosity algebraically in terms of a turbulence scale and intensity that are related, respectively, to the characteristic length dimensions of the flow field and the local mean velocity gradients. This implies an equilibrium between the local turbulence and the mean motion. This requirement of equilibrium has been relaxed in some eddy viscosity models where the intensity and scale of turbulence used to evaluate the eddy viscosity are expressed by partial differential equations for the turbulence kinetic energy and its dissipation rate. This latter class of models is presented in detail, for example, in Refs. 76 and 77. A 1982 conference held at Stanford University [63] was devoted to assessing the merits of existing turbulence models in the prediction of the mean velocity fields for both simple and complex turbulent flows. The flow fields employed as standards for comparison were selected on the basis of their being welldocumented experimentally. It was found that the field models of turbulence, the secondorder closure of the Reynolds stress equations or the twoequation eddy viscosity models, while having a broader range of application than particular algebraic eddy viscosity models, did not show dramatic improvement in accuracy over the simpler models for flow situations similar to those experiments on which the simpler models were based. In view of these observations, and the analytical advantages of the simpler models in the analysis of convection, classical algebraic eddy viscosity models will be used to represent turbulent transport in this chapter. The turbulent flux vector for the local shear stress is given by x, = 15eM~
oy
=(pv)'u'
(6.145)
and the heat flux by D
~T
q7 =ffcpe. kS..= (pv)'i' o)i
(6.146)
The quantity eM is called the eddy diffusivity for momentum, and e~/is the eddy diffusivity for heat. They are related through the turbulent Prandtl number Pr, = eM/e/4. Although these eddy diffusivities act in the same manner as the kinematic viscosity and thermal diffusivity in laminar flow, the critical difference is that the eddy diffusivities are not properties of the fluid but are dependent largely on the dynamic behavior of the fluid motion. In this section the fluid dynamic bases for evaluating these eddy diffusivities are given. They will then be used in a variety of convective heating situations to yield formulas useful in engineering computations.
Mean Velocity Characteristics for Constant Fluid Properties.
Mean velocity distributions measured in turbulent boundary layers when the fluid properties are uniform (low speeds and small temperature differences between the free stream and surface) are described in the reviews of Refs. 78 and 79. At a given station, the turbulent boundary layer is composed of two regions with velocity profiles described by the "law of the wall" and the "law of the wake" after Coles [80, 81]. Wall Region. The region near the wall possesses a universal velocity profile when the data are correlated in terms of the coordinates u ÷ = u/u* and y+ = u*y/v. T h e quantity u* = X/~w/p is called the friction velocity and is the appropriate characteristic velocity in this region. The corresponding characteristic length is v/u*.
6.48
CHAPTER SIX
The nearwall region is composed of three layers as shown in Fig. 6.31. The layer immediately adjacent to the surface (y+ ~< 5) is called the laminar sublayer where, because of the presence of the surface, the turbulence has been damped into a fluctuating laminar flow. In this layer, viscosity predominates over the eddy viscosity, and the velocity distribution may be approximated by u÷:y +
(6.147)
40, ° oeo•
• •
•
30
o• ° o o
•
•
I +
20
S 1
I
2
5
I
zo
,
I
~o
~oo
I
200
,
~o~c~z&o
so~o~o,ooo
y* FIGURE 6.31 Universal velocity profile for an incompressible turbulent boundary layer near the surface ("law of the wall") [79]. (Reprinted by permission of Academic Press.)
For y+ > 50, the turbulent processes completely dominate the local shear, and the resulting correlation can be represented by 1
u ÷ =  In y÷ + B K
(6.148)
The parameter ~: is called the von K~irm~in constant, and the value that fits most of the data is 0.41. The corresponding value of B is 5.0. Intermediate between these layers is the buffer layer, where both shear mechanisms are important. The essential feature of this data correlation is that the wall shear completely controls the turbulent boundary layer velocity distribution in the vicinity of the wall. So dominant is the effect of the wall shear that even when pressure gradients along the surface are present, the velocity distributions near the surface are essentially coincident with the data obtained on plates with uniform surface pressure [82]. Within this region for a flat plate, the local shear stress remains within about 10 percent of the surface shear stress. It is noted that this shear variation is often ignored in turbulent boundary layer theory. Wake Region. The experimental data toward the outer edge of the boundary layer do not correlate in a plot of u +(y+). Correlation of these data can be achieved by utilizing the boundary layer thickness 8 as the characteristic dimension and expressing the velocity as a decrement relative to its value at the boundary layer edge. Such a "wake" correlation is shown in Fig. 6.32 for a plate with uniform pressure. Although ~5 is rather arbitrary because of the asymptotic approach of the velocity to its freestream value, no serious error results in these correlations if a consistent definition such as u(8) = 0.995Ue is adopted. The velocity distribution for the combined wall and wake region of a fully developed turbulent boundary layer was formulated by Coles [81] as
FORCED CONVECTION, EXTERNAL FLOWS
,y+ + B +   w
u ÷=In K
6.49
(6.149)
K
where w(y/6) is the wake function indicated in Fig. 6.33. This function is approximated quite well by 1  cos (ny/8). Equation 6.149 applies to equilibrium boundary layers, as defined by Clauser [79], where (6*/%) dP/dx is constant along the surface. Under these conditions and at large Reynolds numbers, the parameter ~t in Eq. 6.149 is independent of position.
,.,.~
i
I
~o _
I
o x = 0 5 rn x = I.Om
,.
•
x"b,,,
•
x=
I.Sm
x:2.5 m • x:3.2 m x
'~,,
• x= 3.9 m
~' ¢ ~ ~ o X
, x= 5.3 m
I 5
i
o
05
1.0
1.5
2.5
2.0
2 + IOglo(y/8)
FIGURE 6.32 Velocity decrement for an incompressible turbulent boundary layer away from the surface ("law of the wake") [83].
2.0
08 04
Coles [84] used Eq. 6.149, with ~: = 0.41 and B = 5.0, to evaluate cl/2, 6, and ~ from a large set of equilibrium boundary layer data. In this evaluation, Coles excluded data for y+ < 50, where Eq. 6.149 does not apply, and data for values of y near the boundary layer edge, where Eq. 6.149 provides poor values of the slope du÷/dy÷. For flat plates with uniform freestream velocities, Coles found agreement with the data when ~ = 0.62. On the other hand, if experimental skin friction data obtained at very high Reynolds numbers are used to define ~, a value of 0.55 is favored. The use of these different values of ~ produces differences of only a few percent in the skin friction and boundary layer integral parameters, such as the displacement thickness
/ 0.2
/
and the momentum thickness 04
0.6
08
to
0 =
1
dy
(6.151)
yl~ FIGURE 6.33 Eq. 6.149 [81].
Coles' "law of the wake" function in
The value of ~ recommended tive p r e s s u r e g r a d i e n t s is
in r e g i o n s o f z e r o o r posi
6.,50
C H A P T E R SIX
5* dP = 0.55 + 0.47 m ~ % dx
(6.152)
In regions of negative pressure gradients, ~ can be estimated by doubling the constant 0.47 in Eq. 6.152. The parameter ~ is influenced by low Reynolds numbers even when the boundary layer is fully turbulent. In zeropressuregradient flow, ~ can be represented approximately by x = 0"6(Re°Re0470) °85
(6.153)
for Re0 < 5000; ~ = 0.55 for Re0 > 5000. If Eq. 6.149 is used throughout the boundary layer to define the integral parameters, the displacement thickness becomes 5*
1+ ~
fc r
(6.154)
and the shape factor is g* ( 2 + 3.179~ + 1.5 x2 cX~f/2) 1 H = ~ = 1 1+~
(6.155)
The approximations introduced in the integrations leading to these equations, that is, utilizing Eq. 6.149 throughout the sublayer and buffer layer and ignoring the nonzero value of ~gu÷/Oy÷ near the outer edge of the boundary layer, are valid at Reynolds numbers of practical interest. Equation (6.149) also leads directly to the skin friction expression 1
V'c/2

1 ~c 2~t In Re~, +  In +B+~ ~ ~ 1 + ~t 1
(6.156)
For 2000 < Re0 < 50,000 and 0.55 < ~t < 2.0 and with ~ = 0.41 and B = 5.0, Eqs. 6.155 and 6.156 yield skin friction coefficients within +10 percent of the wellestablished LudwiegTillman correlation equation [82] Cr= 0.123 2
x 10 0"678H Re0°268
(6.157)
Given the momentum thickness at some location on a body with streamwise pressure gradients, the evaluation of the local skin friction coefficient from Eq. 6.157 requires knowledge of the shape factor H. Alternatively, the same input information can be used to evaluate both the skin friction coefficient and the shape factor by solving Eqs. 6.1526.156 iteratively. Clauser [79] made two observations regarding turbulent boundary layer velocity profiles that have direct influence on convective heating analyses. First, he noted that the region characterized by the "law of the wall" equilibrates very quickly after it is disturbed by some external force. Thus, this region is nearly in equilibrium with its local surface boundary conditions. In contrast, the region characterized by the "law of the wake" possesses a long memory of the upstream events. Clauser also found that the eddy diffusivity in the wake region is essentially independent of the distance from the surface y and is related to the local boundary layer thickness, which reflects the growth of the boundary layer from the leading edge. The resulting expression for the eddy diffusivity in the wake region on a flat plate or a surface with an equilibrium turbulent boundary layer ((5*/xw)(dP/dx)= constant) is ~ V
where g = v + eM.
U eS *
0.018 ~
P
(6.158)
FORCED CONVECTION, EXTERNAL FLOWS
Flat Plate With Zero Pressure Gradient. 6.154 and 6.155 reduce to
With values of ~ = 0.41, B = 5.0, and R = 0.55, Eqs.
 3.8
cl
(6.159)
H = ~ = 1  6.6
and
6.51
(6.160)
The skin friction in the range 5000 < R0 < 50,000 can be expressed as c~= 2.44 In Re,, + 4.4
(6.161)
A convenient relationship expressed explicitly in terms of the momentumthickness Reynolds number that agrees with Eqs. 6.160 and 6.161 within a few percent in the range 5000 < Re0 < 50,000 is given by cI _ 1 2 (2.43 In Re0, + 5.0) 2
(6.162)
This form of equation originated with von K~irm~in; however, the coefficients appearing here are slightly different because they are based on more recent data. It is often convenient to evaluate the local skin friction coefficient with the following simpie equations: cI 2
0.0128
c/ 2
0.0065
1~..1/4
Re0, < 4000
(6.163)
Re0. > 4000
(6.164)
x'~ 0e
o,al/6 x'~.,,Oe
after Blasius and Falkner [85], respectively. Equations 6.1626.164 are compared with highReynolds number data [86] in Fig. 6.34. 102
..... " ~
~Eq. 6.163 •
Eq. 6.162Eq. / 6.164/I7 " ~
~  ' ~ =,~,c,~~ ~z~::,,,~,~:~..,~o~.." _
(M
"u 1 0  3
,.,=
102
i03
104
I0 ~
Re8e
FIGURE 6.34 Comparison of local skin friction coefficients predicted by Eqs. 6.1626.164 with the experimental data of Ref. 86.
6.52
CHAPTER SIX
On an impervious flat plate with uniform pressure, the momentum integral equation is
Xw Deu2
_
dO
(6.165)
dx
This equation leads directly to an expression for the local length Reynolds number, Rexe =
foRe0~ dRe0e cfl2
(6.166)
If the laminar and transitional zones are short, the turbulent boundary layer can be assumed to begin at the leading edge of the plate. In this case, for surface length Reynolds number of a few million, Eq. 6.163 can be employed in Eq. 6.166 to establish the following simple relationship: cy_ 0.029__~6 2 R ~Xe 1/5
(6.167)
For length Reynolds numbers in the tens of millions, the upstream portion of the boundary layer is not significant, and Eq. 6.164 may be used in Eq. 6.166 all the way from Re0e = 0 tO yield cI_ 0.013____~1 2 R ~Xe 1/7
(6.168)
A more general equation proposed by SchultzGrunow [83] that covers the full Reynolds number range is given by Q = 0.185 2 (log Rexe) 2584
(6.169)
Eddy Diffusivity Models. The mean velocity data described in the previous section provide the bases for evaluating the eddy diffusivity for momentum (eddy viscosity) in heat transfer analyses of turbulent boundary layers. These analyses also require values of the turbulent Prandtl number for use with the eddy viscosity to define the eddy diffusivity of heat. The turbulent Prandtl number is usually treated as a constant that is determined from comparisons of predicted results with experimental heat transfer data. The presence of "wall" and "wake" regions in the turbulent boundary layer is reflected in the distribution of the eddy viscosity. In the outer wake region, Clauser's empirical form (Eq. 6.158), has been adopted for finitedifference boundary layer computations, although Ref. 87 suggests that the constant in Eq. 6.158 be altered to 0.0168. In analyses, however, the need for defining the wake region eddy viscosity has not been critical, largely because of the nearness of the value of the turbulent Prandtl number to unity and the use of the Crocco transformation, as demonstrated in the next section. Evaluation of the wall region eddy viscosity has been facilitated by the fact that the shear stress remains essentially constant across the wall layer. On a fiat plate with uniform freestream conditions, the shear stress drops less than 10 percent from its wall value across the wall layer. The equation governing lowspeed flow on a flat plate in the absence of a pressure gradient is U ~~9u +V
~9u ~ [(V+eM)~9_~] ~)y~)y
(6.170)
If the shear is equal to the wall shear throughout the region of interest, Eq. 6.170 reduces to the Couette flow approximation
du (v + eM) dy
_
Xw p
(6.171)
FORCED CONVECTION, EXTERNAL FLOWS
6.53
Expressed in nearwall coordinates, u ÷ = u/u* and y÷ = u * y / v , Eq. 6.171 becomes eM D
v
~
1

du+/dy +
1
(6.172)
Equation 6.172 directly relates the mean velocity data to the eddy viscosity. From Eqs. 6.147 and 6.172, it is seen that the sublayer is physically identified with zero eddy viscosity or no turbulent transport. Beyond y÷ = 50, Eqs. 6.148 and 6.172 indicate EM P
 ~y+ 1
(6.173)
For 5 _ M 2 , am = 0.6 when M1 < M2. The graphic technique for using this figure to establish pwVw is to determine the intersection of the curves in Fig. 6.44 with a straight line drawn from the origin with slope
ilw ilc ( Ml lam i2~w i2w \~2 ] In an ablating system, the change in the enthalpy of the coolant, ilw ilc, includes the effective heat of ablation, the heat absorbed by phase change, and the chemical processes that take place in the char, if present. In evaporating systems, ilw  i~c includes the heat of vaporization.
SurfaceRoughness.
Up to this point, the turbulent boundary layer has been assumed to form on a surface that is aerodynamically smooth, namely, a surface whose roughness elements are small compared with the thickness of the viscous sublayer. As many surfaces in practical appli
F O R C E D CONVECTION, E X T E R N A L FLOWS
6.67
0.8
0.6
"
.
o3
0.4
\ ~'0~
~4.35 ~3.2
0.2
0
I
2
3
4
[ PwVw/(peueSto)](M2/Ml)am F I G U R E 6.44 Compressibility effects on the reduction of the Stanton n u m b e r by surface mass transfer on bodies with zero axial pressure gradient and including effects of foreign gas injection.
cations are not aerodynamically smooth, the effects of surface roughness must be accounted for in describing the hydrodynamic and thermal behavior of turbulent boundary layers Skin Friction. The earliest investigation of the effects of surface roughness was conducted by Nikuradse [127], who determined the friction factors for flow through artificially roughened pipes. The pipes were coated with various sizes of sand grains in dense arrays. Nikuradse found that the friction factor could be correlated with two parameters, Red and ks~d, where ks is the size of the sand grains. For a given roughness, the friction factor becomes independent of the Reynolds number when the latter is sufficiently large. Similar behavior is observed in a turbulent boundary layer over a rough flat plate, where Prandtl and Schlichting [128] showed that the important parameter is ks/x. The local skin friction coefficient is shown in Fig. 6.45 as a function of Re/with x/ks as a parameter. The lowest curve represents the skin friction for a smooth flat plate (Eq. 6.167). The other curves apply for rough plates with ueks/v and x/ks as parameters. The region above the dashed line is defined to be "fully rough." At a point on a moderately rough surface where x and x/ks are fixed and the Reynolds number is low, the skin friction coefficient is the same as on a smooth surface. An increase in velocity or Reynolds number causes the skin friction coefficient at that point to rise from its smoothsurface value. When the Reynolds number reaches a critical high value, the skin friction coefficient increases asymptotically to a constant value. The dashed line defines the critical Reynolds numbers, above which fully rough conditions exist. With uniform roughness and fixed unit Reynolds number, the skin friction coefficient along the plate behaves as indicated by the curves generally parallel to, but higher than, the smoothplate curve. Schlichting [65] gives the following correlations for local and average skin friction coefficients in the fully rough regime:
x/_~.~
cr= 2.87 + 1.58 In ks ]
_ (
x 25
q = 1.89 + 1.62 In ks ]
(6.230)
(6.231)
6.68
CHAFFER SIX
3
x 103
I X 104
.~
x 104
\ \ \
N I .x.lO3 ~ i
I0
3
~
~
I x 105
x 10 `5
3
X
~
~ .
%
o o o
'/
5 x I0 I
~
5
x/k,
ueks/v
~
"~'~

x/ks = const
2
.,
~
x 10z
5
....
x IOs
"~"
4
x 104 3
2.5
x I05 x 106
1.5
I
105
2
5
2
l0 s
5
2
107
5
2
10e
5
2
109
5
uexlv
FIGURE 6.45 Variation of local skin friction coefficient along a sandroughened flat plate [65]. (Reprinted by permission of McGrawHill.)
Effects of surface roughness are also evident in the boundary layer mean velocity profiles shown in Fig. 6.46. The profiles still exhibit a nearwall logarithmic behavior, but with a dependence on the roughness Reynolds number k ÷ = ksu*/v. The law of the wall for a rough surface may be written as u + =  1I n Y+ +5.0 _ Au+ K
30 _
I
I
I I I I II
I
I
(6.232)
I I I III I
I
I
I I I II~
I . t
oOO_O
20_
_
Smooth:u+=0~
In y+ + 5 . 0
OA?Ix~ o o
°i ~° °~° ° ~
_ooO _
_
000
~ o °°°° 70. For surfaces having a wide distribution of particle size, Au + remains nonzero even for small k ÷ based on average particle size. For this geometry, even a small number of large particles disturbs significantly the nearwall flow. For larger values of k ÷, Au+ becomes proportional to In k ÷ with the proportionality constant nearly equal to 1/~:. The law of the wall then becomes u ÷ = 1 In y + B*
25I 2(
• a ....
(6.233)
W L. Moore F.R. Homo ProndtISchlichting sond groin roughness Colebrook, White • 95% smooth, 5% Iorge groins O 48% smooth, 47% fine groins, 5% lorge groins o 9,5% uniform sond, 5% Iorge groins • 97.5% uniform sond, 2.,5% Iorge groins A uniform sond
....
I0
I00 k+
I000
I0,000
FIGURE 6.47 Effect of roughness size and type on universal velocity profiles in a turbulent boundary layer over a rough flat plate [130]. (Reprinted by permission of Hemisphere Publishing.)
where B* is a function of the surface geometry but independent of k + provided k + > 70. The behavior of the velocity profile (Eq. 6.233) and the constant value of the skin friction coefficient in the fully rough regime suggest that the integral parameters also attain unique values. Based on the data of Ref. 131, Kays and Crawford [74] recommend the following correlation for the skin friction coefficient on a fully rough flat plate: cI = 0.168 2 [In (864 0/ks)I:
(6.234)
where 0 is the m o m e n t u m thickness. Heat Transfer The Stanton number over a rough surface behaves similarly to the skin friction coefficient; at sufficiently high roughness Reynolds numbers k ÷, the Stanton number becomes independent of the freestream velocity. At a given Rex or Rea, roughness causes an increase in local Stanton number over the smoothplate value. These effects are shown in Fig. 6.48 for five values of the freestream velocity. The geometry of the rough surface used in these experiments was the densest array of spheres of radius r as shown in Fig.
6.70
CHAPTER
SIX
0.001 _11
!
I
I

Freestream

o ue:

0 ue : 139 f t / s
0.005
I
I
I
I
1 1 I!
!
!
1
1
velocity
32 ft/s
o ue : 9 0 f t / s 0
°~o~o s
A Ue = 190 f t / s ~" Ue = 2 4 2 f t / s
Smooth plate / St = 0.0153 Re~, ° 2 5
~
~
0.001
~1 1
1
1
I
1
1 1 11
I02
1
1
I
1
I 1 I I
I03

I04 ReF
FIGURE 6.48 Roughsurface Stanton number versus energythickness Reynolds number [129]. (Reprinted by permission of the authors.)
20.24 of Schlichting [65]. For a smooth plate, data for all five velocities collapse on a single correlation: St = 0 . 0 1 2 9 R e r °'25 Pr °'4 =
0.0153Rer°25
for air at room temperature
(6.235)
The effect of roughness is seen as an increase in the local Stanton number with increasing Ue at a fixed value of the Reynolds number based on energy thickness Rere. The roughness Reynolds number range corresponding to the five values of Ue is also shown in Fig. 6.48. The same data plotted as St against FE/r in Fig. 6.49, however, follow a single correlation showing no dependence on Ue. The Stanton number data of Ref. 131 in the fully rough regime on the same test surface can be correlated as St = 0.00317
F) °.175
(6.236)
in t h e r a n g e 1.5 < F/r < 10.0.
0.001
I
I
i
I
!
!
I
I
1
1
I ..... 1
I
I
I
I
Freestream velocity o ue= 3 2 f t l s

0.005
_
_
o ue : 9 0 f t l s
0 Ue : 139 f t / s a ue = 190 f t / s 
o~
~ ue : 242 f t l s
0.001
0.7
~.o
Fir
~o
FIGURE 6.49 Roughsurface Stanton number versus normalized energy thickness for a flatplate boundary layer [129]. (Reprinted by permission of the authors.)
FORCED
CONVECTION,
EXTERNAL
FLOWS
6.71
Streamwise Curvature. Streamwise surface curvature, e.g., on a highly cambered turbine blade, has a significant effect on the local rate of convection. Convex surfaces tend to reduce convection rates from those on a flat surface experiencing boundary layers with the same thickness and edge conditions, whereas concave surfaces tend to increase convection rates In laminar boundary layers, these effects can be evaluated by transforming the cartesian coordinates of the analysis to an orthogonal set with x representing the distance along the curved surface and y locally normal to the surface. The principal change in the governing equations is that a(). 1 Oy IS replaced by 1 + y/r~
~
o
a( ) Oy '
where rc is the local streamwise radius of curvature of the surface. Here, it is assumed that the boundary layer is very thin compared with re, which is positive for a convex surface and negative for a concave one. At sufficiently small values of 8/r~, the instabilities inherent in flow over a concave surface do not generate TaylorGOrtler vortices, and the change in convection, or skin friction, for positive or negative r~ is similar but of opposite sign. For boundary layers that are somewhat thicker, the next level of approximation requires additional terms identified with secondorder boundary layer theory; see Ref. 132, where it is shown that the skin friction coefficient for a laminar boundary layer on a curved surface is related to its flatplate counterpart by cf C/flat plate
 1  0.87______88
(6.237)
Fc
When TaylorGrrtler vortices develop, the boundary layer on a concave surface possesses an additional mixing mechanism and is capable of transferring heat and momentum at a greater rate than suggested by Eq. 6.237. The early heat transfer experiments by Thomann [133] demonstrated that a turbulent boundary layer behaves qualitatively the same as a laminar boundary layer, but that the magnitude of the curvature effect is about an order of magnitude greater. Bradshaw's careful review [134] of this topic identified the relatively large effect of the streamwise curvature in a turbulent boundary layer as not due to the geometric effects on the mean motion but due to extra rates of strain on the turbulence production. In this view, as the surface curvature changes, the local turbulence and mean motion are out of equilibrium, which implies a breakdown of the eddy viscosity concept. Others [135, 136] have retained the eddy viscosity approach by modifying it empirically to alter its magnitudea new mixing length formmand allowing for nonequilibrium of the mean motion and the turbulence through an empirical lag equation. This approach employs finite difference calculations of the boundary layer equations in partial differential form. It has been shown [137] that finite difference computations of the mean conservation equations that utilize the transport equations for the components of Reynolds stress lead to solutions for skin friction that represent curvature effects without model modifications but merely through geometric changes appropriate to thin layers. FreeStream Turbulence a n d Unsteadiness. It was shown in Fig. 6.27 that the freestream turbulence level significantly affects local and overall heat transfer from single cylinders in cross flow. This is caused primarily by early transition of the laminar boundary layer on the forward portion of the cylinder and subsequently by delayed separation of the turbulent boundary layer from the surface of the cylinder. Freestream turbulence and unsteadiness also affect, to varying degrees, the heat transfer behavior of a turbulent boundary layer in the absence of transitionpoint shift and separation. A summary of the available experimental data on the effects of freestream turbulence on heat transfer was presented by Kestin [138]. The experiments of Refs. 139141 showed that for freestream turbulence intensities ranging from 0.75 percent to about 4 percent, the skin friction and heat transfer coefficients at a fixed Rex remained practically unchanged. A later study by Hancock [142] revealed a significant dependence of the boundary layer momentum thickness on the freestream turbulence level. Consequently, significant effects were observed when skin friction data were compared at the same momentumthickness
6.72
CHAPTER SIX
Reynolds number Re0. The correlated data are shown in Fig. 6.50, where the increase in the skin friction is plotted versus a parameter that accounts for both the intensity and the scale of the freestream turbulence. The effect of freestream turbulence is primarily in the outer region of the boundary layer, where the law of the wake is modified, and, hence, the boundary layer integral thicknesses are modified. I
I
I
•

