HEAT TRANSFER A Basic Approach M. NECATI OZISIK Profesor, Mechanical and Aerospace Engineering "North Carolina State Untersity McGraw-Hill Book Company [New York St. Louis San Francisco Avekland Bogoté Hamburg [London Madrid Mexico Montreal New Debt Panama Paris Sto Paulo Singrpore Sydney Tokyo Toronto pp NeHEAT TRANSFER ‘A Basle Approach INTERNATIONAL EDITION Copyright © 1985 Exclusive rights by McGraw-Hill Book Co — Singapore for ‘manufacture and export. This book cannot be reexported from the country to which itis consigned by McGraw-Hill. 567890KHL 943210 Copyright® 1985 by McGraw-Hill, Inc. All ighs reserve. 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 Ina data base or retrieval system, without the prior written permission of the publisher. ‘This book was set in Times Roman. ‘The editors were Anne Murphy and Madelsine Eichberg. ‘The production supervisor was Leroy A. Young. Library of Congress Cataloging in Publication Data zisk, M. Necat. Heat transfer. Includes bibliographical references and indexes. ‘When ordering this tie use ISBN 0-07-Y66460-9 Printed in Singapore.To GUL and HAKANChapter 2 a 22 CONTENTS Preface Introduction and Concepts ‘Conduction Convection Radiation ‘Combined Heat Transfer Mechanism Units, Dimensions, and Conversion Factors Summary of Basic Relations Problems References Shor jography of Textbooks in Heat Transfer Conduction—Basic Equations One-Dimensional Heat Conduction Equation Three-Dimensional Heat Conduction Equation Boundary Conditions Summary of Basie Equations Problems References One-Dimensional, Steady-State Heat Conduction The Slab The Cylinder The Sphere Composite Medium “Thermal Contact Resistance Ciel Thickress of Fslation3s 3 39) Chapter 4 4 3 45 “1 48 Chapter 5 sa 32 33 4 ss 56 3} Chapter 6 Fined Surfaces ‘Temperature Dependent K(7) Summary of Basic Relations Problems References Transient Conduction and Use of ‘Temperature Charts, LLumped-System Analysis Slab--Use of Transient-Temperature Charts Long Cylinder and Sphere—Use of Transient:Temperature Chars ‘Semi-infnte Solid Use of Transent-Temperature Charts Product Solution Use of Transient-Temperature Charts ‘Two-Dimensional, Steady-State Heat Conduction — Use of Conduction Shape Factors “Transient Heat Conduction ina Stab — Analytic Solution Transient Heat Conduction ina Siab—Use of Tabutated Solutions Problems References Finite-Difference Methods for Solving Heat Conduction Problems ‘One-Dimensional, Steady-State Heat Conduction Boundary Conditions ‘Two-Dimensional, Steady-State Heat Conduction “Methods of Solving Simultaneous Algebraic Equations Cylindrical and Spherical Symmetry Unsteady Heat Conduction Explicit Method Unsteady Heat Conduction Implicit Method Problems References Convection—Concepts and Basic Relations Flow over a Body Flow inside a Duet Concepts on Turbulence Equations of Motion Equation of Energy Dimensionless Parameters Boundary-Layer Equations Problems References 226 2 238 2s 253 261 261 2m mChapter 8 81 82 83 85 87 Chapter 9 4 93 4 9s 37 99) 9.10 on 912 Chapter 10 wor 102 Forced Convection for Flow inside Ducts Hydrodynamicallyand Thermally Developed Laminar Flow Thermally Developing, Hydrodynamically Developed Laminar Flow Simultaneously Developing Laminar Flow Turbulent Flow inside Ducts ‘eat Transfer to Liquid Metals ‘Analogies between Heat and Momentum Transfer in ‘Turbulent Flow Heat Transfer Augmentation ‘Summary of Correlations Problems References Forced Convection for Flow over Bodies Drag Coeficient for Flow over a Fat Pate Heat Transfer Coefficient for Flow over a Flat Pate Flow acrogs a Single Circular Cylinder Flow across a Single Noncircular Cylinder Flow across a Single Sphere Flow across Tube Bundles Heat Transfer in High-Speed Flow over a Flat Plate ‘Summary of Correlations Problems References Notes Free Convection Dimensionless Parameters of Free Convection ‘An Approximate Analysis of Laminar Free Convection on 8 Vertical late Correlations of Free Convection on a Vertical late Free Convection on a Horizontal Pate Free Convection on an Toclined Pate ‘Free Convection on a Long Cylinder Free Convection on a Sphere ‘Simplified Equations for Air “Mechanism of Free Convection in Enclosed Spaces Correlations of Free Coavestion in Enclosed Spaces Combined Free and Forced Convection ‘Samumary of Correlations for Heat Transfer in Free Convection Problems References Boiling and Condensat Film Condensation Theory ‘Comparison of Film Condensation Theory with Experiments nm 231 22, 9 30s 310 a 38 332 336 39 37 382 382 361 416 418 6 nm 83 inane nner103 104 los 104 107 108 109 Chapter