•
•
~

Preferred curve
•
/ / ° ,,,,/ , ° ;,,," .J/"
0.2
•
.
•
o
/¢ O.I

,,,
,,,"
il','," ,,',
/.v
ii \~ "
ix ..x~~¢xxl I
I
xx
1.0
Uae
u;" x I 0 0 / (
2.0
..~
+ 2.0)
FIGURE 6.50 Effect of freestream turbulence on skin friction for a fiatplate turbulent boundary layer [142]. (All details about different data points are given in Ref. 142.) (Reprinted by permission of the author~)
The problem of turbulent boundary layers with an oscillatory free stream has received considerable attention recently. Such flows are encountered with turbine blades, reciprocating cylinder walls, and helicopter rotor blades. The experiments of Refs. 143146 have shown that even at amplitudes as large as 40 percent of the mean and frequencies ranging from quasisteady to twice the bursting frequency, approximately ~ u/6, the mean velocity and turbulence intensity profiles in the boundary layer remain unaffected and are indeed the same as those measured with freestream velocity distributions held steady at the mean value. This shows that there is apparently no energy transfer between the imposed organized oscillations and the random turbulent fluctuations in the boundary layer. This being the case, the behavior of unsteady turbulent boundary layers can be predicted satisfactorily using turbulence models developed for steady flows [145].
TRANSITIONAL BOUNDARY LAYERS Transitional Boundary Layers for Uniform FreeStream Velocity Because the transition zone from a laminar to a turbulent boundary layer often covers a major portion of the exposed surface of a body, it is necessary to be able to predict the rapidly
FORCEDCONVECTION, EXTERNAL FLOWS 6.73 changing convection rate in this zone. The position of the onset of turbulence and the extent of the transition zone for a specific configuration depend on many factors such as the scale and spectral content of the freestream turbulence and sound field, the free stream Mach number, and the surface characteristics of smoothness, waviness, temperature, compliance, and mass transfer. To date, there is no universal correlation of these factors that will permit the prediction of the position and extent of the transition zone. What is presented here is a technique for predicting transitional boundary layer convection on a plate, given the position of the transition region. If the transition zone is not well known, one design approach is to arbitrarily assign a series of positions of the onset of turbulence and to set the length of the transition zone equal to the length of the fully laminar boundary layer. The sensitivity of the final design to changes in the position of transition must be determined; a high degree of sensitivity suggests the need for careful experimentation with prototype models. The contents of this section are an extension of the work of Ref. 147 to include the effects of variable fluid properties. The ideas employed are based on the observations of Ref. 148 that on a flat plate the distribution of 13, the fraction of time a surface point is covered by a fully turbulent boundary layer, is closely approximated by a Gaussian integral curve throughout the transition zone, i.e.,
~(x) = I x P(st) ds,_
(6.238)
(Y
%
where ~ is the standard deviation of the transition location about its mean ~t, and P(st) is the probability that transition initiates between st and st + dst:
P(st) = ~
1
[ a ( s ~  s t ) 2] exp  ~ N
(6.239)
Use of Eq. 6.239 requires that the transition take place sufficiently downstream so that the boundary layer is fully laminar at all times near the leading edge (st > 2~). If the term q';t(x, s,) is defined as the heat flux at point x in a turbulent boundary layer with instantaneous transition from laminar to turbulent flow at st, and q'~t(x) is defined as the heat flux at point x with a laminar boundary layer beginning at the leading edge, the heat flux caused by the intermittency of turbulence in the transition zone is then ,,
,,
~x e
"
ds,
qw(X) = [1  ~(x)]qwl(X) + 30 (st)qwt(X, st) .~
(6.240)
The first term is the product of the laminar heat flux and the fraction of time the boundary layer is laminar at x. The second term accounts for both the fraction of time the boundary layer is turbulent and the effect of the moving transition location. The term q",(x, st) is sufficiently complex mathematically that Eq. 6.240 is normally solved by numerical integration. If it is assumed that the energy thickness remains unchanged as the laminar boundary layer changes instantaneously into a turbulent boundary layer, then
q"t(x, s,)
{
st
qwt(X," O) = 1x where
I1
36.9
" cI RO (/Zl 5/4 \([[lpp i ~e)]}1/5 ~e t,]~J~' 5/8\peUeSt 3/8
~rll3 \zt j
z~ _ ie + Pr 1/2 (Ue2/2)  iw Zt ie + Pr 1/3 (u 2/2)  iw
TM
(6.241)
(6.242)
The form of Eq. 6.241 applies for Re/, < 4 x 10 6, where the Blasius skin friction equation (Eq. 6.16) is reasonably accurate and 2 S t / Q = P r 2/3 a n d P r °4 for laminar and turbulent flow, respectively. It also uses the laminar reference enthalpy approach to define l.t'p'/~ePe (see the section on uniform freestream conditions) and uses the turbulent boundary layer transfor
6.74
CHAPTER SIX
mations Fc and FRO, which are assumed insensitive to Reynolds number variations (see the section on ideal gases at high temperature). Thus, given Mae, Te, Tw, st, G, Eq. 6.240 provides the distribution of heat flux in the prescribed transition zone by techniques consistent with those for the fully laminar and fully turbulent boundary layers given previously.
COMPLEX CONFIGURATIONS The material presented earlier was confined to steadystate flows over simply shaped bodies such as flat plates, with and without pressure gradients in the streamwise direction, or stagnation regions on blunt bodies. The simplicity of these flow configurations allows reduction of the problems to the solution of steadystate ordinary differential equations. The evaluation of convective heat transfer to more complex threedimensional configurations, characteristic of real aerodynamic vehicles, involves the solution of partial differential equations. Even when the latter are confined to steadystate problems, they require extensive use of computers in the solution of finite difference or finite element formulations. Nonsteady flows further complicate the problems by introducing another dimension, namely, time. Recent years have shown considerable progress in the development of methods for solving these more complex problems. Larger and faster computers have become more accessible and solution algorithms are more efficient. Complex flow fields undergoing chemical reactions between many species are being performed routinely. For simpler configurations, timedependent calculations of the dynamic behavior of chaotic turbulent flows have been performed to provide numerical experiments with much more detail than can be provided by physical experiments. With regard to accuracy, laminar flows can be solved accurately provided care is given to specifying a computational mesh that can resolve shock waves and/or the regions very near surfaces. The introduction of surface mass transfer or nonuniform streamwise surface temperatures is straightforward. With transitional or turbulent boundary layers, however, the state of the art is less satisfactory. A truly predictive method of computing the flow over practical threedimensional shapes, where the laminar, transitional, and turbulent boundary portions of the flow have been computed in a timedependent manner and with mesh spacing that resolves all the significant eddies, is not currently available and is likely to be too expensive for general engineering use far into the foreseeable future. Practical mainstream studies of turbulent flows have been confined to steadystate computations dealing only with mean motions. Experimental guidance has been used to define the transition regions and the turbulent regions have been computed with a hierarchy of turbulence models ranging from the simple algebraic eddy viscosity models presented here to secondorder models where the concept of the eddy viscosity can be dropped and the Reynolds stresses themselves are evaluated with field equations. Secondorder turbulence models depend on partial differential equations describing measures of the turbulence, i.e., the kinetic energy, the dissipation rate, and/or the individual Reynolds stresses. Thus the number of field equations defining a flow field is increased significantly. Twoequation models that define the eddy viscosity in terms of the turbulence kinetic energy and the dissipation rate (or a dissipation rate per unit kinetic energy) add two partial differential equations to the system and represent a 20 to 40 percent increase in computer storage needs; in addition, these models often reduce convergence rates because of their stiffness. Reynolds stress computations more than double computational costs for similar reasons. Despite their complexity and cost, the secondorder models become advantageous for flows involving steep pressure gradients, separated boundary layers, or surface curvature. The models also often lack universality, working very well for one or more flows but then failing when applied to some other type of flow. As a consequence, codes with particular turbulence models must be verified by comparing their results with data from physical or numerical experiments for similar flow fields. Only when an algorithm, a mesh field, and a
FORCED CONVECTION, EXTERNAL FLOWS
6.75
turbulence model have been verified for a particular shape and set of flow conditions can the technique be applied with confidence to other generally similar configurations and freestream flow conditions. In their early stages of development, these codes should be applied to flat plates and stagnation regions and compared to the timetested methods shown in the previous sections as a first step in the code verification process. Also, because computations with these codes are costly, they are used sparingly. For example, in the computation of the total heat transferred to a body on a trajectory, standard practice is to employ these codes at specific, but widely separated, intervals and then to interpolate for the times between these solutions with guidance for the influence of various parameters from formulas such as those presented earlier.
NOMENCLATURE Symbol, Definition, SI Units, English Units A A* A÷ a* 11" 11
am B B* B B*
8i Bh Bm b b* b* bl C
Ce Cew C,. C* cl cr cm
Ch Cmi
constant, Eqs. 6.119, 6.142, and 6.201 coefficient, Eq. 6.132, Table 6.5 van Driest wall damping parameter, Eq. 6.178 coefficient, Eq. 6.124 constant in reference enthalpy method exponent, Eq. 6.122, Table 6.3 molecular weight ratio exponent, Fig. 6.44 constant, Eq. 6.142 function, Eq. 6.201, Fig. 6.39 coefficient, Eq. 6.132, Table 6.5 coefficient in law of the wall, Eq. 6.148 surface geometry function, Eq. 6.233 blowing function, Eq. 6.218 blowing function, Eq. 6.225 blowing function, compressible flow, Eq. 6.140 exponent in viscosity law for liquids coefficient, Eq. 6.124 constant in reference enthalpy equation blowing function, Eq. 6.220 constant value of Cr, Eq. 6.43 normalized viscositydensity product referred to boundary layer edge conditions ChapmanRubesin constant, Eq. 6.53 normalized viscositydensity product referred to conditions at r, Eq. 6.39 constant in reference enthalpy equation local skin friction coefficient average skin friction coefficient local skin friction coefficient in terms of wall properties, Eq. 6.105 coefficient, Eqs. 6.143 and 6.144 mass transfer coefficient for species i
6.76
CHAPTER SIX
cp Cv D2 ~b 5~o d d* dr E
Ec F
F~
FRx FRo Fo f f(x) H h h h= I
i I
i
J K, k k+ k
ks L L L,
Le l /+ M
M~ Ma m m*
specific heat at constant pressure: J/(kg.K), Btu/(lbm'°F) specific heat at constant volume: J/(kg.K), Btu/(lbm°F) cross flow forced convection coefficient, Table 6.7 diffusion coefficient: m2/s, ft2/s binary diffusion coefficient: m2/s, ftE/s body or tube diameter: m, ft exponent in powerlaw velocity profile, Table 6.4 wind tunnel diameter or height: m, ft coefficient for turbulent boundary layer, Eq. 6.175 Eckert number, V2/2i dimensionless boundary layer stream function, Eq. 6.41 turbulent boundary layer transformation function, ?s/cl, Table 6.8 turbulent boundary layer transformation function, Rexe/Rexe, Eq. 6.209 turbulent boundary layer transformation function, Re0e/Re0e, Table 6.8 function, Table 6.6: (W/m2)[m/(N.m2)] in, [Btu/(s.ft2)](ft/atm) la, Eq. 6.134 Blasius stream function, Eq. 6.12 denotes function of x form factor in twodimensional flow, 8*/0 local heat transfer coefficient: W/(m2.K), Btu/(h.ft2.°F) average heat transfer coefficient for row of tubes: W/(m2.K), Btu/(hft 2.°F) average heat transfer coefficient for infinite array of tubes: W/(mE.K), Btu/(h.ft 2°F) total enthalpy per unit mass: J/kg, Btu/lbm, I = i + (u 2 + v 2 + w2)/2 normalized total enthalpy, Eq. 6.39 normalized total enthalpy, Eq. 6.45 enthalpy per unit mass: J/kg, Btu/lbm diffusion flux: kg/(m2.s), lbm/(ftE.s) mass concentration of species i thermal conductivity: W/(m.K), Btu/(s.ft.°R) roughness Reynolds number, wall layer coordinates index, Eq. 6.76 roughness size: m, ft length of porous section in film cooling: m, ft reference length, e.g., Eq. 6.76 and Eq. 6.89: m, ft freestream turbulence length scale, Fig. 6.50: m, ft Lewis number, pSOcp/k, or ~o/~i~ mixing length: m, ft normalized mixing length, lu*/v molecular weight: kg/kmol, lbm/lbmmol mean molecular weight of coolant: kg/kmol, lbm/lbmmol Mach number exponent, Eq. 6.119 constant, Eq. 6.142
FORCED CONVECTION, EXTERNAL FLOWS Nu
Nusselt number, St ReL Pr
Nu
mean or average Nusselt number, St Ret Pr number of species constant, Eq. 6.142 Deissler empirical constant exponent, Table 6.7 pressure: N/m E, lbf/ft 2 (atm)
n n* nD nl
P
P( ) Pr PrF PrM Prr Pr, q,,
probability function of transition, Eq. 6.239 Prandtl number, ktcp/k, single gas or liquid frozen Prandtl number, gas properties weighted over all species present effective Prandtl number, Eq. 6.182 Prandtl number based on total properties for gas in chemical equilibrium turbulent Prandtl number, eM/e, heat flux: W/m E, Btu/(s.ft 2)
q"
average heat flux (over space): W/m E, Btu/(sft 2)
R
enthalpy profile function, Eq. 6.195 universal gas constant: mE/(s2.K) or J.kmoll.K 1, ft2/(s 2°R)
R
Reynolds number based on diameter d effective Reynolds number, Eq. 6.202 Reynolds number based on length L; L = x, 0, 5*, 5, or F ReL r radius of spherical roughness: m, ft r dimensionless temperature variable, Eq. 6.28 r(O) recovery factor streamwise radius of curvature: m, ft rc !,. nose radius: m, ft S enthalpy profile function, Eq. 6.181 St local Stanton number, h/pcpUe St average Stanton number, h/pcpue St (x, s) local Stanton number on a plate with a temperature jump at x  s S distance to location of surface temperature discontinuity: m, ft Sj downstream of surface temperature jump: m, ft sj upstream of surface temperature jump: m, ft distance from leading edge to transition from laminar to turbulent flow: m, ft St T temperature: K, °R constant in viscosity law for liquids: K, °R Tc T~ Sutherland constant: K Tu intensity of turbulence, U'/Ve porous section temperature Tw~ t time: s normalized local enthalpy, Eq. 6.91 t+ normalized temperature parameter, Eq. 6.184 normalized freestream enthalpy, Eq. 6.93 ~e Red
Reeff
n
+
6.77
6.78
C H A P T E R SIX
U. U U U* U+ AU + U Ue t Ue Ui
V. 1;
v~ W W W X X
Xeff Xi Xt X+
Yo(rl) Y y+
normalized freestream enthalpy, Eq. 6.92 velocity component normal to leading edge of yawed cylinder, Fig. 6.18 normalized freestream velocity: V../(10,000 ft/s), V~./(3048 m/s) velocity component in x direction: m/s, ft/s friction velocity, V~w/p: m/s, ft/s normalized velocity, u/u* shift in normalized velocity due to roughness internal energy per unit mass: J/kg, Btu/lbm freestream velocity component in the x direction: m/s, ft/s freestream RMS fluctuation of velocity: m/s, ft/s velocity component in the xi direction: m/s, ft/s freestream velocity: m/s, ft/s velocity component in the y direction: m/s, ft/s velocity component in the y direction at surface normalized by u* velocity component in the z direction: m/s, ft/s normalized cross flow velocity, Eq. 6.90 Coles wake function, Fig. 6.33 rectangular coordinate: m, ft dimensionless tube spacing in the freestream direction effective distance, laminar flow, Eq. 6.110 generalized orthogonal coordinate: m, ft dimensionless tube spacing normal to the freestream direction normalized Xl distance, Eq. 6.187 dimensionless temperature function, Eq. 6.19 rectangular coordinate, normal to surface: m, ft normalized distance from wall, u*y/v
Greek Symbols
m
F F Au +
6r E qED i
van Driest function, Table 6.8 fraction of time a surface point is covered by a fully turbulent boundary layer, Eq. 6.238; van Driest function, Table 6.8 Prandtl number function, Eq. 6.191 pressure gradient parameter, Eq. 6.94 normalized temperature parameter, Eq. 6.200 function, Eq. 6.192 energy thickness, Eq. 6.228b: m, ft roughness effect on velocity profile boundary layer thickness: m, ft; or incremental value displacement thickness, twodimensional flow, Eq. 6.150: m, ft thermal boundary layer thickness: m, ft sum of molecular kinematic and eddy viscosities turbulent eddy diffusivity for species i: m2/s, ft2/s
F O R C E D C O N V E C T I O N , E X T E R N A L FLOWS
~H
EM
11 q~ rlH 0 0c 00 K
A P p
9 D
t~ 1; "l;w
thermal eddy diffusivity: m2/s, ft2/s normalized thermal eddy diffusivity, Eq. 6.187 momentum eddy diffusivity; momentum eddy diffusivity in the x direction: m2/s, ft2/s normalized momentum eddy diffusivity, Eq. 6.187 transformed boundary layer variable, Eqs. 6.11, 6.36 and 6.89 function, Eq. 6.89 transformed boundary layer variable, Eqs. 6.11, 6.36, and 6.89 modified transformed boundary layer variable, Eq. 6.43 thermal boundary layer thickness parameter, Figs. 6.2 and 6.3 momentum thickness, twodimensional flow, Eq. 6.151: m, ft cone angle: tad, deg ratio of Sutherland constant to edge temperature von K~irmfin mixing length constant, Eq. 6.148 sweepback angle: rad, deg dynamic viscosity: kg/(m.s), lbm/(ft's) kinematic viscosity: m:/s, ft2/s parameter in Coles wake velocity distribution, Eq. 6.149 density: kg/m 3, lbm/ft 3 standard deviation of transition location about its mean shear stress between fluid layers: N/m 2, lbf/ft 2 shear stress at wall: N/m 2, lbf/ft 2 stream function: m2/s, ft2/s, Eq. 6.10
Subscripts aw C
d e
eft F i i i
] l r, ref st T t w
0
6.79
adiabatic wall conditions initial coolant condition in reservoir diameter evaluated at the boundary layer edge effective value frozen properties component i of a mixture constantproperty case defining orthogonal direction, Eq. 6.104 component y of a mixture laminar reference condition stagnation point on body total properties turbulent wall zero mass transfer
6.80
CHAPTER
SIX
1
c o m p o n e n t 1 ( c o o l a n t ) o f mixture
2
c o m p o n e n t 2 (air) of mixture
oo
freestream conditions
Superscripts m e a n value '
r a n d o m fluctuating value; d u m m y variable, reference property, ordinary derivative
*
properties to be evaluated at reference enthalpy or t e m p e r a t u r e condition

constantproperty case
REFERENCES 1. H. Blasius, "The Boundary Layer in Fluids With Little Friction," NACA Tech. Mem. 1256, 1950, transl, of "Grenzschichten in Fl0ssigkeiten mit kleiner Reibung," Z. Math. Phys. (56): 137, 1908. 2. E. Pohlhausen, "Der W~irmeaustausch zwischen festen Krrpern und F10ssigkeiten mit kleiner Reibung und kleiner W~irmeleitung," Z. angew. Math. Mech. (1): 115121, 1921. 3. E. R. G. Eckert and O. Drewitz, "The Heat Transfer to a Plate in Flow at High Speed," NACA Tech. Mem. 1045, 1943, transl, of "Der W~irme0bergang an eine mit grosser Geschwindigkeit l~ings angestrrmte Platte," Forschung auf dem Gebiete des Ingenieurwesens (11): 116124, 1940. 4. G. W. Morgan and W. H. Warner, "On Heat Transfer in Laminar Boundary Layers at High Prandtl Numbers," J. Aeronaut. Sci. (23): 937948, 1956. 5. L. Crocco, "Lo strato limite laminare nei gas," Monografie Sci. di Aeronaut. 3, Rome, 1946, transl, as North American Aviation Aerophys. Lab. Rep. APL/NAA/CF1038, 1948. 6. A. P. Colburn, "A Method for Correlating Forced Convection Heat Transfer Data and a Comparison With Fluid Friction," Trans. Am. Inst. Chem. Eng. (29): 174210, 1933. 7. R.A. Seban, "Laminar Boundary Layer of a Liquid With Variable Viscosity," in Heat Transfer Thermodynamics and Education, Boelter Anniversary Volume, H. A. Johnson ed., pp. 319329, McGrawHill, New York, 1964. 8. L. Howarth, "Concerning the Effect of Compressibility in Laminary Boundary Layers and Their Separation," Proc. Roy Soc. London (194): 1642, 1948. 9. D. R. Chapman and M. W. Rubesin, "Temperature and Velocity Profiles in the Compressible Laminar Boundary Layer With Arbitrary Distribution of Surface Temperature," J. Aeronaut. Sci. (16): 547565, 1949. 10. T. von K~irm~in, "The Problem of Resistance in Compressible Fluids," Atti del Convegno della Fondazione Alessandro Volta 1935, 223326, 1936. 11. E. R. van Driest, "Investigation of Laminar Boundary Layer in Compressible Fluids Using the Crocco Method," NACA Tech. Note 2597, 1952. 12. G. B. W. Young and E. Janssen, "The Compressible Boundary Layer," J. Aeronaut. Sci. (19): 229236, 288, 1952. 13. E. R. van Driest, "The Laminar Boundary Layer With Variable Fluid Properties," North American Aviation Rep. AL1866, Los Angeles, 1954. 14. M. E Romig and E J. Dore, "Solutions of the Compressible Laminar Boundary Layer Including the Case of a Dissociated Free Stream," Convair Rep. ZA7O12, San Diego, 1954. 15. R. E. Wilson, "RealGas LaminarBoundaryLayer Skin Friction and Heat Transfer," 1 Aerosp. Sci. (29): 640647, 1962. 16. N. B. Cohen, "BoundaryLayer Similar Solutions and Correlation Equations for Laminar Heat Transfer Distribution in Equilibrium Air at Velocities up to 41,000 Feet per Second," NASA Tech. Rep. R118, 1961. 17. G. T. Chapman, "Theoretical Laminar Convective Heat Transfer and Boundary Layer Characteristics on Cones at Speeds of 24 km/sec," NASA Tech. Note D2463, 1964.
FORCED CONVECTION, EXTERNAL FLOWS
6.81
18. M. W. Rubesin and H. A. Johnson, "A Summary of Skin Friction and Heat Transfer Solutions of the Laminar Boundary Layer on a Flat Plate," Proc. 1948 Heat Transfer Fluid Mech. Inst.; also Trans. A S M E (71): 383388, 1949. 19. A. D. Young, "Boundary Layers," in Modern Developments in Fluid Dynamics: High Speed Flow, L. Howarth ed., vol. 1, chap. 10, p. 422, Oxford University Press, New York, 1953. 20. E. R. G. Eckert, "Survey on Heat Transfer at High Speeds," Wright Air Development Center Tech. Rep. 5470, Dayton, 1954. 21. E. R. G. Eckert, "Engineering Relations for Heat Transfer and Friction in High Velocity Laminar and Turbulent Boundary Layer Flow Over Surfaces With Constant Pressure and Temperature," Trans. A S M E (78): 12731283, 1956. 22. M.W. Rubesin, "The Effect of an Arbitrary Surface Temperature Variation along a Flat Plate on the Convective Heat Transfer in an Incompressible Turbulent Boundary Layer," NACA Tech. Note 2345, 1951. 23. H. S. Carslaw and J. C. Jaeger, Conduction of Heat in Solids, 2d ed., Oxford University Press, New York, 1959. 24. R. Bond, "Heat Transfer to a Laminar Boundary Layer With Nonuniform Free Stream Velocity and Nonuniform Wall Temperature," Univ. California Inst. Eng. Res. Rep., University of California, Berkeley, CA, 1950. 25. M.J. Lighthill, "Contributions to the Theory of Heat Transfer Through a Laminar Boundary Layer," Proc. Roy. Sci. London (202): 359377, 1950. 26. R. Eichhorn, E. R. G. Eckert, and A. D. Anderson, "An Experimental Study of the Effects of Nonuniform Wall Temperature on Heat Transfer in Laminar and Turbulent Axisymmetric Flow Along a Cylinder," Wright Air Development Center Tech. Rep. 5833, Dayton, 1958. 27. B. M. Leadon, "The Status of Heat Transfer Control by Mass Transfer for Permanent Structures," in Aerodynamically Heated Structures, P. E. Glaser ed., p. 171, PrenticeHall, Englewood Cliffs, NJ, 1962. 28. H.W. Emmons and D. Leigh, "Tabulation of the Blasius Function With Blowing and Suction," Harvard Univ. Combust. Aerodynamics Lab. Tech. Rep. 9, Harvard University, Cambridge, MA, 1953. 29. J. P. Hartnett and E. R. G. Eckert, "Mass Transfer Cooling in the Laminar Boundary Layer With Constant Fluid Properties," Trans. A S M E (79): 247254, 1957. 30. G. M. Low, "The Compressible Laminar Boundary Layer with Fluid Injection," NACA Tech. Note 3404, 1955. 31. J. E Gross, J. P. Hartnett, D. J. Masson, and C. Gazley Jr., "A Review of Binary Laminar Boundary Layer Characteristics," Int. J. Heat Mass Transfer (3): 198221, 1961. 32. E. R. G. Eckert, A. A. Hayday, and W. J. Minkowycz, "Heat Transfer, Temperature Recovery and Skin Friction on a Flat Plate With Hydrogen Release into a Laminar Boundary Layer," Int. J. Heat Mass Transfer (4): 1729, 1961. 33. P. A. Libby and P. Sepri, "Laminar Boundary Layer With Complex Composition," Phys. Fluids (10): 21382146, 1967. 34. P. A. Libby and K. Chen, "Laminar Boundary Layer with Uniform Injection," Phys. Fluids (8): 568574, 1965. 35. R. Iglisch, "Exact Calculations of Laminar Boundary Layers in Longitudinal Flow Over a Flat Plate With Homogeneous Suction," NACA Tech. Mem. 1205, 1949, transl, of "Exakte Berechnung der laminaren Grenzschicht an der 1/ingsangestr6mten ebenen Platte mit homogener Absaugung," Schriften der Deutschen Akademie der Luftfahrtforschung, Band 8 B, Heft 1, 1944. 36. H. G. Lew and J. B. Fanucci, "On the Laminar Compressible Boundary Layer Over a Flat Plate With Suction or Injection," J. Aeronaut. Sci. (22): 589597, 1955. 37. A. J. Pallone, "Nonsimilar Solutions of the Compressible Laminar Boundary Layer Equations With Applications to the Upstream Transpiration Cooling Problem," J. Aerosp. Sci. (28): 449456, 492, 1961. 38. J. T. Howe, "Some Finite Difference Solutions of the Laminar Compressible Boundary Layer Showing Effects of Upstream Transpiration Cooling," NASA Mem. 22659A, 1959. 39. W. Mangler, "Zusammenhang zwischen ebenen und rotationssymmetrischen Grenzschichten in kompressiblen Fltissigkeiten," Z. angew. Math. Mech. (28): 97103, 1948.
6.82
CHAPTER SIX
40. P. A. Libby, "Laminar Boundary Layer on a Cone With Uniform Injection," Phys. Fluids (8): 22162218, 1965. 41. I. E. Beckwith and N. B. Cohen, "Application of Similar Solutions to Calculations of Laminar Heat Transfer on Bodies With Yaw and Large Gradient in High Speed Flow," NASA Tech. Note D625, 1961. 42. C. B. Cohen and E. Reshotko, "Similar Solutions for the Compressible Laminar Boundary Layer With Heat Transfer and Pressure Gradient," NACA Rep. 1293, 1956. 43. E H. Clauser, "Turbulent Boundary Layers in Adverse Pressure Gradients," J. Aeronaut. Sci. (21): 91108, 1954. 44. L. Lees, "Laminar Heat Transfer Over BluntNosed Bodies at Hypersonic Flight Speeds," Jet Propulsion (26): 259269, 1956. 45. D. R. Hartree, "On an Equation Occurring in Falkner and Skan's Approximate Treatment of the Equation of the Boundary Layer," Proc. Cambridge Philosoph. Soc. (33): 223239, 1937. 46. A. N. Tifford, "The Thermodynamics of the Laminar Boundary Layer of a Heated Body in a HighSpeed Gas Flow Field," J. Aeronaut. Sci. (12): 241251, 1945. 47. S. Levy, "Heat Transfer to ConstantProperty Laminar Boundary Layer Flows With PowerFunction FreeStream Velocity and WallTemperature Variation," J. Aeronaut. Sci. (19): 341348, 1952. 48. I. E. Beckwith, "Similar Solutions for the Compressible Boundary Layer on a Yawed Cylinder With Transpiration Cooling," NASA Tech. Rep. R42, 1959. 49. D. R. Davies and D. E. Bourne, "On the Calculation of Heat and Mass Transfer in Laminar and Turbulent Boundary Layers, I: The Laminar Case," Q. J. Mech. Appl. Math. (9): 457466, 1956. 50. A. N. Tifford and S. T. Chu, "Heat Transfer in Laminar Boundary Layers Subject to Surface Pressure and Temperature Distributions," Proc. 2d Midwestern Conf. Fluid Mech., p. 363, 1952. 51. A. Pallone and W. Van Tassell, "Stagnation Point Heat Transfer for Air in the Ionization Regime," A R S J. (32): 436437, 1962. 52. H. Hoshizaki, "Heat Transfer in Planetary Atmospheres at SuperSatellite Speeds," A R S J. (32): 15441551, 1962. 53. J. G. Marvin and G. S. Deiwert, "Convective Heat Transfer in Planetary Gases," NASA Tech. Rep. R224, 1965. 54. J. G. Marvin and R. B. Pope, "Laminar Convective Heating and Ablation in the Mars Atmosphere," A I A A J. (5): 240248, 1967. 55. P. DeRienzo and A. J. Pallone, "Convective StagnationPoint Heating for Reentry Speeds up to 70,000 fps Including Effects of Large Blowing Rates," A I A A J. (5): 193200, 1967. 56. C. E Hansen, "Approximations for the Thermodynamic and Transport Properties of HighTemperature Air," NASA Tech. Rep. R50, 1959. 57. S. Bennett, J. M. Yos, C. F. Knopp, J. Morris, and W. L. Bade, "Theoretical and Experimental Studies of HighTemperature Gas Transport Properties," AVCO Corp. RADTR657, 1965. 58. W. E Ahtye, "A Critical Evaluation of Methods for Calculating Transport Coefficients of Partially Ionized Gas," NASA Tech. Mem. X54, 1964. 59. R. W. Rutowski and K. K. Chan, "Shock Tube Experiments Simulating Entry Into Planetary Atmospheres," Lockheed Missiles and Space Co. LMSD 288139, vol. 1, part 2, 1960. 60. J. S. Gruszczynski and W. R. Warren, "Measurements of Hypervelocity Stagnation Point Heat Transfer in Simulated Planetary Atmospheres," General Electric Space Sci. Lab. R63SD29, 1963. 61. L. Yee, H. E. Bailey, and H. T. Woodward, "Ballistic Range Measurements of StagnationPoint Heat Transfer in Air and Carbon Dioxide at Velocities up to 18,000 feet per second," NASA Tech. Note D777, 1961. 62. R. M. Nerem, C. J. Morgan, and B. C. Graber, "Hypervelocity Stagnation Point Heat Transfer in a Carbon Dioxide Atmosphere," A I A A J. (1): 21732175, 1963. 63. S. J. Kline, Proceedings of the AFOSRHTTMStanford Conference on Complex Turbulent Flows, Stanford University Department of Mechanical Engineering, Stanford, CA, 1982. 64. R. Hilpert, "W~irmeabgabe von geheizten Dr~ihten und Rohren in Luftstrom," Forsch. Ingenieurwes. (4): 215224, 1933. 65. H. Schlichting, Boundary Layer Theory, 6th ed., McGrawHill, New York, 1968.
FORCED CONVECTION, EXTERNAL FLOWS
6.83
66. V. T. Morgan, "The Overall Convective Heat Transfer From Smooth Circular Cylinders," in Advances in Heat Transfer, T. E Irvine Jr., and J. P. Hartnett eds., vol. 11, pp. 199264, Academic, New York, 1975. 67. A. Zukauskas, "Heat Transfer From Tubes in Cross Flow," in Advances in Heat Transfer, J. P. Hartnett and T. E Irvine Jr. eds., vol. 8, pp. 93160, Academic, New York, 1972. 68. E. Schmidt and K. Wenner, "W/armeabgabe tiber den Umfang eines angeblasenen geheizten Zylinders," Forsch. Ingenieurwes. (12): 6573, 1941. 69. J. Kestin and P. E Maeder, "Influence of Turbulence on Transfer of Heat from Cylinders," NACA TN 4018, 1954. 70. W. M. Kays and A. L. London, Compact Heat Exchangers, 2d ed., McGrawHill, New York, 1964. 71. J. O. Hinze, Turbulence: An Introduction to Its Mechanism and Theory, 2d ed., McGrawHill, New York, 1975. 72. E Bradshaw, ed., Turbulence, SpringerVerlag, New York, 1976. 73. H. Tennekes and J. L. Lumley, A First Course in Turbulence, MIT Press, Cambridge, MA, 1972. 74. W. M. Kays and M. E. Crawford, Convective Heat and Mass Transfer, 2d ed., McGrawHill, New York, 1980. 75. E. R. van Driest, "Turbulent Boundary Layer in Compressible Fluids," J. Aeronaut. Sci. (18): 145160, 216, 1951. 76. B. E. Launder and D. B. Spalding, Mathematical Models of Turbulence, Academic, New York, 1972. 77. W. Rodi, "Turbulence Models and Their Applications in Hydraulics," International Association for Hydraulic Research StateoftheArt Paper, Delft, the Netherlands, June 1980. 78. J. Kestin and P. D. Richardson, "Heat Transfer Across Turbulent Incompressible Boundary Layers," Int. J. Heat Mass Transfer (6): 147189, 1963. 79. E H. Clauser, "The Turbulent Boundary Layer," in Advances in Applied Mechanics, H. L. Dryden et al. eds., vol. 4, Academic, New York, 1956. 80. D. Coles, "The Law of the Wall in Turbulent Shear Flow," in Sonderdruck aus 50 Jahre Grenzschichtforschung, H. Goertler and W. Tollmien eds., Friedrich Vieweg & Sohn, Brunswick, Germany, 1955. 81. D. Coles, "The Law of the Wake in Turbulent Boundary Layer," J. Fluid Mech. (1): 191226, 1956. 82. H. Ludwieg and W. Tillman, "Investigation of the Wall Shearing Stress in Turbulent Boundary Layers," NACA Tech. Mem. 1285, 1959, transl, of "Untersuchungen fiber die Wandschubspannung in turbulenten Reibungsschichten," Ing. Arch. (17): 288299, 1949. 83. E SchultzGrunow, "New Frictional Resistance Law for Smooth Plates," NACA Tech. Mem. 986, 1941, transl, of "Neues Widerstandsgesetz far glatte Platten," Luftfahrforsch. (17): 239246, 1940. 84. D. Coles, "The Young Person's Guide to the Data," in Proceedings of the Computation of Turbulent Boundary Layers1968, AFOSRIFPStanford Conf. 1968, D. Coles and E. A. Hirst eds., Stanford University, Stanford, CA, 1968. 85. G. B. Schubauer and C. M. Tchen, "Turbulent Flow," in Turbulent Flows and Heat Transfer, C. C. Lin ed., sec. B, pp. 119122, Princeton University Press, Princeton, NJ, 1959. 86. D. W. Smith and J. H. Walker, "SkinFriction Measurements in Incompressible Flow," NASA Tech. Rep. R26, 1959. 87. T. Cebeci and A. M. O. Smith, Analysis of Turbulent Boundary Layers, Academic, New York, 1974. 88. T. von Kfirmfin, "The Analogy Between Fluid Friction and Heat Transfer," Trans. A S M E (61): 705710, 1939. 89. D. B. Spalding, "Heat Transfer to a Turbulent Stream From a Surface With a Stepwise Discontinuity in Wall Temperature, International Developments in Heat Transfer," in Conf. Int. Dev. Heat Transfer, part 2, pp. 439446, ASME, New York, 1961. 90. E. R. van Driest, "On Turbulent Flow Near a Wall," J. Aeronaut. Sci. (23): 10071011, 1956. 91. W. M. Kays, R. J. Moffat, and W. H. Thielbahr, "Heat Transfer to the Highly Accelerated Turbulent Boundary Layer With and Without Mass Addition," J. Heat Transfer (92): 499505, 1970. 92. P. S. Klebanoff, "Characteristics of Turbulence in a Boundary Layer With Zero Pressure Gradient," NACA Tech. Note 3178, 1954.
6.84
CHAPTER SIX
93. E. R. van Driest, "The Turbulent Boundary Layer With Variable Prandtl Number," in Sonderdruck aus 50 Jahre Grenzschichtforschung," H. Goertler and W. Tollmien eds., Friedrich Vieweg & Sohn, Brunswick, Germany, 1955. 94. R. G. Deissler, "Analysis of Turbulent Heat Transfer, Mass Transfer, and Friction in Smooth Tubes at High Prandtl and Schmidt Numbers," NACA Rep. 1210, 1954. 95. M.W. Rubesin, "A Modified Reynolds Analogy for the Compressible Turbulent Boundary Layer on a Flat Plate," NACA Tech. Note 2917, 1953. 96. C. Ferrari, "Effect of Prandtl Number on the Heat Transfer Properties of a Turbulent Boundary Layer When the Temperature Distribution Along the Wall Is Arbitrarily Assigned," Z. angew. Math. Mech. (36): 116135, 1956. 97. D. B. Spalding, "Contribution to the Theory of Heat Transfer Across a Turbulent Boundary Layer," Int. J. Heat Mass Transfer (7): 743761, 1964. 98. D. B. Spalding, "A Unified Theory of Friction, Heat Transfer, and Mass Transfer in the Turbulent Boundary Layer and Wall Jet," Aeronaut. Res. Council (England) ARCCP829, 1965. 99. B. E Blackwell, W. M. Kays, and R. J. Moffat, "The Turbulent Boundary Layer on a Porous Plate: An Experimental Study of the Heat Transfer Behavior With Adverse Pressure Gradients," Stanford Univ. Dept. Mech. Eng. Rep. HMT16, Stanford University, Stanford, CA, August 1972. 100. R. A. Antonia, "Behaviour of the Turbulent Prandtl Number Near the Wall," Int. J. Heat Mass Transfer (23): 906908, 1980. 101. B. A. Kader, "Temperature and Concentration Profiles in Fully Turbulent Boundary Layers," Int. J. Heat Mass Transfer (24): 15411544, 1981. 102. W. C. Reynolds, W. M. Kays, and S. J. Kline, "Heat Transfer in the Turbulent Incompressible Boundary Layer, ImConstant Wall Temperature," NASA Mem. 12158W, 1958. 103. A. Seiff, "Examination of the Existing Data on the Heat Transfer of Turbulent Boundary Layers at Supersonic Speeds From the Point of View of Reynolds Analogy," NACA Tech. Note 3284, 1954. 104. D. B. Spalding and S. W. Chi, "The Drag of a Compressible Turbulent Boundary Layer on a Smooth Flat Plate With and Without Heat Transfer," J. Fluid Mech. (18): 117143, 1964. 105. E. R. van Driest, "The Problem of Aerodynamic Heating," Aeronaut. Eng. Rev. (15): 2641, 1956. 106. D. Coles, "The Turbulent Boundary Layer in a Compressible Fluid," Rand Corp. Rep. R403PR, Santa Monica, CA, 1962. 107. S. C. Sommer and B. J. Short, "FreeFlight Measurements of Turbulent Boundary Layer Skin Friction in the Presence of Severe Aerodynamic Heating at Mach Numbers From 2.8 to 7.0," NACA Tech. Note 3391, 1955. 108. E. J. Hopkins, M. W. Rubesin, M. Inouye, E. R. Keener, G. G. Mateer, and T. E. Polek, "Summary and Correlation of SkinFriction and HeatTransfer Data for a Hypersonic Turbulent Boundary Layer on Simple Shapes," NASA Tech. Note D5089, 1969. 109. J. P. Hartnett, "Mass Transfer Cooling," in Handbook of Heat Transfer Applications, W. M. Rohsenow, J. P. Hartnett, and E. N. Gani6 eds., chap. 1, McGrawHill, New York, 1985. 110. H. S. Mickley, R. C. Ross, A. L. Squyers, and W. E. Stewart, "Heat, Mass, and Momentum Transfer for Flow Over a Flat Plate With Blowing or Suction," NACA TN 3208, 1954. 111. W. M. Kays, "Heat Transfer to the Transpired Turbulent Boundary Layer," Stanford Univ. Dept. Mech. Eng. Rep. HMT14, Stanford University, Stanford, CA, June 1971. 112. R. M. Kendall, M. W. Rubesin, T. J. Dahm, and M. R. Mendenhall, "Mass, Momentum, and Heat Transfer Within a Turbulent Boundary Layer With Foreign Gas Mass Transfer at the Surface, Pt. 1: Constant Fluid Properties," Itek Corp. Vidya Div. Rept. 111, 1964. 113. W. M. Kays and R. J. Moffat, "The Behavior of Transpired Turbulent Boundary Layers," Stanford Univ. Dept. Mech. Eng. Rep. HMT20, Stanford University, Stanford, CA, 1975. 114. M. W. Rubesin, "An Analytical Estimation of the Effect of Transpiration Cooling on the Heat Transfer and Skin Friction Characteristic of a Compressible Turbulent Boundary Layer," NACA Tech. Note 3341, 1954. 115. W. H. Dorrance and E J. Dore, "The Effect of Mass Transfer on the Compressible Turbulent Boundary Layer Skin Friction and Heat Transfer," J. Aeronaut. Sci. (21): 404410, 1954.
FORCED CONVECTION, EXTERNAL FLOWS
6.85
116. M. W. Rubesin and C. C. Pappas, "An Analysis of the Turbulent Boundary Layer Characteristics on a Flat Plate With Distributed Light Gas Injection," NACA Tech. Note 4149, 1958. 117. E. L. Knuth and H. Dershin, "Use of Reference States in Predicting Transport Rates in HighSpeed Turbulent Flows With Mass Transfer," Int. J. Heat Mass Transfer (6): 9991018, 1963. 118. R. L. P. Voisinet, "Influence of Roughness and Blowing on Compressible Turbulent Boundary Layer Flow," Naval Surface Weapons Center TR 79153, Silver Spring, MD, June 1979. 119. C. C. Pappas and A. E Okuno, "Measurements of Skin Friction of the Compressible Turbulent Boundary Layer on a Cone with Foreign Gas Injection," J. Aerosp. Sci. (27): 321333, 1960. 120. H. S. Mickley and R. S. Davis, "Momentum Transfer for Flow Over a Flat Plate With Blowing," NACA Tech. Note 4017, 1957. 121. T. Tendeland and A. E Okuno, "The Effect of Fluid Injection on the Compressible Turbulent Boundary LayerThe Effect on Skin Friction of Air Injected Into the Boundary Layer of a Cone at M = 2.7," NACA Res. Mem. A56D05, 1956. 122. R. J. Moffat and W. M. Kays, "The Turbulent Boundary Layer on a Porous Plate: Experimental Heat Transfer With Uniform Blowing and Suction," Stanford Univ. Dept. Mech. Eng. Rep. HMT1, Stanford University, Stanford, CA, 1967. 123. C. C. Pappas and A. E Okuno, "Measurements of Heat Transfer and Recovery Factor of a Compressible Turbulent Boundary Layer on a Sharp Cone With Foreign Gas Injection," NASA Tech. Note D2230, 1964. 124. E. R. Bartle and B. M. Leadon, "The Effectiveness as a Universal Measure of Mass Transfer Cooling for a Turbulent Boundary Layer," Proc. 1962 Heat Transfer and Fluid Mechanics Inst., Stanford University Press, Stanford, CA, pp. 2741, 1962. 125. J. E. Danberg, "Characteristics of the Turbulent Boundary Layer With Heat and Mass Transfer at Mach Number 6.7," Proc. 5th U.S. Navy Syrup. Aeroballistics, U.S. Naval Ordnance Lab., 1961. 126. C. J. Scott, G. E. Anderson, and D. R. Elgin, "Laminar, Transitional and Turbulent Mass Transfer Cooling Experiments at Mach Numbers From 3 to 5," Univ. Minnesota Inst. Tech. Res. Rep. 162, 1959. 127. J. Nikuradse, "Laws of Flow in Rough Pipes," NACA Tech. Mem. 1292, November 1950, transl, of "Str6mungsgesetze in rauhen Rohren," VDI Forschungsheft, No. 361, 1933. 128. L. Prandtl and H. Schlichting, Das Widerstandsgesetz rauher Platten, Werft, Reederei, Hafen 14, 1934. 129. J. M. Healzer, R. J. Moffat, and W. M. Kays, "The Turbulent Boundary Layer on a Rough Porous Plate: Experimental Heat Transfer With Uniform Blowing," Stanford Univ. Dept. Mech. Eng. Rep. HMT18, Stanford University, Stanford, CA, 1974. 130. T. Cebeci and P. Bradshaw, Momentum Transfer in Boundary Layers, Hemisphere, Washington, DC, 1977. 131. M. M. Pimenta, R. J. Moffat, and W. M. Kays, "The Turbulent Boundary Layer: An Experimental Study of the Transport of Momentum and Heat With the Effect of Roughness," Stanford Univ. Dept. Mech. Eng. Rep. HMT21, Stanford University, Stanford, CA, 1975. 132. M. D. Van Dyke, "HigherOrder BoundaryLayer Theory," in Annual Reviews of Fluid Mechanics, pp. 265292, Palo Alto, CA, 1969. 133. H. Thomann, "Effect of Streamwise Curvature on Heat Transfer in a Turbulent Boundary Layer," J. Fluid Mech. (33/2): 383392, 1968. 134. P. Bradshaw, "Effects of Streamline Curvature on Turbulent Flow," AGARDograph No. 169, 1973. 135. R. E. Mayle, M. E Blair, and E C. Kopper, "Turbulent Boundary Layer Heat Transfer on Curved Surfaces," J. Heat Transfer (101): 521525, August 1979. 136. S. A. Eide and J. P. Johnston, "Prediction of the Effects of Longitudinal Wall Curvature and System Rotation on Turbulent Boundary Layers," Stanford Univ. Dept. Mech. Eng. Rep. PD19, Stanford University, Stanford, CA, November 1974. 137. D. C. Wilcox and M. W. Rubesin, "Progress in Turbulence Modeling for Complex Flow Fields Including Effects of Compressibility," NASA Tech. Paper 1517, April 1980. 138. J. Kestin, "Effect of FreeStream Turbulence on Heat Transfer Rates," in Advances in Heat Transfer, T. E Irvine Jr., and J. P. Hartnett, eds., vol. 3, pp. 132, Academic, New York, 1966.
6.86
CHAPTER SIX
139. A. Edwards and B. N. Furber, "The Influence of FreeStream Turbulence on Heat Transfer by Convection From an Isolated Region of a Plane Surface in Parallel Air Flow," Proc. Inst. Mech. Eng. (170): 941, 1956. 140. W. C. Reynolds, W. M. Kays, and S. J. Kline, "Heat Transfer in the Turbulent Incompressible Boundary Layer, IVmEffect of Location of Transition and Prediction of Heat Transfer in a Known Transition Region," NASA Mem. 12458W, 1958. 141. J. Kestin, P. E Maeder, and H. E. Wang, "Influence of Turbulence on the Transfer of Heat From Plates With and Without a Pressure Gradient," Int. J. Heat Mass Transfer (3): 133154, 1961. 142. P. E. Hancock, "The Effect of FreeStream Turbulence on Turbulent Boundary Layers," PhD thesis, Imperial College, London, 1980. 143. S. K. E Karlsson, "An Unsteady Turbulent Boundary Layer," J. Fluid Mech. (5/2): 622636, 1959. 144. M. H. Patel, "On Turbulent Boundary Layers in Oscillatory Flow," Proc. Royal Soc. London A (353): 121144, 1977. 145. J. Coustiex, R. Houdeville, and M. Raynaud, "Oscillating Turbulent Boundary Layer With a Strong Mean Pressure Gradient," Proc. 2d Syrup. Turbulent Shear Flows, London, pp. 6.126.17, 1979. 146. P. G. Parikh, W. C. Reynolds, R. Jayaraman, and L. Carr, "Dynamics of an Unsteady Turbulent Boundary Layer," Proc. 3d Symp. Turbulent Shear Flows, Davis, CA, pp. 8.358.40, 1981. 147. E. R. van Driest, "Turbulent Boundary Layer on a Cone in a Supersonic Flow at Zero Angle of Attack," J. Aeronaut. Sci. (19): 5557, 72, 1952. 148. G. B. Schubauer and E S. Klebanoff, "Contribution on the Mechanics of Boundary Layer Transition," NACA Rep. 1289, 1956.
CHAPTER 7
RADIATION John R. Howell The University of Texas at Austin
M. Pinar MengO9 University of Kentucky
INTRODUCTION The field of radiative transfer is undergoing major advances in the capability to analyze complex problems. A welldeveloped theoretical base exists that can be applied to the solution of most (but not all) engineering problems. Major increases in computational speed and capacity have opened the way to solutions of problems that include complex geometries, spectral effects, and inhomogeneous properties. The need for such a capability is driven by applications that include hightemperature manufacturing processes and materials processing, improved efficiency, and more accurate design methods necessary for energy conversion devices, the use of new materials, hypersonic flow analysis, and others. Some areas of radiative transfer analysis that had seen sparse research for some time have undergone a renaissance. This is true for radiative transfer among surfaces with no participating medium. The need to provide fast and accurate computer visualization for use in data depiction, virtual reality, realtime animation, and other areas requiring accurate modeling of radiative transfer has revived research on surface radiative transfer algorithms. These same needs have brought forth new research on surface property measurement and modeling so that accurate spectral and directional effects can be included in computer visualization. Other areas of radiative transfer have been driven by increased capability of analysis due to the great strides in computer capability. Just a few years ago, twodimensional problems of radiative transfer in enclosures with a participating medium were at the edge of computational capability. Now, these are routine, and many threedimensional cases have been analyzed. Because of the need in applications such as utility steam generator design to analyze threedimensional geometries with up to tens of thousands of surface and volume computational elements, much research is now focused on further increases in computational speed. Massively parallel computers may well provide the required computational capability for such problems. In this chapter, we present the fundamentals of radiative transfer analysis. We begin with a review of the properties of the ideal radiating body, the blackbody, by describing radiative exchange among ideal (black) surfaces and then extending the analysis to surfaces with real radiative properties. We further extend the analysis to the case of a medium between the bounding surfaces that can absorb, emit, and scatter radiation. We provide information on the radiative properties of such participating media and conclude with a discussion of methods 7.1
7.2
CHAPTER SEVEN for treating combinedmode heat transfer problems in which radiative transfer is important. We present the bounds of applicability of the theory based on current stateoftheart and available data. Additional information can be found in the recent texts by Siegel and Howell [1], Brewster [2], and Modest [3].
Radiation Intensity and Flux Any substance at a finite temperature emits electromagnetic energy in discrete energy quanta called photons. The energy of each photon is equal to hv = hc/~,, where h is the Planck constant, c is the speed of light, and v and ~, are the frequency and wavelength of the emitted energy, respectively. As it will be outlined below, with decreasing wavelength, or increasing frequency, the energy associated with photons increases. Thermal radiation is associated with a temperature range of approximately 30 to 30,000 K and wavelength range of 0.1 to 100 ~tm. For most practical purposes, however, we are interested in a wavelength range of 0.4 (nearUV) to about 15 ~tm (nearIR). The photons whose energies correspond to this wavelength/temperature range are capable of changing the discrete vibrational, rotational, and electronic energy states of atoms and molecules of the material on which they are incident, and this, in turn, changes the internal energy and the corresponding temperature of the material. As a result of this, energy is transferred from a hot object to a colder one via thermal radiation as opposed to transfer by phonons when the objects are in contact. Radiative heat transfer from one small volume or surface element to another is determined by accounting for the energies of photons of all wavelengths, emitted in all directions over a certain time interval. Depending on the location of each element and its orientation with respect to others, the amount of radiant energy exchange between elements will vary. In order to determine the contribution of each element to the radiation balance, we introduce a fundamental and mathematically convenient quantity termed radiation intensity. By definition, the fractional radiant energy de~ propagating through (or originating from) an infinitesimally small area dAn in the direction ~(0, ~), confined within an infinitesimally small solid angle df~ around fi(0, ~), within a wavelength interval d~, around the wavelength of ~,, and within a time interval of dt is called the radiation intensity: I~(~) = I~(0, ~)=
z
^0
A
I
X FIGURE 7.1 Definition of radiative intensity.
lim
(aa,ata.ax.ao~o
dez(O, ~) dAn df~ d~ dt
(7.1)
The direction of propagation ~ is defined in spherical coordinates in terms of the zenith angle 0 and azimuthal angle ¢~ (see Fig. 7.1). dAn is the area normal to the direction of propagation t5 and is equal to dA cos 0, with 0 being the zenith angle (i.e., the angle between the surface normal h and ~). The solid angle is defined as the ratio of infinitesimal area normal to ~, to the square of the distance between two infinitesimal surface elements exchanging radiation (see Fig. 7.2). Note that radiation intensity may vary as a function of location ~= ?(x, y, z), direction ~ =~(0, ¢), time t, and wavelength ~,; therefore, it is a function of seven independent parameters. For most calculations of interest, the transient nature of radiation intensity is not critical. Recently, however, with the advances in femto and picosecond pulsed lasers, the transient radiative transfer applications have started becoming important (see, e.g., Ref. 4); nevertheless, we will not cover these advances in this chapter.
RADIATION z
rdOrsinSd~
/'"
,/~',',d "T I~'~N
\
F I G U R E 7.2
For heat transfer predictions, the radiative heat flux through a surface (with normal h) is required, which is determined by integrating the radiation intensity incident from all directions (i.e., 4n steradians) as
~(?): I,
= sinSdOd~
x Definition of solid angle.
7.3
i~(?,h)h. h d~
(7.2)
= 4~
where h is the unit vector normal to the surface (see Fig. 7.1). Note that by replacing t5 with h. h, which is equivalent to the direction cosine for a given surface, the flux across the surface is obtained. Radiative flux on an opaque (i.e., nontransmitting) surface element is obtained by evaluating the integral over a single hemisphere (i.e., f~ = 2rt steradians). The total radiative flux is determined by integrating ~ ( ? ) over the entire wavelength spectrum.
Blackbody Radiation The blackbody is the standard against which the behavior of all real radiating materials is gauged. It has welldefined characteristics that are firmly based in theory and experiment. Here, these characteristics will be outlined, as understanding them is paramount to conceptualizing the radiative transfer phenomenon.
General Definitions and Characteristics.
The most important attributes of a blackbody can
be listed as: • A blackbody is defined as a surface or volume that absorbs all incident radiation. This includes radiation at every wavelength and from every direction. • The blackbody is the best possible emitter of radiation at every wavelength and in every direction. • Radiation emitted by a blackbody increases monotonically at every wavelength with absolute temperature. • Radiation within an isothermal enclosure with blackbody boundaries is isotropic; that is, uniform in all directions. With such qualities, the blackbody is seen to be a convenient standard for comparing the properties of real materials. All real materials will reflect some incident radiant energy and are thus not perfect absorbers. Because they do not absorb as much as the ideal blackbody, they must emit less than an ideal blackbody to remain in thermal equilibrium with their surroundings. A real surface thus emits less than the blackbody (again, at every wavelength and in every direction). It is possible to construct a nearly ideal blackbody by the artifice of defining the surface over the small entrance to a deep cavity as the blackbody. Little of the radiation crossing this fictitious "black" surface reflects back through the cavity opening, especially if the internal cavity surfaces are made as nonreflecting as possible and are oriented so as not to face the cavity opening directly. If the interior cavity surfaces are maintained at a uniform constant temperature and the cavity opening is small, then the radiation within the cavity is isotropic and the energy leaving the cavity opening will be quite close to that of a blackbody at the cavity temperature. These observations are the basis for producing experimental blackbodies used in making comparisons of radiation from real material surfaces for property measurements. To analyze radiative heat transfer, quantitative measures of the blackbody characteristics described in this section must be provided.
7.4
CHAPTERSEVEN Blackbody Intensity, Emissive Power. It can be shown that the intensity leaving a black surface is independent of 0 and ¢; that is, the blackbody emitted intensity is isotropic [1]. This fact provides another convenient benchmark for comparing the behavior of real surfaces. For a blackbody, the spectral intensity is given by the Planck distribution of blackbody intensity [5]: 2C1 I~b = n2~5(eC~l,~r_ 1)
(7.3)
where T is the absolute temperature (in K) and C~ and C2 are constants with values
C~ = hc:o = 0.59552 • 108 W ~tm4/m2 C2 = hco/k = 14,388 lamK with
(7.4)
h = 6.626075 • 1034 Js (Planck constant)
(7.5)
k = 1.380658 • 1023 J/K (Boltzmann constant)
(7.6)
Co = 2.99792458.108 m/s (speed of light in vacuum)
(7.7)
and n is the index of refraction of the medium. To determine the energy leaving a black surface in all directions, l~b is integrated over the hemisphere of solid angles df~ = sin 0 dO de to give the spectral emissive power of a blackbody exb: 2/1;C1 1) =rtI~ e~b = fo~=o f;'~ =o l~b COS 0 sin 0 dO d~ = n2~5(eC2/,~r_
(7.8)
Figure 7.3 depicts the blackbody function for different temperatures, including the solar temperature of 5762 K. Equation 7.8 can be simplified by dividing by n3T 5 to give e~b _ 2rtC1 n3T 5 (n~,T)5(e c2/"~r 1)
(7.9)
This relation is plotted in Fig. 7.4. 1 08
E
l.
107
lo6
1 05 n°
1 04
~
1 03
~
1 02
,90
IO~
o
/
I OO 101
2
3
4
5 6
10 0
2
3
4
5 6
Wavelength (~m)
FIGURE 7.3 The Planck distribution of blackbody emissive power.
1 01
2
a
4
RADIATION ~i~'''
15
7.5
a.0
to
,¢ i
..............
E12
::t. &
e
0.8
o O r
E
~
9
.......
0.6
v
>'
"10 O
,
O
"" X
0.4
6
O t~
t.
.o
e
.a
3
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
0.2
o t'O U.
0.0 3
4
5
6 7
1 00
1.5
3
2
4
5
6 7
1 01
1.s
2
n~,T (ILU'nK) FIGURE 7.4 Normalized and fractional blackbody emission distributions.
Note that, with decreasing temperature, the area under the curve (i.e., the total energy emitted by the blackbody) decreases. The maximum value of the curve shown in Fig. 7.4 can be found mathematically by taking the partial derivative of Eq. 7.9 with respect to wavelength and setting the result equal to zero. The maximum of the curve occurs at a fixed value of C3 = (n~][")max
 2897.8
~tmK
(7.10)
This relation is known as Wien's displacement law, and it provides a convenient means of determining where the wavelength of peak radiated energy occurs for a given temperature. For example, the solar radiation spectrum has a peak at about ~ = 0.5 lam, as Tsun = 5762 K. On the other hand, an object at room temperature (Troom = 300 K) would emit maximum energy at about X = 10 l,tm. It is obvious that, as absolute temperature increases, the wavelength at which the maximum emissive power occurs becomes smaller. Values of e~/n3T 5 are tabulated in many sources (usually for n = 1; see Refs. 1, 2, 3, and 6) and easily computed from Eq. 7.9. If exp(C2/n~,T) >> 1, the Planck blackbody distribution can be simplified to 2/~C1
e~b ~n2~5 eC2/"~T
(7.11)
This expression, known as Wien's distribution of blackbody radiation, is accurate over a wide range of wavelengths and temperatures. Equation 7.9 can be integrated with respect to wavelength to obtain the total blackbody emissive power (i.e., the rate of energy emitted by a blackbody into all directions at all wavelengths):
eb = f:o exb d~, = n2(yT 4
(7.12)
Equation 7.12 is known as the StefanBoltzmann equation, and 0 is the StefanBoltzmann constant, which has the value
0 = 2C1~5/15C4= 2~5k4/(15h3c 2) = 5.6705" 10s W/(m2K 4)
(7.13)
7.6
CHAPTERSEVEN The value of ~ calculated from the accepted values of Planck's constant h, the Boltzmann constant k, and the speed of light in a vacuum Co agrees with experimentally determined values within experimental error of 1.4 parts in 10 4 [7]. Equation 7.12 shows that the rate of energy emitted by a blackbody increases in proportion to the absolute temperature to the fourth power, so that radiation will generally be the dominating heat transfer mode at high absolute temperatures. For many practical applications, the blackbody emission is needed within finite, relatively narrow wavelength intervals, rather than the entire spectrum. To calculate the blackbody energy between wavelengths ~1 and k2, we write
f~i2
C1 f~2 n2T4 ~3d ~
(7.14)
C1 [fo~2n2T4 e~3d~ fo~,n2T l 4 e~3d~ ] ~ 1 ~ 1
(7.15)
e~ d~ = C~2 ~1
e~  1
C4
= (Fo_,~zr Fo_,~,r)Eo(T)
(7.16)
where ~ = C2/n)~T. Here, F0,~r is the fraction of the energy in the blackbody distribution that lies below some given n;~T value and is called the fractional blackbody function of the first kind. This fraction is defined as ( e ~ / r ') d(~.r) F0,~r =
e~ d;~ =
fo~(e~/TS)d(~.T)
(7.17)
£~e~d;~
where the F fraction can be evaluated numerically and is listed in most textbooks. An analytical formulation for this integral was suggested [8], resulting in 15~[~(
3~2
Fo,,~r = __~ ._
p3
6~
+ ~1 + ~J2 +
63)]
(7.18)
This relation converges very quickly, usually within three terms of the series except at very large values of ~. Figure 7.4 also depicts the fractional blackbody function versus n~T product.
Nonblack Surfaces and Materials No real materials act as blackbodies, so measures of their deviation from blackbody behavior are used to define the spectral, directional, and temperature dependence of real surfaces relative to those of a blackbody. The notation used is to denote a spectrally dependent property with a ~, subscript as before, and to denote a directionally dependent property with a prime ('). These symbols are omitted when the property in question has been averaged over one or both of the dependencies
Emissivity.
The ability of a surface to emit radiation in comparison with the ideal emission by a blackbody is defined as the emissivity of the surface. The emissivity can be defined on a spectral, directional, or total basis.
Directional Spectral Emissivity* ' * The prime denotesthe directionalquantity. Note that the radiation intensityI is directionalby definition.
(7.19)
RADIATION
Figure 7.5 depicts the angular distribution of I~b and Ix schematically for an ideal blackbody and a typical real surface. Note that only the azimuthal angle (0) dependence is shown for the sake of clarity. Integrating the emitted energy over all wavelengths at a particular direction results in the directional total emissivity.
z
"~
7.7
0
Directional Total Emissivity I e' FIGURE 7.5 Schematic of angular intensity distributions leaving an ideal blackbody (Ixb)and a real surface (I~).
Ib

nf~ =o Ixd)~ n ~ =
(~T4
e'XI~bd)~
=o (~T4

nI (~T4
(7.20)
Integrating the energy emitted over all directions at a particular wavelength gives hemisphericalspectral emissivity.
HemisphericalSpectral Emissivity eX ex  l~2'~fov2 e~. COS 0 sin 0 dO d~ ekb ~ =0 =o
(7.21)
Integrating the emitted energy over both wavelength and direction and comparing with the similar integrated quantity for a blackbody yields hemispherical total emissivity.
Hemispherical Total Emissivity e
eb
f~=o~~=ofo~'2=oe~I~, cos 0 sin 0 dO dO d~, ~T 4 n f~=o exI~ d~,
f~ 8xe~bd)~ =0
(~T4
(jT 4
(7.22)
The various integrated emissivities allow calculation using data for the detailed emissivities.
Absorptivity.
The absorptivity is the property that defines the fraction of the incident energy that is absorbed by a surface. This property may also be dependent upon the direction and wavelength of the incident radiation. These properties will usually depend on the temperature of the absorbing surface (Ts); however, notation to indicate this fact is omitted.
Directional Spectral Absorptivity d3qx'a d3q~a ¢x'~ d 3qx,i Ix,i cos 0 dO
(7.23) dk
where the superscript 3 indicates that the flux q is a dependent function of three variables: wavelength ~,, direction (0, ~), and location r(x, y, z). Integrating the absorbed energy over all wavelengths at a particular direction of incidence results in directional total absorptivity.
Directional Total Absorptivity tz' 
d2qa n f(__o o(~Ix,id~, d2qi [** lx,i d)~ Jx
(7.24)
=0
By integrating Eq. 7.23 over all incident directions, the energy absorbed from all directions at a particular wavelength is obtained as hemisphericalspectral absorptivity.
7.8
CHAPTER
SEVEN
HemisphericalSpectral Absorptivity d2qLa
=o :o (Z~ILi COS 0 sin 0 dO d¢ 
(7.25)
d2qx,if,2'~fev2 =0
/~,~ cos e sin e de dO =0
Finally, integrating the absorbed energy over both wavelength and direction and comparing with the integrated incident energy gives hemispherical total absorptivity.
Hemispherical Total Absorptivity cx 
dq, dqi
f: ff2~£rc/2(X'~ILi

=o =o _o f; ~2~ £rd2 =0
=0
f:rc~rc/2
COS 0 sin 0 dO dO dX =
/L~ COS 0 sin 0 dO d~ d~
=0
o(L cos 0 sin 0 dO dO o =o ~2rt ~n/2 L cos 0 sin 0 dO d~ =0
(7.26)
0=0
Kirchhoff's Law.
Through an energy balance at thermodynamic equilibrium, it can be shown that the directionalspectral emissivity is always equal to the directionalspectral absorptivity of a surface, or ' = E~.' (XZ.
(7.27)
This relation is known as Kirchhoff's law. Equation 7.27 may be substituted into the various relationships for the integrated emissivity or absorptivity. However, it does not follow that such quantities as directional total, hemisphericalspectral, or hemispherical total emissivity and absorptivity are necessarily equal. In fact, the integrated properties are only equal if certain restrictions are met. These are given in Table 7.1. TABLE 7.1
Requirements for Application of Kirchhofrs Law
Relation
Restrictions
I: o~ = ~{ II: o~x= ¢x
None Incident radiation has equal intensity from all angles, or cx;.= G. are independent of angle Incident radiation has a spectral distribution proportional to that of a blackbody at the temperature of the surface, or o~ = e~.are independent of wavelength One restriction each from II and III above
III: (z' = ~'
IV: a = e
Reflectivity.
Reflectivity is the property of a surface that defines the fraction of incident energy that is reflected by the surface. This property depends not only on the wavelength and directional characteristics, but it must also describe the directional distribution of the reflected radiation. It therefore has more independent variables than the properties discussed so far. Integrated properties are defined by integration over incident angle, reflected angle, wavelength, and combinations of these. Many of the reflectivities are rarely used in practice; the most useful ones are defined here. Note that the same notation is used as for absorptivity and emissivity, except that a double prime indicates a reflectivity that depends on both direction of incidence and direction of reflection. The most fundamental reflectivity defines the intensity reflected into a particular direction resulting from energy incident from a given direction. At a particular wavelength, this is as follows.
Bidirectional Spectral Reflectivity P7 = IL i COS
Ix'r 0 i sin
(7.28)
0 i doe dOi
RADIATION
7.9
where double prime (") indicates the directional nature of both incident and reflected radiant energy. Integrating P'~ over all angles of reflection gives:
DirectionalHemispherical Spectral Reflectivity f2n ,~o~ ' 2r=O l ~ , r c o s O r s i n O r d O r d O r l~.,iCOS Oi sin 0 i d O i d~i
,,,= I,~d
=
f2,~ for2 p~cOSOrsinOrdOrdf~r *r=0 r= 0
(7.29)
Because of the reciprocity of the bidirectional reflectivity P'~, the hemisphericaldirectional spectral reflectivity for isotropic incident intensity P~r is equal to the directionalhemispherical reflectivity p'~. Here, a single prime (') is used to denote the directional nature of incident radiation. The fraction of the incident energy from all directions that is reflected into all directions at a particular wavelength is written as hemispherical spectral reflectivity.
Hemispherical Spectral Reflectivity f2~ f?2 P~.Jl~.i c o s t~i = 0 rr~
2n
f
0i
sin
Oi dOi dd~i
i = 0
(7.30)
~d2
f
i = o 0i = o
I~,ic°sOisinOidOidt~ i
By integrating over all wavelengths, the reflectivities defined above become total values and are given as follows.
Bidirectional Total Reflectivity
f;
Ik. r
dX
f; p 7 I~.~d)~
=0
O" =
=0
=
cos Oi sin OidOi dt~i f; ,) Izj
(7.31)
f;_ o l~,i
DirectionalHemispherical Total Reflectivity
f;=
,, P~.,i l~..i d~.
p~ =
(7.32)
f; = 0 lz.i d~, Hemispherical Total Reflectivity
~ f2. [,2 P~.il'~.i sin OidOi dt~idX f). ff2 p'il~ sin dOi dt~i p= ': ~; i2. ;ff2 g.j cos O~sin O~dOi d~ d)~  f)~ f),2 I~ cos O~sin Oi dOi d~ COS 0 i
= 0
~i = 0
= 0
~i = 0
0
COS 0 i
~
0i = 0
t~i = 0
Oi = 0
0
0
0i
(7.33)
0i = 0
Relations among Surface Properties.
Because incident energy onto an opaque surface is either reflected or absorbed by the surface, it follows that one unit of radiant energy incident on an opaque surface at a given wavelength and from a given direction will have a fraction absorbed and the remainder reflected, or 1 = tx~ + p'~
(7.34)
Similarly, for radiation at all wavelengths, it follows that 1 = o( + P'
(7.35)
7.10
CHAPTER SEVEN
Total radiation onto a surface at all wavelengths and from all directions follows a similar conservation relation, 1 = (x + p
(7.36)
Kirchhoff's law (Eq. 7.27) may be used to replace the absorptivity in these relations with emissivity if the restrictions of Table 7.1 are observed. Thus, data on one of the radiative properties can often be used to generate the others, although care must be used to avoid violating the restrictions of Table 7.1. If the surface is not opaque but semitransparent, then Eqs. 7.347.36 need to include transmissivity x and be replaced by the following equations: 1 = (x~' + p~ + x '
1 = oc' + p' + x' 1 = (x + p + x
(7.37) (7.38) (7.39)
where x~., x', and x are defined similar to the corresponding oc equations (see Eqs. 7.23, 7.24, and 7.25). Values f o r Surface Properties. In solving radiative transfer problems involving surfaces, property values must be available. Radiative surface properties given in the literature show wide variations, because the values are altered greatly by the presence of surface contaminants (oxidation, fingerprints, etc.), the presence of surface texture or roughness (machining marks, grain structure), and tailored surface characteristics such as coatings, films, or geometric structures specifically designed to affect spectral or directional characteristics. Thus, tabulated values serve at best as rough guidelines to the properties of engineering surfaces. Tabulated property values for selected substances are given in the first two tables in App. A (Tables A.7.1 and A.7.2). The available data for radiative properties of metallic and nonmetallic surfaces can be found in a number of data bases [9, 10]; they are too extensive to be included in this chapter. Also, the reader is referred to a number of recent handbooks and reviews for more detailed data [ 1 1  1 7 ] . In the absence of tabulated or measured properties for a given surface, various options are available. The behavior of a surface can be computed based on fundamental theories, such as Maxwell's electromagnetic wave theory; the surface characteristics can be assumed based on extrapolation from the behavior of similar surfaces; a model of the surface behavior can be constructed based on simplified assumed surface characteristics; or greatly simplified characteristics can be assumed to be accurate enough for use. In the third table in App. A (Table A.7.3), the spectral complex index of refraction data for a number of metals are listed (from Ref. 16), which can be used to determine the surface absorption, reflection, and transmission characteristics as discussed below. Computed Properties. For perfect dielectric materials and for highly conducting materials, electromagnetic wave theory can be used to predict radiative properties [1, 18]. These predictions are based on the assumption that the surface is optically smooth; that is, the surface roughness is small compared with the wavelength of the radiation (otherwise, the problem may be numerically I Ii" Ir (0) intractable). In addition, the composition of the material must be well known so that necessary properties such as the simple refractive index n and the attenuation coefficient k that comprise the complex refractive index rn = n  ik can be assigned (Fig. 7.6); this implies that the surface is not nl V contaminated by oxide films or other impurities. If these N 2 :: • ,...~. ' + .  ::,  , ; ' . ..':....".; "..:'f.: : ::~  c ..,:., . : ~ ,"'..'... restrictions are met, then properties can be computed with reasonable accuracy from the following relations, which : .:: : :.:,: .:." .~. . :.'.,":i : ;:.".:. '..[...2 ,..: apply for radiation from medium 1, assumed to be non or weakly absorbing (i.e., nl = 1 and kl  0), onto the surface of FIGURE 7.6 Definition of radiative properties at absorbing/reflecting surface of medium 2: the interfaces.
RADIATION
7.11
• Dielectrics (insulating materials, with n = n 2 / n l and kl = k2 = 0) Normalhemispherical total or spectral reflectivity p, = ( n  1) 2 (n + 1)2
(7.40)
Directionalhemispherical total or spectral reflectivity
P" 1 I(nec°sO[n2sin20]l/2)2 =2
( [n2sin2011/2  COS 0 ~2]
n 2 c o s 0 + [ n 2  s i n 20] 1/2 +
[n 2  s i n 2 0  ~ + cos 0
)J
(7.41)
  N o r m a l total emissivity
41"/ ~ ' = (n + 1) 2
(7.42)
Hemispherical total emissivity
1 e=~
3n+1) n1) 6(n+1)2
n2 .2_1)2 (.1)2n3 
(n 2+1)3
I n n. +. 1+
n2+2n1) (n 2 + 1)(n 4  1)
8n4n4+a) (n 2 + 1)(n 4  1) 2 In n (7.43)
• Metals (electromagnetically attenuating materials; for incidence on a surface through most gases, nl = 1 and k~  0):   N o r m a l hemispherical reflectivity P; = ( n 2  nl) 2 + ( k 2  k,) 2 = ( n 2  1) 2 + (k2) z (n2 + nl) 2 + (k2 + kl) 2 (n2 + 1) 2 + (k2) 2
(7.44)
  N o r m a l spectral emissivity (nl = 1 and k~ = 0),
[ re\1/2
eL. = 3 6 . 5 /   /
 464
re 2
(7.45)
where re is the electrical resistivity of the metal in ohm  cm and ~, is in ~m   N o r m a l total emissivity (n~ = 1 and kl = 0) e ' = 0.578(re T) in  0.178re T + 0.0584(re T) 3/2
(7.46)
Hemispherical total emissivity (nl  1 and k~ = 0) e = 0.766(re T) in  [0.309  0.0889 In (re T)]reT O.Ol75(reT) 3/2
(7.47)
The relations of Eqs. 7.407.47 as well as Kirchhoff's law (with the restrictions of Table 7.1) may be used to find other properties from the electromagnetic theory relations. The relations given for both dielectrics and metals are for unpolarized incident radiation; if polarization is important, then more detailed analysis must be used (see Refs. 1, 3, 18, 19). Also, the refractive index and absorption index may show spectral dependence, in which case the computed radiative properties will also be spectral in nature. Property Approximations. Because of the lack of accurate radiative properties in many situations, it is common practice to invoke certain approximations for the property behavior. The most common assumptions are that the surface properties are independent of wavelength (a gray surface), independent of direction (a diffuse surface), that the surface behaves as an ideal mirror (a specular surface), or that the surface is black. The assumption of a graydiffuse surface is the most commonly invoked. For a surface that is truly both gray and diffuse, Kirchhoff's law applies for all of the property sets; that is, tx = e, and the computation of radia
7.12
CHAPTERSEVEN tive exchange among surfaces and in enclosures is considerably simplified. However, for surfaces with directional or spectral property variations, these assumptions can introduce considerable error into the energy exchange calculations, particularly when radiation is the dominant heat transfer mode.
RADIATIVE EXCHANGE: ENCLOSURES CONTAINING A NONPARTICIPATING MEDIUM In this section, we deal with energy exchange among surfaces and in enclosures when no medium is present between the surfaces that affects the transfer; that is, no scattering, absorption, or emission occurs within the medium. Such effects are covered in the next major section of this chapter.
Black Surfaces The radiative exchange between two black surfaces depends only on the absolute temperatures of the surfaces and their shapes and relative positions. If the fraction of blackbody energy leaving area element d A j and incident on area element d A k is defined as dFdjdk, then the energy emitted in wavelength interval d~, around ~, by black element j and incident on element k is d3q~,dj_dk and given by (see Fig. 7.7)
dAk rodionteneroy receivedby dAk n..~/'~' JI~///A dFdj.dtt = rodlon] e'nerg~"yleov'ingdA, i'~'nolIdir"ections " /" Uk,,~~" J //// //"
i// / ~ / /
///~///~t Sl.k
nj
~
d
A
j
FIGURE 7.7 Nomenclatureand schematicfor radiative exchange between two surfaces.
d3q~.,dj, ak = e ~ , j d A j d F a j _ a k d ~
(7.48)
where, again, the superscript 3 means that the flux is a function of three independent parameters, wavelength, direction, and spatial coordinate. The dFdjdk is called the configuration factor, shape factor, view factor, or angle factor. It is independent of wavelength because for black surfaces, the directional distribution of emission from a surface does not depend on wavelength but is diffuse at every wavelength. Equation 7.48 gives the energy leaving surface j that is incident upon (and therefore absorbed by) black surface k. The energy leaving k and absorbed by j is given by a similar equation: d 3q~,,dk , dj = e~b,k d A kdFdk  dj d~,
(7.49)
RADIATION
7.13
The net exchange between the two surface elements d3qx, dj_dk (instead of d3qk,d]~dk) is then
d3qk,djdk = d3qk,dj~dk  d3qk,dk ~dj = exb,jdAjdFdj_dkd~, exb,kdAkdFdkdjd~,
(7.50)
If surfaces j and k are at the same temperature, then the net energy exchange must be zero, and it follows from Eq. 7.50 that
dAkdFdkdj = dAjdFdj_dk
(7.51)
Substituting this reciprocity relation into Eq. 7.50 results in
d3q~,djdk = (exb,j  exb,k)dAjdFdjdkd~,
(7.52)
Integrating over all wavelengths gives
d2qd~dk = (eb,~ e~k )dAjdFdjdk = 6( T~ T~)dAjdFdj_dk
(7.53)
If surface k is finite in extent and dAk is an element of that surface, then Eq. 7.53 can be integrated over the entire area of surface k to find the net energy exchange of surface element j with surface k:
dqdjk = t~ ~ [T 4 T~] dFdjdk dAj
(7.54)
"A k
Isothermal Surfaces.
where
If surface k is isothermal, Eq. 7.54 can be easily integrated to obtain
dqdjk = ~[T 4 T~,]Fd~_kdA~
(7.55)
Fdj_ k = ~ dFdj_dk
(7.56)
"A k
If surface j is also isothermal, then Eq. 7.55 can be integrated to give
dqjk = c~[T~ T~] ~ dFdj_k dA~ = c~[T4"A J where
T4IF~_~Ai
(7.57)
1
Fj_k=~j fAj Fdyk dAj
(7.58)
By writing the net radiative energy flux on surface k (rather than j), it is easily found that
A~dF~_dk = dAk Fdkj AkdFk dj = dAj Fej_ k AkFk_j= AjFj_k
(7.59)
These reciprocity relations will be used in later sections. If there are N surfaces forming an enclosure, then the net radiative energy flux on surface j is given by summing the contributions from all surfaces forming the interior of the enclosure: N
dqj= ~ ~. (T~ T4k)dFdj_kdaj
(7.60)
k=l
A total of N temperatures and radiative energy fluxes must be known to solve for the others. If the temperatures of all of the N surfaces are available, then the radiative fluxes at all surfaces are easily computed from Eq. 7.60. If radiative energy fluxes are given at M of the N surfaces, and temperatures are given at the others, then the set of M linear equations must be
7.14
CHAPTERSEVEN solved for the unknown temperatures. The methods given in the section "Exchange among Gray Diffuse Surfaces" can be used. Nonisothermal Surfaces. If the temperature of a surface varies, determination of the radiative energy flux or temperature profile on all surfaces of an enclosure requires the solution of integral equations (see Eq. 7.54). For the general case of all surfaces having varying temperatures, the relation for an element on surface j becomes
dq(rj)dj_~ = ~ fA [T4(rj)  T~,(r~)] dFdjdk dAj
(7.61)
k
and, summing over all surfaces in the enclosure, N
dq(rj)dJ: O ~l= fAk [T4(rj)  T4(rk)] dFdj_ak dAj
(7.62)
Configuration Factors. To solve for the radiative transfer among surfaces using the previous blacksurface equations, expressions for the configuration factors must be available. Many factors for common geometries have been derived and presented in the literature. Compilations are given in Siegel and Howell [1] (42 factors and references to 234 factors available in the literature); Brewster [2] (13 factors); Modest [3] (51 factors); and Howell [20] (over 278 factors with an annotated bibliography). Some useful factors are illustrated in App. B. Configuration Factor Algebra. When the configuration factor FA, between two surfaces is known, the reciprocity relation (Eq. 7.59) can be used to find FBA. Other relations can also be developed that allow simple calculations of new factors from known factors. If surface B can be subdivided into N nonoverlapping surfaces that completely cover surface B, then N
Fa_B : FA1 + FA2+ FA3+"" + FAN= ~, FAn
(7.63)
n=l
because all energy fractions from surface A to parts of surface B must equal the fraction of the total energy leaving A that is incident on all of B. Suppose that surface 1 is completely enclosed by a set of M surfaces. In that case, all energy leaving surface A must strike some other surface forming the enclosure. In terms of configuration factors: M El_ 1 ~ El_ 2 ~ El_ 3  k  . ' ' d FI_ M = ~ F1mm=l
FIGURE 7.8 Perpendicular right isosceles triangles joined along their short sides.
1
(7.64)
Note the term FI_I must be included in the summation if surface A is concave to account for the fraction of energy leaving surface A which is incident on itself. The reciprocity relations plus Eqs. 7.63 and 7.64 form the basis of what is called configuration factor algebra. Using these relations, new factors can be computed from a set of known factors; sometimes, factors can be generated from the algebra alone. The procedure is best illustrated by example. Consider two right isosceles triangles that are joined along a short side, as shown in Fig. 7.8. The triangles are perpendicular to one another. To find F1_2, note that an enclosure can be formed by first joining the free corners of the triangles by a line of length I as shown in Fig. 7.9. This forms a corner cavity with the third congruent triangle. The enclosure is completed by placing an equilateral triangle of side l
RADIATION
7.15
h h
l
A1
h
FIGURE 7.9 Construction of corner cavity by addition of line connecting free corners of triangle.
FIGURE 7.10 Completion of enclosure by addition of equilateral triangle, surface 4.
(and area A4) over the cavity formed by the three isosceles triangles, which have equal areas A1, A2, and A3. This is shown in Fig. 7.10. Now, apply configuration factor algebra. Eq. 7.64 gives El1 + El2 + El3 + F14 = 1
(7.65)
Because surface 1 is planar, F~_~= 0. By symmetry, F~_2= F~_3. Thus, Eq. 7.65. reduces to 2F1_2 + Fl_4 = 1
(7.66)
For surface 4 of the enclosure, Eq. 7.64 plus the use of symmetry gives F4_~ + F42 + F43 + Fan = 3F4_1 = 1
(7.67)
Using reciprocity, Eq. 7.59 results in
A4
A4
F~_4 = ~  F41 3A~
(7.68)
1  F,_4 1  (A4/3A,) F~_2 = ~ = 2
(7.69)
Substituting into Eq. 7.66 results in
Using geometry, A1 =
h2/2 and A4 = X/312/4 = k/3h2/2, giving F12 =
1  (lIVe)
2
 0.21132
(7.70)
This is the desired answer. The other configuration factors can be d e t e r m i n e d in a similar manner. Siegel and Howell [1] note on p. 220 that, for an Nsurfaced enclosure of all planar or convex surfaces (i.e., Fi_i = 0 for all i), N ( N  3)/2 factors must be found from a catalog of factors or by calculation. The remaining factors can then be d e t e r m i n e d by configuration factor algebra. If M of the surfaces (M < N ) are concave (i.e., have Fi_i > 0), then [ N ( N  3)/2] + M factors must be known before configuration factor algebra can determine the remaining factors. The presence of symmetry may further reduce the n u m b e r of factors that must be known before the rest can be determined.
7.16
CHAPTERSEVEN When the values of certain factors are known only approximately, then the constraints imposed on the factors by reciprocity and conservation in an enclosure can be used to refine the known values. Methods for this purpose have been proposed in Refs. 2124.
Exchange Among Gray Diffuse Surfaces When an enclosure can be assumed to be made up of a set of black or gray diffuse surfaces, welldeveloped techniques are available for determining the exchange among the surfaces. In general, one boundary condition (either surface temperature or heat flux) must be specified for every surface of the enclosure. It is possible by more advanced techniques to obtain solutions for the case when some surfaces have both conditions prescribed and others have neither condition specified. This situation causes the solution methods described here to fail, as the equation set is then illconditioned. Socalled inverse solution methods must be invoked. Methods of handling such a problem are given for blacksurfaced enclosures in Ref. 25 and for enclosures with black or gray diffuse surfaces in Refs. 26, 27, and 28. Use o f Configuration Factors. For diffuse nonblack surfaces, the radiation leaving the surface is made up of both diffusely emitted and diffusely reflected energy. Thus, the directional distribution of all radiation is diffuse. It follows that we can continue to use the concept of configuration factors to determine the fraction of the leaving radiation that strikes another surface. Thus, all of the information generated about configuration factors for black surfaces also applies to the more general case of diffusely reflecting and emitting surfaces. If the surfaces are diffuse and have wavelengthdependent properties, the methods to be outlined here will apply for energy exchange at each wavelength; a final summation or integration over all wavelengths will then be necessary to compute total radiant exchange. The outgoing radiation energy flux from a given location on surface k, qo,k is made up of the emitted and reflected flux from that surface, or qo,k = Ek(ST2 + Pkqi,k
(7.71)
where all quantities are evaluated at a particular location on surface k. The quantity qi,k is the radiation flux incident at the given location from all other surfaces in the enclosure, including surface k itself, if it is concave. The quantity qo,k is often called the radiosity of the surface, and qi,k, the irradiance. Note that, contrary to the practice in most of heat transfer, these energy fluxes carry a directionalitythe radiosity is the portion of the radiant energy flux with the component away from the surface, while the irradiance is the portion directed toward the surface. The net radiative heat flux leaving surface k(qk) is the difference between the radiosity and the irradiance, or qk =qo, kqi, k
(7.72)
This flUX corresponds to the usual concept used in heat transfer, as the net energy flux is taken as positive if in the direction parallel to the surface normal of the position on k. The final equation for energy transfer quantifies the irradiance as the sum of the radiant energies reaching a location on surface k from all other areas on the enclosure surface. This relation can have various forms depending on the degree of approximation used in the analysis. Circuit Analogy. Simple problems in radiant transfer can be diagrammed and formulated in analogy with a simple electrical circuit [29]. This is done by observing that Eq. 7.57 for black surfaces is in the form of Ohm's Law if dqj_k is analogous to the current and (T 4 T 4) is analogous to the driving potential difference. In this case, the corresponding resistance is 1/(~Fj_kAj). The analogy can also be extended to gray diffuse surfaces. In making this analogy, all of the usual assumptions are still present: all surfaces are gray and diffuse, and the
RADIATION
7.17
radiosity leaving each surface is uniform over the surface (implying that the temperature, heat flux, and incident radiation are uniform over each surface). The circuit analogy is not a useful approach when more than about four surfaces are treated, and it will not be elaborated here. For more details, see Refs. 1 and 3. Net Radiation Method. A more powerful method for describing radiative transfer is the net radiation method. In this method, radiative energy balances are constructed for each surface, and the resulting set of equations is then solved. (Some equations as written may fail in the limit of black surfaces and must be slightly modified starting from the original relations.) U n i f o r m Surface Radiosity. Here, first, we limit consideration to gray diffuse surfaces with uniform radiosity. In that case, Eqs. 7.71 and 7.72 apply. Because of the assumption of gray diffuse surfaces, Eq. 7.71 can be rewritten as qo,k = Ek(YT~ + (1

(7.73)
Ek)qi, k
An additional equation for the radiative energy incident on surface k is N
qi, k a k = ~ qo, jajFj_k = Ak ~" qo, jFkj
(7.74)
j=i
The three equations, 7.727.74, can be written for each of the N surfaces in the enclosure. If either Tk or qk is prescribed for each surface, this results in 3N equations in 3N unknowns: the unknowns being q~,k, qo,k, and either qk or Tk values for each surface. For m of the N surfaces in the enclosure having specified temperatures and the remaining N  m surfaces having specified heat flux, q~.k can be eliminated to give the useful forms N
[Skj  (1  ek )Fkj]qo, j = Ek(YT4
1O m
qr  X Wm '~m' Im; m
~m < 0
(7.134)
7.32
CHAPTER
SEVEN
m
q: = Z w.,l~.,11~; ~,. < 0
(7.135)
m
In this formulation, the angular derivative term (i.e., the second term in Eq. 7.133) is the most difficult to evaluate. Since the original work of Carlson and Lathrop [33], a direct differencing procedure has been used to simplify this term, such as:  ~ r
~)
(7.136) r
Wm
As mentioned by Jamaluddin and Smith [59], the m _+ 1/2 directions correspond to the angular range of Wm. An explicit relation can be obtained assuming a uniform radiation field in the corresponding angular domain: (7.137)
or,, +,/2  a,,_ ,/2 = w,,~,,
If Eq. 7.133 is multiplied by 2rtrdrdz and integrated over the ringshaped volume element, a difference formulation of the RTE is obtained:
Cm(Ai+ ,Ii+1  A,Ii)  (A,+ ,  A,)[(0tm + 1/21,,, + 1/2  am,/21m1/2)/Wm] + gm(Bj+, lj+ i  Bjlj)
~v~
=  ( K + ~) Vplm + ~:Vplh + ~
~ OP(UZ, PZ')Wmlm
(7.138)
tt!
Here, let us assume that the phase function is a linearlyanisotropic one, expressed as: • (~, f£) = 1 + a0 cos ¢ = 1 + ao(¢m¢', + grog:,)
(7.139)
Using a centraldifferencing scheme, the intensities can be related to Im li + li +, = L,, +,/2 + Ira,/2 = lj + lj +, = 2Im
(7.140)
where Im is the intensity at the center of the volume element: Im = { , , A L + fJml,,_ ,/2 + gmBlj + Vp('~l b Jr Ols) ¢,,,A + f5,,, + lamB + Vp(~ + cy)
with
(7.141)
A = Ai + A~ + 1 B = Bj+ Bj+I [~m = ((/'m + 1/2 + 0 { : m  , / 2 ) ( A i +
l  Ai)/Wm
[ l + ao(~m~'m + bt.,~'m) ] g = ~" w,,,L,, m
4rt
Equation 7.141 is written for positive ~,, and It,, values. For other combinations, similar equations need to be developed. Note that Vp is the volume of the computational cell. For the solution of the governing equations, an iterative scheme is followed [69]. After determining the intensity at a cell center (see Eq. 7.141), the intensity downstream of the surface element can be determined via extrapolation using Eq. 7.140. The central differencing used in Eq. 7.140 may result in negative intensities, particularly if the change in the radiative field is steep. A numerical solution to this problem was recommended by Truelove [67], where a mixture of central and upward differencing is used:
li+ l =(1 + f)Im  f L
(7.142)
RADIATION
7.33
where 0 < f < 1 and ~m,~l,m> 0. I f f i s 1.0, it is central differencing, and if f = 0.0, it is upwind differencing, and the intensity is always positive. For a scattering medium, the first solution excludes the in and outscattering terms; they are included in further iterations. Tables 7.2 and 7.3 list the quadratures used for the $2 and $4 approximations.
The Ordinates and Weights for the $2 Approximation in Axisymmetric Cylindrical Enclosures
TABLE 7.2
SN Point (m)
~t
Ordinates rl
~
Weights w
1 2 3 4
0.5 0.5 0.5 0.5
0.7071 0.7071 0.7071 0.7071
0.5 0.5 0.5 0.5
n n n n
TABLE 7.3
The Ordinates and Weights for the S 4 Approximation in Axisymmetric Cylindrical Enclosures SN Point (m)
la
1 2 3 4 5 6 7 8 9 10 11 12
0.2959 0.2959 0.9082 0.2959 0.2959 0.9082 0.9082 0.2959 0.2959 0.9082 0.2959 0.2959
Ordinates 11 0.2959 0.2959 0.2959 0.9082 0.9082 0.2959 0.2959 0.9082 0.9082 0.2959 0.2959 0.2959
~
Weights w
0.9082 0.9082 0.2959 0.2959 0.2959 0.2959 0.2959 0.2959 0.2959 0.2959 0.9082 0.9082
n/3 n/3 n/3 n/3 n/3 n/3 n/3 n/3 n/3 n/3 n/3 n/3
Figure 7.17 depicts the comparison of S 2 and S 4 DO approximation predictions [69] with Monte Carlo calculations [91] in an absorbing, emitting, and scattering medium. Overall, the accuracy of the agreement is acceptable, for a wide range of optical thicknesses and single scattering albedo values considered. Note that the extension of the DO approximation to threedimensional rectangular enclosures was attempted by a number of researchers (see, e.g., Refs. 65 and 85). Even though the formulation of the threedimensional model will not be given here, its governing equations can be derived with little difficulty. As shown in Refs. 65 and 85, the original SN quadratures yield accurate results in threedimensional solutions; they are listed in Table 7.4 for $2, $4, $6, and $8 approximations. By the LambertBeer law, the radiative transfer equation is derived from a macroscopic point of view using the principle of the conservation of radiative energy. In order to solve the RTE using any of the available methods, certain mathematical assumptions are introduced, and the physics of the problem are retained. If the assumptions are correct, then the RTE model yields physically acceptable results.
Statistical Models.
7.34
CHAFFER SEVEN 50 "'I
I
I
Discrete Ordinates
_O..................~_ •
I
S4
I
Monte Carlo I / ~ = 0 . 2 5 m  I
   S2
• / ~ =0.50 mI
 ¢  ~      ~_"SL'~
,_= o=
~
* B : 1.00 m'
3o
I
20

_
•
"6
0
i
I
I
I
i
0
0.2
0.4
0.6
0.8
l
1.0
Single Scattering Albedo, ~o= cr/~
FIGURE 7.17 Comparison of $2 and $4 predictions with the Monte Carlo results in cylindrical geometry (adapted from Ref. 69).
O n the other hand, it is possible to d e t e r m i n e the radiative energy balance everywhere in the m e d i u m by considering a large n u m b e r of photons, which originate from each volume/ surface element, p r o p a g a t e in all directions, and are absorbed and scattered. The radiative energy gain or loss in each e l e m e n t is calculated by considering the effect of an infinitely large n u m b e r of photons, as each p h o t o n affects the total energy balance. In doing so, the direction and the path length of each p h o t o n bundle are d e t e r m i n e d statistically. This procedure, called
TABLE 7.4 Ordinates and Weights for the SN Approximations in Rectangular Geometries (First Quadrant Values) SN Approximation
Point (m)
Ordinates
Weights
la
~
rl
w
$2
1
0.5773503
0.5773503
0.5773503
1.5707963
$4
1 2 3
0.2958759 0.9082483 0.2958759
0.2958759 0.2958759 0.9082483
0.9082483 0.2958759 0.2958759
0.5235987 0.5235987 0.5235987
$6
1 2 3 4 5 6
0.1838670 0.6950514 0.9656013 0.1838670 0.6950514 0.1838670
0.1838670 0.1838670 0.1838670 0.6950514 0.6950514 0.9656013
0.9656013 0.6950514 0.1838670 0.6950514 0.1838670 0.1838670
0.1609517 0.3626469 0.1609517 0.3626469 0.3626469 0.1609517
$8
1 2 3 4 5 6 7 8 9 10
0.1422555 0.5773503 0.8040087 0.9795543 0.1422555 0.5773503 0.8040087 0.1422555 0.5773503 0.1422555
0.1422555 0.1422555 0.1422555 0.1422555 0.5773503 0.5773503 0.5773503 0.8040087 0.8040087 0.9795543
0.9795543 0.8040087 0.5773503 0.1422555 0.8040087 0.5773503 0.1422555 0.5773503 0.1422555 0.1422555
0.1412359 0.0992284 0.0992284 0.1712359 0.0992284 0.4617179 0.0992284 0.0992284 0.0992284 0.1712359
RADIATION
7.35
a statistical or MonteCarlo approach, is straightforward to implement, although, computationally, it may be expensive. The Monte Carlo method seeks to replace mathematical descriptions of radiative transfer such as those described in the subsections on the zonal method and the differential and moment method with a simulation of the physical processes that are occurring. The method simulates radiative transfer by following the histories of small amounts of radiative energy, often called energy bundles. The bundles are followed throughout the series of events such as emission, reflection, scattering, and absorption that occur over the bundle lifetime. By following a sufficiently large number of bundles, the radiative flux or flux divergence distributions can be found over surfaces or within volumes of participating media. The Monte Carlo method results in statistically averaged results. The accuracy of the results depends upon the number of sample bundles chosen for study in the same way that experimental accuracy depends upon the number of replications of the experiment. This dependence on the number of samples is the greatest drawback of the method, as it requires a tradeoff between computer time and statistical accuracy of the results. Brief Outline ofthe Monte Carlo Method. Monte Carlo is based on determining the probability of a particular event in the history of an energy bundle. Whenever such an event occurs based on the physics of the problem, a choice is made such that, after many such events, the choices have the correct distribution. This is done by relating the variable being chosen to a random number R, where 0 < R _N i
'
xi=(ni/Ni) W
L
i 0=sinl(R01/2)
. is i = 2? .,
]~
for k = 1 to M, qk/ffT 14 = S k/N 1 for m = 1 to M, qm/ffT14 = Sm/N 1
.... Ifi=l,j=2 I
if i=2, j1 xj=xi+Ltar~cos~
!
s!o P
I
Ifj=l, k = INT(Mxj/W)+I, SkSk+l If j=2, m = INT(Mxj/W)+I, Sm=Sm+l
F I G U R E 7.19
Flowchart for the Monte Carlo solution.
I
7.38
CHAPTERSEVEN absorptions at positions on surfaces 1 and 2, respectively, and M is the number of Ax increments on the surfaces. For a second example, consider a gray absorbing medium with uniform properties contained between black infinite parallel plates at different temperatures. Find the heat transfer between the plates and the temperature distribution in the medium, assuming radiative equilibrium (i.e., no sources or sinks in the medium). Energy bundles are followed through their histories after emission from each surface until absorption at a boundary. Because of the assumption of radiative equilibrium, any bundles absorbed within a medium volume element must be balanced by an emission from that element; this is simulated by simply reemitting an absorbed bundle in a new direction and continuing the history until final absorption at a boundary. The medium temperature distribution is computed by equating the emission from the element to the absorption. The flow chart for this case is shown in Fig. 7.20. Table 7.6 gives the results for the net radiative flux between the plates at an optical thickness of 1.0 for various numbers of energy bundles, compared with the exact solution of this problem, which is 0.5532. The solution was programmed according to the flow chart in Fig. 7.20. Discussion of Monte Carlo methods for radiative transfer calculations have been given in Refs. 61, 93, and 94. It was used to model radiative heat transfer in large scale furnaces by, among others, Taniguchi and Funazu [95], Xu [96], Gupta et al. [91], and Richter [97]. In general, Monte Carlo techniques do yield reliable results if certain precautions are taken and sufficient numbers of photon bundles are utilized and biasing techniques are applied [61]. As mentioned by Howell [94], one of the main drawbacks of a Monte Carlo technique is the grid
InputT I,T 2, NI,L N2=N 1(T2/TI)4, a I i=l Y1=0 I ni=0 L. ni=ni+l I
] I
i=2 y2=L
1_ ["
Yo=Yi I is ni > Ni
~
0=sinl(R01/2) ~ I I ( )~ j=TRUNC(y/Ay)+ 1 S:=S:+I coJs0~I2R "~ Y=Y°(l252i)(c°s0/a)lnR [
. isi=2?
,
If y>L, then Sw2=Sw2+l h If yy>L? Stop FIGURE 7.20 MonteCarlo flowchart for radiative exchange between opposed black plates of finite width.
RADIATION
7.39
TABLE 7.6 Resultsof Monte Carlo Example Problem; Normalized Radiative Flux at Surface 1 for Various Sample Sizes N1, bundles emitted by surface 1
Radiative flux at surface 1, ql/cT 4
100 500 1000 5000 10,000 50,000
0.592 0.542 0.553 0.561 0.567 0.5564
incompatibility with the conservation of mass and energy equations. This problem, however, is likely to be eliminated with increasing computational power of computers. The main advantage of the Monte Carlo techniques is its possible application to any complex geometry. Recent work on the Monte Carlo technique includes that by Farmer and Howell [98101], who investigated the optimal strategies for implementing the Monte Carlo technique on both serial and parallel computers. For up to 32 parallel processors, the CPU time on the parallel machines was very nearly inversely proportional to the number of processors, indicating that the method is very well suited to parallel machines. With the advancement of computers, it is theoretically possible to solve any differential or integral equation using an appropriate numerical discretization scheme. The challenge here is to devise a scheme such that the physics of the problem are retained. All the radiative transfer models developed in the past attempt to do this; however, they also take advantage of different analytical techniques to reduce the complexity of equations and the required computational power as much as possible. In numerical solution techniques, the computational requirements may be decreased by using certain numerical simplifications and/or by choosing coarser grids. In general, it is safe to say that, with the availability of supercomputers, finer grids compatible with the grid scheme used for the solution of the other conservation equations can be used with little difficulty and computational penalty. Therefore, numerical techniques are likely to have wide acceptance in the future to model the radiative heat transfer in complicated geometries, especially for nonscattering media (see Refs. 102 and 103). Raithby and Chui [103] suggested a finite volume formulation of the RTE in onedimensional planar and twodimensional rectangular enclosures. The model is capable of handling scattering by the medium and yields very good agreement with the exact solutions. In principle, it is very similar to the discrete ordinates approximation, the main difference being the selection of the angular discretization to be used. In DO approximations, this selection is analytically determined, whereas Raithby and Chui use an arbitrary criterion for the selection. Therefore, a trialanderror approach is required to determine the optimum angular discretization. The finite element techniques have been used to model radiative heat transfer in multidimensional enclosures. Razzaque et al. [104, 105] considered the combined conductionradiation problem in an absorbing/emitting medium. Chung and Kim [106] and Stikmen and Razzaque [107] allowed for isotropic scattering in the medium. Tan [108] introduced a product integration method, which is similar to the finite element technique. This approach requires much less computational time than the other finite element solution techniques. Burns et al. [109, 110] formulated a detailed finite element radiative transfer code that uses grids generated by commercial FEM grid generators. The code computes the local radiative flux divergence based on a given temperature distribution in the medium; the computed divergence can then be used in the energy equation in a commercial FEM code for treating
N u m e r i c a l Models.
7.40
CHAPTERSEVEN combinedmode problems. Note that the finite element approach was also used for the solution of the reduced equations of the DO approximations [73, 74, 76].
Hybrid Techniques. Almost all the methods discussed before have certain disadvantages. It is sometimes possible to combine the features of two or more methods to develop a more efficient technique to model radiative heat transfer in furnaces. For example, if the furnace geometry is very complicated, the zone method cannot be used effectively, as it is quite difficult to determine all the required exchange factors. Here, a Monte Carlo technique can be adopted, as done by Vercammen and Froment [111], to calculate the exchange factors between volume and surface elements, and then the radiative heat transfer between each element is calculated using the zonal method. Edwards [112] also suggested a similar approach where he obtained the exchange factors using a Monte Carlo technique and calculated the radiative flux distribution using a radiosityirradiosity approach. With this hybrid approach, exchange factors in most complicated geometries can be calculated with little difficulty. Also, a significant amount of time is saved by using the zone method instead of a Monte Carlo technique to determine the radiative heat flux distribution in the medium. In addition to that, possible statistical errors due to a Monte Carlo technique are avoided. The basic principles of Monte Carlo techniques have been used by Lockwood and Shah [113] in developing the socalled discrete transfer method. Instead of choosing the direction of the intensities originating (due to emission, scattering, or reflection) from each volume/ surface element randomly, they suggested a deterministic approach. Although the method was shown to be accurate and computationally efficient for nonscattering media, it did not yield accurate results if scattering particles were present. Recently, Selquk and Kayakol [114] evaluated this approach and outlined the problems related to its implementation. They noted that, for relatively simple, homogeneous, and nonscattering media, this approach yielded about 10 percent error for the radiative source term near the corners, even if 64 rays were considered. Additionally, it required three times more CPU time than the $4 approximation to converge in threedimensional rectangular enclosures [85]. An alternative approach similar to the discrete transfer method was reported by Taniguchi et al. [115, 116, 117]. They showed that, for nonscattering media, the method yields very good results with significant computational time savings over standard Monte Carlo techniques. Richter [97] suggested a similar semistochastic approach, where he developed a solution scheme based on the principles of a Monte Carlo technique, yet the directions of photons emitted by each volume/surface element were predetermined. He applied this approach to several largescale furnaces and showed that, if the scattering is not accounted for, the model yields reasonable results even if as few as 10,000 photons are considered. Another hybrid approach is to combine the diffusion approximation for optically thick media with the Monte Carlo technique. The diffusion method can be applied in geometric or spectral regions that are optically thick with good accuracy; Monte Carlo is used in geometric or spectral regions with intermediate or thin optical depth and near boundaries. Farmer and Howell [98, 99] have implemented two forms of such a hybrid, finding good accuracy and greatly reduced computer time over conventional Monte Carlo.
Strategies for Choosing a Radiative Transfer Model
One of the most important decisions an engineer has to make regarding the modeling of radiative heat transfer in largescale furnaces is the choice of an appropriate radiative transfer model. Considering the fact that the solution of the RTE requires information about the temperature and spatial distribution of radiative properties in the medium, the radiative transfer model is to be coupled with the models for flow, chemical kinetics, turbulence, and so on. Therefore, the design engineer must choose a model that will be compatible with the solution techniques for the other governing equations. The model should also be reliable and able to predict accurately the radiative flux and the divergence of radiative flux distributions in the medium. In addition, the model should be computationally efficient.
RADIATION
7.41
It is not always necessary or desirable to choose the most accurate radiative transfer model available. If the accuracy of the radiative property data used in predictions is not as good as the accuracy of the model itself, it is difficult to justify the extra computational effort required by a refined model. A simpler approximate model may be more appropriate. It is important to ask the following questions before a specific model is chosen. (1) Is the medium geometry simple? (2) Are there steep temperature and species concentration distributions in the medium? (3) Are there anisotropically scattering particles? (4) If there are, what kind of scattering phase function approximations can be used for them? Having some approximate answers to these questions will help to expedite the selection process. For the solution of the radiative transfer equation, there are several different models available in the literature, as summarized in the foregoing sections. None of these models, however, can be used on a universal basis and applied to all different types of practical problems. It is up to the researcher to decide which model should be used for what type of application. In order to make such a choice, he or she should know the advantages, disadvantages, range of applicability, and versatility of each model. In this section, we will attempt to introduce some simple guidelines to make the selection procedure less timeconsuming. For this purpose, the advantages and disadvantages of different models are listed in Table 7.7 to help the reader in choosing an appropriate model. As discussed previously, the radiative transfer equation is written in terms of radiation intensity, which is a function of seven independent parameters. The RTE is developed phenomenologically and is a mathematical expression of a physical model (i.e., the conservation of the radiative energy). It is a complicated integrodifferential equation. There is no available analytical solution to the RTE in its general form. In order to solve it, physical and mathematical approximations are to be introduced individually or in tandem. We can consider possible approximations under three different types of broad categories: (1) simplification of the spectral nature of properties; (2) angular discretization of the intensity field; and (3) spatial discretization of the medium (for all parameters). It is important to realize that the solution of the radiative transfer equation is required only to obtain the divergence of the radiative flux vector that is a total quantity (i.e., inte
TABLE 7.7
C o m p a r i s o n of R a d i a t i v e Transfer M o d e l s
Angular resolution Flux methods M u l t i flux a p p r o a c h e s Discrete transfer m e t h o d D i s c r e t e o r d i n a t e s approx. YIX DO E v e n p a r i t y / o d d parity D O Moment methods Moment method Spherical h a r m o n i c s approx. double/quadruple SHA Zone methods Monte Carlo techniques Numerical approaches Finite d i f f e r e n c e t e c h n i q u e s
Finite element techniques
Spatial resolution
Spectral resolution
. . . . . . . . . . . . . . . . . . . . ". . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . •. . . . . . . . . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
Scattering medium
. . . . .
. . .
.
.
.
.
.
.
. .
.
.
. .
. . .
.
.
.
. . .
.
. .
.
.
.
. . .
. .
.
3D
. .
. . .
. .
.
.
. .
. . .
. . .
. .
.
.
. .
. .
. . .
2D
.
.
.
. .
.
.
Potential accuracy: . . . . Very good; ° ° ° Good; ° ° Acceptable; ° Not good. (Ratings are subjective and given for applications to multidimensional complex problems; they are likely to change with the increasing availability of faster computers).
7.42
CHAPTER SEVEN
grated over the entire wavelength range spectrum). Since eventually we need to obtain a spectrally integrated quantity, one natural way to simplify the problem is by using spectrally averaged radiative properties. The averaging over the wavelength spectrum can be performed over predefined narrow ranges or over the entire wavelength spectrum. The radiative flux itself is an integrated quantity over a predetermined hemispherical angular domain. After solving the radiative transfer equation in terms of intensity in all directions, the radiative flux is calculated. If a radiative heat transfer model based on fluxes or similar integrated quantities is developed, then the required mathematical complexity will be greatly reduced. The disadvantage of this approach is the eventual suppression of directional variations of intensity. In other words, if there are localized sinks or sources that contribute to radiative transfer, their effect cannot be predicted accurately by such an approximate model. If smaller angular divisions are employed, then the accuracy is increased. Finally, the medium is to be discretized spatially to perform the numerical calculations. It is preferable to employ the same discretization for radiative transfer calculations as for flow and other scalar field calculations. This is a very timeconsuming approach if, for example, the zonal method is used. The multiflux approximations are more useful for this type of discretization. After this brief background, we can start discussing a logical procedure for selecting a radiative transfer model. If the physical system being considered is not large and has relatively simple geometry, and if a high degree of accuracy is required, then a very narrow discretization of angular radiation field can be chosen, and computations can be carried out by considering the RTE over each of these directions. If the medium is not scattering, then the integral term in Eq. 7.105 vanishes, and the problem becomes relatively straightforward and significantly simpler. This approach is likely to yield a highly accurate solution for radiation intensity. To be able to take advantage of such a detailed solution scheme, one must model the radiative property variations very accurately, and the effect of temperature on these properties should be considered. Once the divergence of the radiative flux is determined, it enters into the energy equation as a source term, which affects the temperature profile in the medium directly and the concentration distributions indirectly. Since the medium properties are dependent on the temperature and concentration distributions, the radiative properties need to be updated before each iteration in a comprehensive model. If the medium is scattering, and if the scattering phase function is not mathematically simple, then direct simulation of the radiative transfer equation becomes extremely prohibitive. There are basically two exceptions to this case. If the scattering particles are highly forwardscattering (i.e., most of the energy scattered is in the same direction as the incident beam), then the phase function may be modeled using a 8Eddington approximation. After introducing appropriate scaling laws, the scattering is modeled with isotropic or linearly anisotropic phase functions. In this case, the problem is still tractable, although not simple. This complication is greatly reduced if the optical thickness is small (i.e., the maximum value of the integral of pathlength extinction coefficient product in the medium is less than 0.1 so that a single scattering assumption can be made). The second exception is if we are interested in the propagation of a collimated light source (i.e., a laser). In this case, since only one incident direction is to be considered, the problem can be modeled by direct simulation, even for multiple scattering media up to intermediate optical thicknesses (z ~ 1). In general, the direct simulation of the radiative transfer equation is to be chosen if a fundamental understanding of radiationcombustion or radiationturbulence interactions is required. If the level of accuracy required is not so high, then a Monte Carlo (MC) simulation of the radiative transfer equation can be considered. The MC technique can be effectively used for complex geometries, and it is possible to account for spectral property variations in detail. Its main drawback is the requirement for extensive computational time, which, probably, will not be an important issue in the near future.
RADIATION
7.43
For comprehensive modeling of combustion systems, a relatively less detailed multiflux or moment method is preferable to the statistical MC approach. For example, the $4 discrete ordinates approximation has been successfully used by several researchers for both cylindrical and rectangular systems. The method allows the user to account for sufficient detail in spectral properties of combustion products and reasonably accurate discretization of the angular radiation field, and it is compatible with the finite difference schemes used for flow and energy equations. The $6 approximation yields slightly better results, and its advantage is more visible if there are large temperature gradients and scattering particles in the medium. Of course, its use can be justified only if the properties are known with good accuracy. It is computationally more costly than the $4 approximation. Because of this, an innovative strategy is to be devised to avoid computational difficulties. For example, the S: or $4 approximation can be used for the initial calculations of the radiative flux distribution, and $6 can be used only for the final iterations. This approach is likely to yield more accurate predictions with little computational penalty. For systems with uniform or fixed distributions of properties, the YIX modification of the SN method should be considered [8790]. On the other hand, the use of even higherorder DO approximations is not warranted given the current level of the accuracy and availability of radiative properties. In multidimensional systems, moment, spherical harmonics, and hybrid multiflux approximations usually do not yield results with the accuracy or efficiency of the DO approximations. For nonscattering media, the zone method yields accurate results; however, it may not be directly compatible with the flow and energy equations. This problem has been solved by considering two separate finite difference schemes: one for radiation calculations and the other for flow and energy equations [118]. Another alternative is to solve the RTE using a finite difference or finite element scheme. This approach will guarantee the compatibility of the equations [103, 109]. Also worth consideration is the semistochastic model suggested by Richter [97], which is very fast for nonscattering media, and can easily be extended to account for scattering.
Solutions to Benchmark Problems
The choice of modeling method, particularly when a participating medium is present, is not yet a clear one. Each method discussed above has drawbacks as well as positive attributes. In an effort to clarify the choice of methods, the American Society of Mechanical Engineers has sponsored a series of workshops in the form of technical sessions and discussions. A particular series of problems was proposed by the workshop organizers, Professor Timothy Tong of Colorado State University and Dr. Russ Skocypec of Sandia, Albuquerque. Contributors were asked to provide solutions to the benchmark problems at the first workshop. The problems were simple geometries in one, two, and three dimensions. A participating medium was specified with given spectrally dependent anisotropic scattering properties and a given model for spectral band absorptance. (Edward's exponential wideband approximation for a given mole fraction of CO2 in nitrogen was specified.) A temperature distribution within the medium was given, and the bounding surfaces were specified as cold and black. Researchers were asked to apply their favorite method of solution and provide the workshop with numerical values of boundary heat flux distributions and the divergence of the radiative flux at various locations within the medium. Solutions were presented based on a generalized zonal method, three Monte Carlo solutions, a modified discrete ordinates method (the YIX method), and two specialized approaches. The conclusions noted by Tong and Skocypec [119] are that the boundary heat flux values were in better agreement than were the flux divergence values, but even the boundary fluxes varied by as much as 40 percent among investigators for twodimensional geometries and as much as a factor of 2 for the threedimensional geometries. The poor agreement found in this exercise has led to continued dialogue among all of those concerned. Solutions have
7.44
CHAPTER SEVEN
been reexamined and modified among many of the researchers, and the conclusions at this time are that the major methods provide good agreement (within a few percent) for two and threedimensional geometries for the case of a gray medium with and without anisotropic scattering, but there remain considerable differences in the solutions when spectral effects are included. It is not clear as yet why this is the case. Recently, benchmarking attempts have been extended to more complicated Lshaped configurations [90, 120, 121,122]. Hsu and Tan [90] have also presented comparisons between different approaches and suggested that the ray effects could be minimized by increasing the order of angular quadrature used. There are not many exact solutions to the RTE in multidimensional enclosures. The few exceptions include those by Cheng [123] and Crosbie and Lee [124] for inhomogeneous media and Selquk [125] and Selquk and Tahiroglu [126] for homogeneous systems, although they are all related to "simple" geometries. Benchmarking studies against the "real" data, as those obtained from careful experiments, are still lacking!
RADIATIVE PROPERTIES FOR PARTICIPATING MEDIA The medium that interacts with radiation may contain particles and gases which absorb and scatter the radiant energy. In combustion chambers, for example, soot, char, flyash, coal particles and spray droplets affect the propagation of radiant energy. Among various gases, carbon dioxide and water vapor are the major participants to radiative transfer, both in combustion chambers and in the atmosphere. In this section, we will give a short review for the radiative properties of gases and then present some easytouse formulations.
Radiative Properties of Gases If an electromagnetic wave is incident on a gas cloud, it interacts with the individual molecules. This interaction can be considered as a radiative transition and results in a change of energy level in each molecule. If the molecule absorbs the energy from the EM wave, there may be a transition between the nondissociated molecular states (boundbound transition), between the nondissociated and dissociated states (boundfree transition), or within the dissociated states. For most radiative heat transfer applications, the energy level of interest is such that, when the EM wave is incident on a molecule, it results in a boundbound transition, if any. According to quantum mechanics, this means that, for an EM wave being absorbed by a molecule, it has to have just the right amount of energy to raise the molecular energy state to higher levels. It is also possible to consider this transition as the absorption of a photon (energy quantum). The energy of a photon is expressed as E  hv, where h is the Planck constant and v is the frequency of the wave, which is related to the wavelength via the speed of light: c  Co/n v X  V ~ o / n  v/~. Here, subscript o is used to denote the vacuum, and n is the real part of the complex index of refraction of the medium. It is obvious that a molecule can absorb the energy in discrete amounts of hv or h c / X , which results in a boundbound transition. This suggests that, if a beam of radiation spanning a wavelength interval is incident on a gas cloud, it will lose its energy at certain wavelengths but will not be affected at others. For most gas molecules, there are several hundreds, even thousands, of possible molecular energy states. Therefore, one expects to find a large number of wavelengths at which the molecules absorb the incident energy, which makes the prediction of gas absorption a very difficult problem. Additionally, each of these absorption wavelengths can be broadened because of pressure and temperature as well as the uncertainty principle of Heisenberg. It is obvious that exact consideration of all these active frequencies/wavelengths may not be
RADIATION
7.45
desirable for most engineering applications, although it will be within the realm of possibility with increasing availability of highspeed, largememory computers. Even though most of the discussion given above refers only to gas absorption, it can be readily extended to emission. It should be understood, however, that the emission is related to the gas temperature, whereas absorption depends on both gas temperature and the temperature of the source. As briefly mentioned under ideal conditions above, a molecule absorbs or emits radiation at a fixed wavelength. Then, the change in the energy of the beam incident along path s within the gas cloud is written as (Eq. 7.105 with scattering neglected):
Line Radiation.
dI. ds
 Kn(Ib n  I n )
(7.147)
where I~ is the spectral radiation intensity and 11 (=1/~,) is the wavenumber (usually expressed in units of cm1). The absorption coefficient nv is equal to the emission coefficient based on Kirchhoff's law. The absorption and emission by gas molecules are realized not at a single frequency but over a very narrow band of frequencies. The reason for this is mainly the change in the energy level of molecular states due to molecular collisions, temperature, pressure, or relative motion of molecules with respect to the beam of radiation. The result is the broadening of absorption/emission lines; the most well known are the collision, natural line, and Doppler broadenings. The shape of these narrow spectral lines is exponential, peaked at the center, with rapid decay away from the center frequency. The corresponding line width is in the order of drl = 0.05 cm 1. Here, 11 = 1/~, is the wavenumber corresponding to the wavelength ~,; the unit cm 1 is read as "wavenumbers." For example, at wavelength of ~, = 1 lam, 11 = 10,000 cm 1, and for d~ = 0.05 cm ~, d~, = drl/112 = 0.001 l,tm. The shape and width of each absorption line are functions of temperature and pressure of the medium. The Lorenz profile is usually used to define the shape of these lines at moderate temperatures under local thermodynamic equilibrium conditions. Although other profiles, such as Doppler or Stark profiles, can also be used to define the line shapes, the Lorenz profile, which adequately describes collisionbroadened lines, is more appropriate for most applications, including those in combustion systems. For more details, the reader is referred to Siegel and Howell [1] or Modest [3]. Models. Given that detailed spectral calculations with wavelength resolution on the order of 10 3 ktm is neither computationally efficient nor justifiable for many engineering applications, it is better to develop more affordable models. A close look at the broadened absorption lines depicts that several of them are positioned very close to each other and may overlap, especially for the vibrationrotation transitions of diatomic and polyatomic molecules. The absorption coefficients of individual lines can be added to find the absorption coefficient of the narrowband:
Narrowband
~:n = ~ nnJ
(7.148)
J
Two of the bestknown models used for this purpose are the Elsasser and the statistical Goody models, both of which employ the Lorenz profile for description of individual line shapes. These models give very accurate predictions over a bandwidth of approximately 50 cm 1, which is considered narrow for most practical purposes. (At ~, = 1 l.tm, this bandwidth translates to about A~, = 0.05 ktm.) Because of this, the model is called the n a r r o w b a n d m o d e l Although this technique is significantly simpler than the linebyline models, it still requires an extensive database about the species considered and significant computational effort. Such a detailed model can be considered useful only if the species concentration distribution is known very accurately, which is usually not the case.
CHAPTER SEVEN
7.46
Detailed discussions of narrowband models were given by Tien [127], Ludwig et al. [128], Edwards [129], and Tiwari [130]. The discussion of these approaches can also be found in standard texts [1, 2, 3].
Wideband Models. The narrowband models introduce significant simplification over the linebyline calculations; however, the accurate predictions depend not only on the rigor of the model but also on the accuracy of input data, such as the local temperature, the temperature profile, the partial pressures of the radiating gases, and so on. In most practical systems, these data are not available with good accuracy. This suggests that even simpler approximations may be more appropriate for calculation of gas properties in practical systems. Water vapor, carbon dioxide, and carbon monoxide are the most important contributors to nonluminous radiation in combustion systems. They are the byproducts of any hydrocarbon combustion and absorb and emit radiation selectively only at certain wavelengths. Although there are other gases such as NOx and SO2 present in the combustion products, their partial pressures are very small, and their contributions can be safely neglected in radiative transfer predictions. Among these, SO2 is more important, as it has usually an order of magnitude higher concentration in largescale furnaces than that of NO. The emissivity of SO2 is about the same order of magnitude as the H20, and, therefore, it is possible to account for the SO2 contribution easily by adding its partial pressure to that of H20 and considering only the CO2 and H20 contributions in the calculations [131]. For combustion gases such as water vapor, carbon dioxide, and carbon monoxide, the number of wavelength ranges important for thermal radiation calculations is not excessive. Figure 7.21 shows the important absorption bands of the CO2 and H20 gases at two different temperatures [132]. For H20, there are four important widebands centered at wavelengths of 1.38, 1.84, 2.7, and 6.3 l.tm. For CO2, there are bands at 2.7 and 4.3 ~tm, and, for CO, there is one band centered at 4.4 ~tm. The width of each of these socalled widebands is an adjustable parameter that is determined by assuming the absorption and emission of radiation in the wideband is equal to the effective absorption and emission of several narrowbands present within the band. Therefore, the shape of the band chosen determines the width. Edwards and his coworkers first developed basic concepts of wideband models (see Ref. 129). There are different variations of the model, such as the box or exponential wideband models. The underlying idea in these approaches is to represent the gas absorption over a relatively wide spectral band (about 0.1 to 0.5 ~tm) with a simple function. If this function is an exponential, the resulting model is called as the exponential wideband model [129,133], which is the most wellknown and successful of all different models. It is possible to simplify this even further by assuming absorption remains constant over a prescribed wavelength interval. This yields the box model, which is not as good as the former one, as expected. I00.0
'1
[!
i'
75.0
i
a. 50.0 (%)
"
,
"" "
r
I

.""
t
i :
'
a
50.0
....,,
(%)
'
NB
..t.~ ~ "
'fv,, •
;)5.0 =.
"
..
"~
I
'i o
t"
.t
ii
!!
;i ""
NB
;
ooo
,ooo
v [cm']
25.0
"
': :!
'
:i
ooo

~"
ooo
,oooo
0
2000
~
.n
4000
n 6000
~/[cm']
FIGURE 7.21 The important absorption/emission bands of combustion gases at different temperatures [132].
8000
I0000
RADIATION
7.47
In this section, we will outline the formulation of the EdwardsMenard model, as it can be readily used for different applications.
Formulation of the Exponential Wideband Models.
In exponential wideband models, the band strength parameter, or an integrated absorption coefficient of a given wideband, is defined as: f**
 Jo ~ drl
(7.149)
where ~ is the spectral absorption coefficient, based on either linebyline calculations or narrowband models. Note that the integration is evaluated from 0 to 0% even though the band itself is finite in size. The EdwardsMenard model assumes an exponentially decaying function for the smoothed absorption coefficient (sometimes written as ~:n  (S/d)n, where S and d are parameters defined similar to those used for individual absorption lines). The exponential wideband model assumes that for a symmetric band, ~n is given as: ~:~ = 
0~
e4(.)/~ = 
(0
0~
e2'. cnl;~
(7.150)
t.O
where o~ is defined by Eq. (149) and co is the socalled bandwidth parameter, which is the width of the band at 1/e of the maximum intensity. Most of the absorption bands of the gases of engineering interest are symmetric, and the corresponding rio wavenumbers are listed in Table 7.8 (adapted from Ref. 3). If the bands are not symmetric, the f(rl) function is slightly modified to represent a cut band shape, which simulates the real profile more accurately. Then, either an upper limit flu or lower limit tit is used, and f(rl) is defined as (flu  rl) or (rl  rll). Note that, for only a few bands (flu or rll), values are reported in Table 7.9. (Even though these correlations were developed more than three decades ago, they are still the most convenient and accurate of all wideband model expressions, see Refs. 1 and 3 for details.) The total band absorption is defined as (7.151)
A = fband~"qd~  fo**( 1  e"X) drl with x being the path length. The normalized total band absorptance is:
A *  A/o = A*(o~, ~, %)
(7.152)
Here, x0 = ~:,~cxis the optical thickness at the band center. Table 7.8 lists all the exponential wideband model correlations as derived by Edwards and Menard for an isothermal gas. They are calculated using the equations given below:
• *(T)
[1exp(~=lUkSk)]V(T) Ct0
tx(T) = ~ W*(T0)
[ (21)1 1  exp
Uo,k~)k
(7.153)
W(To)
~(t) = YPe = Yo~/To O(T)
T O(To) Pe
co(T) : ~X/~T0 [[ where
~P(T) =
(7.154) (7.155)
[(v, + gk + ~Sk 1)!/(g, 1)!vk!]e u~v*
k = 1 v k = vo, k
(7.156)
m
~lI ~ k=l
vk=O
[(vk + g k  1)!/(gk 1)!vk!]e u'v'
7.48
CHAPTER SEVEN TABLE 7.8
Exponential Wide Band Correlations for an Isothermal Gas [133]
1
TABLE
7.9
0 < % < 13
A* = %
[3 < x0 < 1/13
A* = 2 V ~ 0 ~  13
Square root regime
1/13 < x0 0) for 8k
:,',~
(.oo,~) i(oo'2,~.~"~.0o~,~\."X.o3~." 4. A recent review of these techniques has been given by Manickavasagam and Mengtiq [174]. Although most particles in combustion chambers are nonspherical, because of large uncertainties in particle shape and material structure (which affects the complex index of refraction), the use of more sophisticated models to determine particle radiative properties is not always warranted. There are special applications where the exact shape and composition of the particle may be very critical to the understanding of the physical phenomena. For example, it is important to know the soot agglomerate structure to determine the soot formation mechanism in flames using laser/light diagnostics. Also important is the effect of pores and material inhomogeneity in pulverized coal particles on the radiationcombustion interaction. In other words, particle shape and structure are very critical in understanding the
RADIATION 7.57 microscale phenomena being observed in flames; however, for predicting the average radiative properties for radiative heat transfer calculations, the use of simpler models may be sufficient. Simple engineering models for coal/char particles were suggested by Buckius and Hwang [175], Viskanta et al. [176], and Mengtiq and Viskanta [177]. In Ref. 175, the coal/char radiative properties were calculated from the LorenzMie theory for different size distributions and complex index of refraction values, and then a curvefitting technique was employed to obtain simple engineering equations. It was shown that these "empirical" equations are valid for a wide variety of size distributions. The same conclusion was also drawn in Ref. 177, where analytical expressions were developed starting from anomalous diffraction theory. More recently, Im and Ahluwalia [178] and Kim et al. [179] presented a series of approximations for pulverizedcoal and particulategas mixtures. Any of these models can be easily incorporated into complicated global computer algorithms. Skocypec and Buckius [180, 181] presented an analytical formulation to obtain the radiation heat transfer from a mixture of combustion gases and scattering particles. They considered band models for the gases and accounted for the absorption and scattering by particles. They developed charts similar to Hottel charts for combustion gases. The results presented can be used to obtain the average radiative properties if the particle loading information is not known accurately. (See also Refs. 182184 for a discussion on the limits of this formulation.) In the following sections, we will present simple techniques to calculate the optical/radiative properties of soot, fly ash, and coal/char particles. Also, the necessary physical parameters will be summarized.
Soot Particles~Agglomerates.
Soot is one of the most important contributors to radiative transfer in combustion chambers. It is formed during the combustion of almost all hydrocarbons, and in combustion chambers it exists almost everywhere. Unlike combustion gases such as H20 or CO2, which absorb and emit only at certain wavelengths, soot participates in radiative transfer at all wavelengths. Therefore, the radiant energy emitted from a sooty flame is significantly more than, for example, a cleanburning natural gas flame. Although this can be considered as an advantage, having unburned carbon particles emitted in the exhaust is a major drawback. In order to predict the contribution of soot to overall radiative transfer phenomena in combustion chambers, one must know the soot shape, size, size distribution, and optical properties. In a combustion chamber, soot volume fraction or number densities are usually not known. It is generally accepted that the primary soot particles are spherical in shape and about 2060 nm in diameter (see the review by Charalampopoulos [185]). However, depending on the flow and combustion characteristics of the system, they agglomerate to form irregularly shaped large particles. The shape can be clusters of spheres or cylindrical long tails attached to burning coal particles. If the agglomeration is not considered, calculation of required radiative properties of soot particles will be straightforward. Since the size of an individual soot sphere is much smaller than the wavelength of radiation, the Rayleigh limit (for small x = 7tD/~,) to the LorenzMie theory can be used. Then, the soot absorption and scattering efficiency factors are given as
Nl
{
24nxk~
Qabs,Rayleigh 12x Mll = x (n 2 _ k~ + 2) z + 4n~k~
]
Qsca,Rayleigh= 8X4(1  3 ~12) where
M1 = 12 + m 212= N 2 + (2 + N2)2
N1 = 2nk 
I(m 2)
(7.183)
(7.184) M2 = 1 + 2N2
N2 = n 2  k 2  R(m 2)
(7.185)
7,58
CHAPTER SEVEN
Selamet and Arpaci [186, 187] have investigated the accuracy of the Rayleigh approximation for soot particles and proposed a simple extension based on the Penndorf approximation. According to their study, for larger particles, the extinction efficiency should be modified to: [(1_551 aext = Qabs.Rayleigh"k2X3 U~ + 3" ~ where
M4 = 4N] + (3 + 2N2)2
4 M6 ]
6 M_M_~/_~ + ff M1 ] + 3  ~ 1 x
(7.186)
M5 = 4(N2  5) + 7N3
M6 = (N2 + N3  2) 2  9N ]
N3 = (n 2 + k2)2= N] + N 2
(7.187)
These equations can be used up to size parameter of x = 0.8 for soot index of refraction range of 1.5 < n < 4.0 and 0.5 < k _> Xc, although this breaks down in certain cases. For electrons in a metal at room temperature, x is on the order of 1014 s whereas for phonons in solids it is on the order of 1011 s. For gas molecules, the mean free time is given as e/v where f is the mean free path and v is the molecule velocity. The mean free path depends on the number density p and collision crosssectional area > x, x~, transport is ballistic in nature and local thermodynamic equilibrium cannot be defined. This transport is nonlocal in space. One has to resort to timeaveraged statistical particle transport equations. On the other hand, if L >> 2, 2 r and t = x, Xr, then approximations of local thermodynamic equilibrium can be assumed over space although timedependent terms cannot be averaged. The nonlocality is in time but not in space. When both L = 2, 2 r and t = x, Xr, statistical transport equations in full form should be used and no spatial or temporal averages can be made. Finally, when both L >> 2, 2~ and t >> "c, xr, local thermodynamic equilibrium can be applied over space and time leading to macroscopic transport laws such as the Fourier law of heat conduction. Let us consider the last case first since that is the easiest one and it also ties the microscopic transport characteristics to the macroscopic world.
KINETIC THEORY Formulation The kinetic theory of transport phenomena is the most elementary and perhaps the first step toward understanding more complex transport theories [1]. Consider a plane z, across which particles travel carrying mass and kinetic energy. Consider two fictitious planes at z + 2z and
8.4
CHAPTER EIGHT
.
z+g
qz
FIGURE 8.1 Schematicdiagram showing energy flux across a z plane used in kinetic theory.
z  ez on either side of the z plane as shown in Fig. 8.1. Here, fz is the z component of the mean free path f which makes an angle 0 from the direction perpendicular to z. On average, the particles moving down from z + fz contain an energy density u that is characteristic of the location u(z + ez), whereas those moving up from z  fz have characteristic energy density u(z  fz). If the particles move with a characteristic velocity v then the next flux of energy in the positive z direction is
qz = 1/2Vz[U(Z  fz)  u(z + fz)]
(8.1)
where vz is the z component of the velocity and the factor 1/2is used since only half of the total number of particles at each location move up from z  ez or down from z + ez. Using Taylor expansion and keeping only the first order terms, one gets
du du qz =Vzez ~z =  ( c ° $ 2 0)k'e dz
(8.2)
where it is assumed that fz = f cos 0 and Vz = v cos 0. Averaging over the whole hemisphere of solid angle 2r~, one gets
duI~f~n fO¢'12 ] 1 du q z "  v ~ ~z o o COS2 0 sin 0 dO dq0 =  ~ vf dz
(8.3)
where tp is the azimuthal angle and 0 is the polar angle and sin 0 dO dip is the elemental solid angle. Assuming local thermodynamic equilibrium such that u is a function of temperature, one can write the flux as
qz =
.
1 du d T . . . ~ v~ d T dz
1 dT 3 Cv~ dz
.
(8.4)
This is the Fourier law of heat conduction with the thermal conductivity being k = Cvf/3. Note that we have not made any assumption of the type of energy carrier and, hence, this is a universal law for all energy carriers. The only assumption made is that of local thermodynamic equilibrium such that the energy density u at any location is a function of the local temperature. The characteristics of the energy carrier are included in the heat capacity C, velocity v, and the mean free path 2. Neglecting photons for now, the thermal conductivity can be written as
MICROSCALE TRANSPORT PHENOMENA k  1/3[(Eve)/+ (Cue)e ]
8.5 (8.5)
where the first term is the lattice contribution and the second term comes from electrons. In the case of electrons in a metal, the electron contribution is dominant, whereas for semiconductors and insulators, the phonon contribution is dominant. For gas molecules, the heat capacity is a constant equal to C = (n/2)pk8 where n is the number of degrees of freedom for molecule motion, p is the number density, and kB is the Boltzmann constant. The rms speed of molecules is given as v = V'3kBT/m, whereas the mean free path depends on collision cross section and number density as e  (p6)l. When they are put together, one finds that the thermal conductivity of a gas is independent of p and therefore independent of the gas pressure. This is a classic result of kinetic theory. Note that this is valid only under the assumption that the mean free path is limited by intermolecular collision.
Thermal Conductivity of Crystalline and Amorphous Solids Since the thermal conductivity depends on C, v, and e, let us investigate the characteristics of these quantities. The electron heat capacity in a metal varies linearly with temperature C = (n2pk2/2EF) T where EF is the Fermi energy of a metal and p is the electron number density. This is a consequence of the free electron theory of metals in which only electrons within an energy range ks T around the Fermi energy EF are responsible for transport phenomena [2]. Here, kB is the Boltzmann constant (1.38 x 10 23 J/K) and T is the absolute temperature. The Fermi energy of most metals falls in the range of 3 to 10 eV whereas the thermal energy kBT is 0.026 eV at room temperature. Hence, only a small fraction of all the electrons in a metal contribute to energy transport in metals. The velocity relevant for transport is the Fermi velocity of electrons. This is typically on the order of 10 6 rrds for most metals and is independent of temperature [2]. The mean free path e can be calculated from e = VF'r,where x is the mean free time between collisions. At low temperature, the electron mean free path is determined mainly by scattering due to crystal imperfections such as defects, dislocations, grain boundaries, and surfaces. Electronphonon scattering is frozen out at low temperatures. Since the defect concentration is largely temperature independent, the mean free path is a constant in this range. Therefore, the only temperature dependence in the thermal conductivity at low temperature arises from the heat capacity which varies as C o~ T. Under these conditions, the thermal conductivity varies linearly with temperature as shown in Fig. 8.2. The value of k, though, is samplespecific since the mean free path depends on the defect density. Figure 8.2 plots the thermal conductivities of two metals. The data are the best recommended values based on a combination of experimental and theoretical studies [3]. As the temperature is increased, electronphonon scattering becomes dominant. The mean free path for such scattering varies as e o~ T" with n larger than unity. The mean free path of electrons at room temperature is typically on the order of 100 A. The mean free path depends on the material but is independent of the sample, since electronphonon scattering is an intrinsic process. As a result of electronphonon scattering, thermal conductivity of metals decreases at higher temperatures. Although the lattice heat capacity in a metal is much larger than its electronic contribution, the Fermi velocity of electrons (typically 10 6 m/s) is much larger than the speed of sound (about 10 3 m/s). Due to the higher energy carrier speed, the electronic contribution to the thermal conductivity turns out to be more dominant than the lattice contribution. For a semiconductor, however, the velocity is not the Fermi velocity but equal to the thermal velocity of the electrons or holes in the conduction or valence bands, respectively. This can be approximated as v  X/3kBT/m*, where m* is the effective electron mass in the conduction band or hole mass in the valence band. This is on the order of 105 m/s at room temperature. In addition, the number density of conduction band electrons in a semiconductor is much less than
8.6
CHAPTEREIGHT
10
.
.
.
.
.
.
Copper 10 = m
>
1
= m
o
ID o
¢0
Aluminum
10
w
E tO
IDefect Scattering
10
Phonon Scattering
o 10
o
10
1
Temperature
10
2
10
T (K)
FIGURE 8.2 Thermal conductivity of aluminum and copper as a function of temperature [3]. Note that at low temperature, the thermal conductivity increases linearly with temperature. In this regime, defect scattering dominates and the mean free path is independent of temperature. The thermal conductivity in this regime depends on the purity of the sample. The linear behavior arises from the linear relation between the electronic heat capacity and temperature. As the temperature is increased, phonon scattering starts to dominate and the mean free path reduces with increasing temperature. To a large extent, the thermal conductivity of a metal is independent of the purity of the sample.
the electron density in metals. Hence, the electronic heat capacity is also lower than that of metals. This leads to the fact that electrons play an insignificant role in heat conduction in semiconductors. Therefore, as far as heat conduction is concerned, semiconductors and insulators fall in the same class of materials. Phonons are quanta of crystal vibration [2,4]. The physics of phonons is quite similar to that of photons in that they follow BoseEinstein statistics. However, there are some key differences, namely: (1) phonons have a lower cutoff in wavelength and upper cutoff in frequency whereas photon wavelength and frequency are not limited; (2) phonons can have longitudinal polarization whereas photons are transverse waves; (3) p h o n o n  p h o n o n interaction can emit or annihilate phonons and thereby restore thermodynamic equilibrium. Despite these differences, heat conduction by phonons can be studied as a radiative transfer problem. Figure 8.3 shows the thermal conductivity of crystalline diamond samples with different defect concentrations [5]. At low temperatures, the thermal conductivities of all the samples are nearly equal and follow the T 3 behavior [2,4,6]. This arises from the Planck distribution of phonons at low temperatures. The dominant phonon wavelength at low temperatures can be very large as suggested by the relation ~,dom= hvs/3kBT. This is essentially the Wien's displacement law applied to phonons. Hence, the dominant wavelength at low temperatures can be much larger than crystal imperfections such as point defects, dislocations, and grain boundaries. Therefore, mean free path is not limited by the defect scattering but by the size of the crystal. Hence, the mean free path at very low temperatures is temperatureindependent.
MICROSCALE TRANSPORT PHENOMENA
8.7
Phonon velocity is constant and is the speed of sound for acoustic phonons. The only temperature dependence comes from the heat capacity. Since at low temperature, photons and phonons behave very similarly, the energy density of phonons follows the StefanBoltzmann relation t~Ta/vs, where 0 is the StefanBoltzmann constant for phonons. Hence, the heat capacity follows as C o~ T 3 since it is the temperature derivative of the energy density. However, this T 3 behavior prevails only below the Debye temperature which is defined as 0o = hcoo/kB. The Debye temperature is a fictitious temperature which is characteristic of the material since it involves the upper cutoff frequency coo which is related to the chemical bond strength and the mass of the atoms. The temperature range below the Debye temperature can be thought as the quantum requirement for phonons, whereas above the Debye temperature the heat capacity follows the classical DulongPetit law, C = 3~kB [2,4] where 11is the number density of atoms. The thermal conductivity well below the Debye temperature shows the T 3 behavior and is often called the Casimir limit.
10
3
Diamond 10 2
oo 10
Increasing Defect Dens
O
O/mO
=me
>
, m
O "O e= O
o
1
] 10
0
O me=
E !
m
10
1
F
10
2 10
0
Boundary
Defect
Scattering
Scattering
.
.
.
. 10
. 1
.
.
.
.
. 10
2
.
. 10
Temperature T (K) F I G U R E 8.3 Thermal conductivity of diamond as a function of temperature [5]. The solid line represents the T 3 behavior.
As the temperature is increased, ~,dombecomes comparable to the defect sizes and, hence, the defect concentration in the crystal determines the thermal conductivity. This is the regime near the peak of the thermal conductivity which occurs when the temperature is on the order of T = 0o/10. Although defect scattering is temperatureindependent, phononphonon interactions are highly temperaturedependent. There are two types of phononphonon interactions, namely, normal and Umklapp. Normal or Nscattering conserves energy and momentum during collision of two or more phonons, whereas Umklapp or Uscattering conserves energy but does not conserve momentum. Although normal scattering does not directly pose any resistance, it distributes the phonon energy to higher frequencies~ As the temperature is increased beyond 0o/10, C becomes a constant (=3rlkB). Also, the phonon density at high frequency and large wave vectors becomes sufficiently high that phononphonon Umklapp scattering dominates and determines the phonon mean free path.
8.8
CHAPTEREIGHT Hence, the mean free path decreases drastically with temperature which results in a sharp drop in thermal conductivity with increasing temperature. Figure 8.4 shows the thermal conductivity for quartz (crystalline SiO2) and amorphous silica (aSiO2) [7]. The quartz data follows the T 3 behavior at low temperature, peaks at about 10 K, and then drops with increasing temperature. As discussed before, this is the expected trend for a crystalline solid. However, amorphous silica behaves very differently. The value of the thermal conductivity is much lower than that of the crystalline sample for all values of temperature. In addition, the temperature dependence of the conductivity is also vastly different. Hence, the model proposed for crystalline solids cannot be applied for such a case. Note that the relation k = Cve/3 is still valid although the heat capacity and the mean free path cannot be determined by relations used for crystalline solids. In 1911, Einstein proposed a model for heat conduction in amorphous solids. In this model, he assumed that all the atoms vibrate as harmonic oscillators at the same frequency toE. In addition, he also assumed that a particular oscillator (or atom) is coupled to only first, second, and third nearest neighbors. Hence, the vibrational energy of the oscillator can only be transferred to these atoms. A further assumption was that the phases of these oscillators were uncorrelated and were completely random. Using these assumptions, he derived the thermal conductivity to be
k~ 111/3
ke
x2e x
=     O E ~ h /I; (eX 1) 2
(8.6)
where 11 is the number density of oscillators, 0E is the Einstein temperature, and x = 0e/T. The Einstein temperature is defined as 0E = hWF./kB. Unfortunately, the predictions of this theory
10 2
~'
1o I
~
10 0
~, ,~__
10 1
=l= • 1'
o
10
I
i
axis/
aquartz
II ~
/~
ro •
10
i
"'.%
'.%_
2
Silica
"o
i=: o
i
~ /
3
0 ¢=
104
,I~
10 5 106 10 2
T2
CahillPohl :in
/ 101
10 0
10 1
10 2
10 3
Temperature T (K)
FIGURE 8.4 Thermalconductivityof quartz and glassysilica as a function of temperature [7]. The quartz thermal conductivity exhibits a T 3 behavior at low temperature, a peak at about 10 K, then reduction at higher temperatures. This is typical of a crystalline solid. For amorphous glass the thermal conductivity increases as T2 plateaus between 1 to 10 K and then increases monotonically with temperature. Also plotted are the predictions of the CahillPohl and Einstein models. The CahillPohl model provides accurate predictions for temperatures higher than 50 K but cannot predict the low temperature behavior. The Einstein model predictions are much lower than the measured values.
MICROSCALETRANSPORTPHENOMENA
8.9
were far below those of the measured data as can be observed in Fig. 8.4. The major flaw in this theory was in the assumption that the phases of the neighboring oscillators were uncorrelated. The success of the Debye theory was based on the fact that it considered the coherence of a crystal wave for a distance on the order of a mean free path. Cahill and Pohl [8,9] recently developed a hybrid model which has the essence of both the localized oscillators of the Einstein model and coherence of the Debye model. In the CahillPohl model, it was assumed that a solid can be divided into localized regions of size ~,/2. These localized regions were assumed to vibrate at frequencies equal to 0)= 2rtVs/~ where vs is the speed of sound. Such an assumption is characteristic of the Debye model. The mean free time of each oscillator was assumed to be onehalf the period of vibration or x = rt/o~. This implies that the mean free path is equal to the size of the region or ~./2. Using these assumptions, they derived the thermal conductivity to be
kce =
kBTI~3Z Vsi i
~
1)(e x ~ d x
(8.7)
JO
where x = Oi/T and 0 i = vsi(h/kB)(6rrq) v3 is a characteristic temperature equivalent to the Debye and Einstein temperatures. The summation is over the two transverse and one longitudinal polarizations for which the speeds of sound can be different. Note that the agreement between the CahillPohl model and the measured thermal conductivity above 50 K is excellent. The validity of the model is further verified by comparing the measured and predicted thermal conductivities of several amorphous solids at 300 K. Below 50 K, there exists a plateau in the thermal conductivity and a sharp drop at temperatures below 10 K. In the limit T > x, xr is assumed, then the most common simplification is to drop the timevarying term in Eq. 8.12. In addition, if L >> e, er is assumed, then the gradient term can be approximated as Vf = Vf0 such that in the onedimensional case, the BTE can be solved to yield /)f0 f = f o  "CVx bx
(8.16)
where Vx is the x component of velocity. This can be called the quasiequilibrium approximation. The only term that contains lack of equilibrium is the scattering term. Local thermodynamic equilibrium is inherently implied by the approximation df/dx = dfo/dx. However, since the local equilibrium f0 can be defined only over a length scale er the approximation finally boils down to df/dx = Afo/£ r. This and the timescale approximations are also made in the kinetic theory, and, hence, one should expect the same results. Since the equilibrium distribution is a function of temperature, one can express
Ofo a f o O T _  ~)x dT i)x
(8.17)
This leads to the energy flux
OT dfo qx(X) =  ~x f v2x  ~ e D(e) de
(8.18)
The first term containing f0 drops out since the integral over all the directions becomes zero. Equation 8.18 is the Fourier law of heat conduction with the integral being the thermal conductivity k. If one assumes that the relaxation time and velocity are independent of particle energy, then the integral becomes
k = vEx  ~ e D(e) de = vEx
~
e D(e) de = ~ Cv2'l~
(8.19)
This is exactly the kinetic theory result k = Cve/3. Similar derivations and conclusions can be made for Fick's law. Ohm's law is characterized by the relation J = oE where Jl is the current density vector at any point in space, E is the electric field vector, and o is the electrical conductivity. The Fourier law is the energy analog of Ohm's law due to the following reasons. The electric field vector E can be written as the negative gradient of the electric potential E   V ~ and hence is analogous to the negative gradient of temperature. The energy flux vector q in the Fourier law is analogous to the current density vector J in Ohm's law. Using kinetic theory, it can be shown that the electrical conductivity follows the relation o
rlee2Xm m
(8.20)
where tie is the density of electrons, e is the electron charge, Xm is the momentum relaxation time, and m is the electron mass. The ratio of the thermal and electrical conductivity can be expressed by the WeidemannFranz law [2] which is given as follows, where the right side contains only physical constants and is known as the Lorenz number [2]. ~JT  3
~
= 2.44 x 108 Kf~_ ~
(8.21)
Note, however, that electrical conductivity is related to the m o m e n t u m relaxation time, whereas the thermal conductivity is related to the energy relaxation time. They are usually close at room temperature or at very low temperatures.
8.12
CHAPTEREIGHT
Hyperbolic Heat Equation If the Boltzmann transport equation is multiplied by the factor integrated over energy, then the equation transforms into
3qx3__7_ + fj
~x
= _ jf
vxeD(e)de on both sides and
fv,,e O(e) de
"[(X, 13)
(8.22)
The acceleration term is dropped in this equation. Consider the situation that L >> e, er and t  x, x~. Now make the following assumptions: (1) the relaxation time is independent of particle energy and is a constant; and (2) the quasiequilibrium assumption is made for the term igfl3x = (dfoldT)(3T/~gx). Then Eq. 8.22 becomes
3qx
qx
k 3T
37+ =x x ~3x
(8.23)
This is the Cattaneo equation [10], which, in combination with the following energy conservation equation,
OT
Oqx
C  ~ + ~x = 0
(8.24)
leads to the hyperbolic heat equation of the form [10] 32T 3T k 32T x~+ ~tC 3x 2
(8.25)
The solution of Eq. 8.25 is wavelike, suggesting that the teg_m_12erature field propagates as a wave. The speed of propagation of this wave is equal to Vk/Cx which also happens to be the speed of the energy carrier, for example, the speed of sound for phonons. So, this model is nonlocal in time but local in space since the temperature represents a spatially localized thermodynamic equilibrium.
Mass, Momentum, and Energy ConservationHydrodynamic Equations The conservation equations that are encountered in fluid mechanics, heat transfer, and electron transport can be derived as different moments of the BTE [11]. Consider a function ~(p), which is a power of the particle momentum ~(p) = pn where n is an integer (n = 0, 1, 2 , . . . ) . Its average can be described as 1 (~(p)) = p f ,(p)f(p)d3p
(8.26)
where P is the number density of particles. The BTE is now multiplied by ~(p) and integrated over momentum. In general form, this gives the moment equation ~gt
+ m V. (p(p~))  pF " 3~ = p ~ [(W(p,p')~(p')) (W(p',p)~(p))] p"
(8.27)
Note that the momentum of each particle can be divided into two components as follows P=Pd+Pr
(8.28)
where Pd is the average or drift momentum corresponding to a collective motion of particles in response to an external potential gradient, and Pr is the random component of the momentum which arises due to thermal motion and is responsible for diffusion. Note that (p) = Pd since the average of the random component over all the momentum space is zero.
MICROSCALETRANSPORTPHENOMENA 8.13 The zeroth moment is when n = 0 and ¢(p) is a constant. Using this, one gets the continuity or number conservation equation which is Op i)t+ V . (pVd) = So  Si
(8.29)
where Vd is the drift velocity, So is the source or rate of generation rate of particles and Si is the sink or removal rate of particles. In the case of fluids, there are no sources or sinks and hence the fight side is zero. However, when electrons and holes are considered, the zeroth moment equation can be written for each valley of the electronic structure of a semiconductor or a metal. Intervalley scattering due to electronelectron, electronhole, electronphoton, or electronphonon interactions may be responsible for particle exchange between the different valleys and bands. This creates a source or sink in each valley in which case the right side of Eq. 8.29 may be nonzero. However, if all the valleys and bands are considered together, the right side would be zero since charge or mass must be conserved. The m o m e n t u m conservation equation is obtained by taking the first moment, ¢(p) = p = my. This yields the following equation ~(ppd) 1 (~(pp)) ~7+   V (p(pp))  pF = m /)t
(8.30) scat
The second term is the average of a tensorial quantity. However, since the average of odd powers of Pr is zero, we get (pp) = PdPd + p2~i/where ~5i/is the unit tensor. The third term of the left side is what is referred to as a body force term in fluid mechanics. It is perhaps more appropriate to refer to it as a potential gradient term since a thermodynamic force can be written as a gradient of any potential F =  V U. The potential U is a sum of the gravitational potential G, electrochemical potential ~, and so forth. For electrons in a metal or semiconductor, the force can be due to electric or magnetic fields which can also be expressed as a gradient of a potential. The fight side of Eq. 8.30 is the scattering term. Under the relaxation time approximation, the right side can be assumed to follow
(()(~tP))scat ~  DP,~m
(8.31)
where Xmis the m o m e n t u m relaxation time. Therefore, the m o m e n t u m conservation equation becomes
O(OPd_._____~) 1 1 /)t +   V . (PPdPd) +   V(Op 2) =  o V ( G + * + "")  OPd m m '~m
(8.32)
The third term on the left side has the form of the kinetic energy of the random particle motion and is representative of the pressure of the particles. Therefore, Eq. 8.32 can be rewritten in the following form:
~)(pmvd) ~)~+ V
" (pmVdVd)=pV
( ?P+ G + ~ +
"")
pmvd T,m
(8.33)
The second term on the left side is often referred to as the advection term. When this is negligible, Eq. 8.33 under zero acceleration reduces to the form Vd =  ~
m
V
+ G + • + "'"
(8.34)
In the case of electron transport where • = e g + kBT In (p) is the electrochemical potential, one can derive the familiar driftdiffusion equation [11]: J = oe2xm E + pe~m V ( k B T l n O) m m
(8.35)
8.14
CHAPTEREIGHT Here, the first term is the drift term representing Ohm's law with the electrical conductivity being o = pe2"Cm/m. The second term is the diffusion term which gives rise to thermoelectric effects and current flow due to electron concentration gradients. In the case of fluid transport, neglecting the left side of Eq. 8.34 gives v =
m
V
(8.36)
which is equivalent to the Darcy equation for flows in porous media. It is evident that Eq. 8.33 has the familiar form of the NavierStokes equation except for the last term involving collisions. The NavierStokes equation can be derived from the BTE using the ChapmanEnskog approximation where the right side of Eq. 8.33 leads to the diffusion term [1]. The energy conservation equation is obtained from Eq. 8.27 if the second moment is taken t~(p) = p2 since energy e p2/2m. This yields the following equation: =
~9_~ /)t + VJ~ =  p F  V d + p ~p"
P'2/ W(p,p') 2 m /
W(p',p)
~m
(8.37)
where ~ = pe is the energy density in J/m 3 and J~ is energy flux vector in W/m 2 which can be expressed in general form as Jg = Vd~ + q
(8.38)
The term va~ is the advection of energy which comes from the drift contribution, and q is the heat flux vector due to diffusion which arises from the random motion of the particles. This reduces the energy equation to ~[ + V. (Vd~) = pVd" V U  V q + ~
so ~
Si
where U is the sum of all the potentials discussed earlier. Here, the scattering term from Eq. 8.37 is divided into an energy source and an energy sink term which are discussed shortly. The first term on the right side is the work done by a force on the particles and, therefore, must appear in the energy conservation equation. To obtain a relation for q, the next higher moment of the BTE needs to be taken. However, closure is often obtained by assuming the Fourier law q =  k V T. But, recalling the fact that the Fourier law is derived under the assumption of quasiequilibrium in both space and time, this may not always be a valid assumption. A higherorder relation which takes into account nonlocality in time but quasiequilibrium in space is the Cattaneo equation for heat flux described in Eq. 8.23. The energy density ~ of a particle system has contribution from entropic motion as well as from the drift and can be written as 3
1
= ~ pkBT + ~ pray 2
(8.40)
Note that the factor 3/2 is valid for particles such as monoatomic gas molecules and electrons, with only three degrees of freedom of motion, each degree possessing an energy of kBT/2. By multiplying the momentum conservation equation (Eq. 8.33) by Vd and subtracting it out of the energy conservation equation (Eq. 8.39), the thermal energy conservation equation can be derived as
3t + Vd " V T+ ~ TV " Vd= 3pk~ V ' q + ~ so
~
Si
Note that the work term pVd" V U drops out since work increases mechanical energy but does not increase entropy or temperature of a system. Only when this work is dissipated by scattering, the entropy of the system is raised and the temperature increases. The scattering term in Eq. 8.41 can be written as follows [12]:
MICROSCALE TRANSPORT PHENOMENA
1/ So ~ Si = 
"r,E(T~v)
+ ~B
'~m  '~e
8.15
(8.42)
where To is a reservoir temperature and "l~e is the energy relaxation time. The first term on the right side is simply the energy relaxation term with respect to an equilibrium temperature To. The second term comes from the difference between the momentum and energy relaxation processes. The energy relaxation time is different from the momentum relaxation time since a collision may change the particle momentum but not its energy. Even if both these times were the same, the term would be nonzero. This is the contribution of the kinetic energy of the particle to the temperature rise. Hence, this is the fraction of the work done which is dissipated resulting in entropy generation and temperature rise. If we consider electrons that are energized by the work done by an external electric field, the electronphonon interactions eventually dissipate this work and result in energy loss to the phonons. Hence, the reservoir temperature is that of the phonons. Note that although the work term pv. F is not present, the term (2mv2a/3kB'r,m) in Eq. 8.42 represents the dissipated work which adds thermal energy to the system. The ratio (X~/Xm) can be called the Prandtl number Pr of the fluid since the fluid diffusivities are inversely proportional to their respective relaxation times.
Equation of Radiative Transfer for Photons and Phonons Photons and phonons do not follow number conservation as do electrons and molecules. However, they do follow energy conservation. An intensity of photons or phonons can be defined as follows
Ik(r,k,Lt) = v(k,s)f(r,k,s,t)hro(k,s)
(8.43)
where Ik is the intensity with wave vector k, v is the velocity at wave vector k, s is the polarization, and h o is the energy. The intensity can also be defined in terms of frequency o and angle (0,~) in polar coordinates corresponding to the direction of vector k as follows: Io,(r,o,0,~,~,t) = v(o,O,~,s)f(r,o3,0,~,t)hoO(o,s)
(8.44)
where D(co,~) is the density of k states in the frequency range co and co + do. If the BTE of Eq. 8.8 is multiplied by the factor v(c0,0,~,0hoD(o,0, the following equation is obtained:
3l~o(r,o),O,,,s,t) (~Io I ,gt + v . Vlo,(r,o,O,C,s,t) = \~/scat ( 0Io ] where
(8.45)
[W(o3",0',¢'.~" ~ o3,0,¢.Olo,(r,o)',O',O/,s',t)l
~,~/~,~t= (co',o',¢',s') ~ L_w(o),o,o?,s_~ro,,o,,op,~,)io(r,o),o,¢,s,t)j
+ ~ [W(j,f~ >o3,0,~,Oe(r,j,fLt )  W(o,O,~,s >j,f~)Io,(r,o,O,~,s,t)] (~ta)
(8.46)
Here, each W is a scattering rate. It is evident that the scattering term is quite complicated and needs explanation. Equation 8.45 is the conservation of energy based on the intensity at frequency o, polarization s, and direction (0,¢). Consider now the first summation in Eq. 8.46. This increases the intensity I,o(r,o,O,~,s,t) due to scattering in frequency co' >o, polarization s' >s, and direction 0', ~' > 0, ¢. The second term is the loss of intensity I,o(r,o,O,~,s,t) due to scattering to other frequencies, polarizations, and directions. Note, however, that if photons are considered, then this term represents photonphoton scattering and not scattering between photons and other particles. So, this term accounts for scattering among the particle type, either photonphoton or phononphonon, respectively. This is often known as the inscattering term in photon radiative transfer, although scattering is usually considered only in direction 0', ~' > 0, ¢~
8.16
CHAPTER
EIGHT
and not in frequency and polarization. This is because, inelastic photon scattering is normally ignored in engineering calculations unless processes such as Raman scattering are involved. For phonon radiative transfer, however, inelastic scattering such as normal and Umklapp processes are very common and must be accounted for in this term. In addition, such phononphonon scattering is often between different phonon polarizations as allowed by phonon energy and momentum conservation during the collision. The second summation term in Eq. 8.46 is for increase or decrease in intensity Io,(r,(o,O,O,s,t) due to interactions with other particles. The particle type is given a tag ], and the phase space defined by momentum and direction is given a tag f~. For example, an energetic electron in a metal or in the conduction band of a semiconductor can drop in energy by emitting a phonon of a certain polarization (e.g., LOphonon) due to electronphonon interactions. Here, the electron is given a tag j and the phonon is given a tag f2. The frequency, direction, and polarization of this phonon is decided by energy and momentum conservation of the scattering process. In photon radiative transfer, this term is often referred to as the blackbody source term. This is true for the particular case of blackbody radiation. However, in a device such as a semiconductor laser or a lightemitting diode, photons are not emitted in a blackbody spectrum but within a certain spectral band that is decided by the semiconductor electronic band structure. Hence, this term is kept as a general emission term in Eq. 8.46. Similarly, there is a loss term when a phonon or a photon is absorbed by another particle and removed from the system. It is clear that in the most general form as described in Eq. 8.45, the scattering terms pose difficulty for solving. Therefore, the relaxationtime approximation is usually made for convenience, in which case the equation of radiative transfer reduces to
3Io,
I~
Io, e(j,f~)
'1;s
"1;a
37 + v. VI,, . . . . .
+
"1;e
+
ffI~(r,m',O',t) to',O'
'1;s
dO"
dt.o' ~ 4x
(8.47)
The first term on the right side is the outscattering term with % being the scattering relaxation time, the second term is the photon/phonon absorption (or transfer of energy to other particles such as electrons, or photons to phonons or phonons to photons, etc.) where % is the absorption time, the third term is the emission term with 1/% being the emission rate. Here, energy from other particles is converted and contributed to the intensity 1,0. The last term is the inscattering term from other frequencies and solid angles O'. In an even simpler form, the equation of radiative transfer can be written as
aI~ I°I~ I~ e(j,n) + v. Vl,, + ~ 3t Xs "Ca T'e
(8.48)
where the inscattering term is totally ignored but it is assumed that (o' ~ m scattering restores equilibrium that is represented by I °. This is often the assumption made in phonon radiative transfer where interfrequency scattering restores phonon equilibrium. The equation of radiative transfer will not be solved here since solutions to some approximations of the equation are well known. In photon radiation, it has served as the framework for photon radiative transfer. It is well known that in the optically thin or ballistic photon limit, one gets the heat flux as q  ~(T 4  T~) from this equation for radiation between two black surfaces [13]. For the case of phonons, this is known as the Casimir limit. In the optically thick or diffusive limit, the equation reduces to q   k p V T where kp is the photon thermal conductivity. The same results can be derived for phonon radiative transfer [14,15].
NONEQUILIBRIUM ENERGY TRANSFER The discussion in the previous sections concentrated on transport by a single carrier, that is, heat conduction by electrons or phonons, charge transport by electrons, and energy transport
MICROSCALETRANSPORTPHENOMENA
8.17
by photons. Relatively little attention was paid to energy transfer processes between the energy carriers. For example, Joule heating occurs due to electronphonon interactions whereas radiative heating involves photonelectron and electronphonon interactions. These are examples of what are commonly called heat generation mechanisms. Traditionally, it is assumed that electrons and phonons within a solid are locally under equilibrium such that a heat generation term can be added to the energy conservation equation. For example, Joule heating during electron transport is usually modeled as IER where I is the current and R is the electrical resistance. Such a term is added to the energy conservation equation for the whole solid. Such an equation uses a single temperature T to describe the solid at a point r and time t. It inherently assumes that there is equilibrium between the electrons and the phonons. However, this is not quite the picture in many cases. The equilibrium between electrons and phonons can be disrupted by several processes. For example, in the presence of a sufficiently high electric field, electrons can be energized and thrown far out of equilibrium from the phonons. Such nonequilibrium conditions can now be achieved in contemporary technology where electronic devices with submicrometer feature sizes undergo highfield transport. In the case of radiative heating in a metal, for example, the electrons are again thrown out of equilibrium from the lattice due to excitations by ultrashort laser pulses that are on the order of 100 fs. Such lasers are now available and are widely used in physics, chemistry, and materials processing. Therefore, it is clear that when modern engineering systems involving transport phenomena become small and fast, the energy diss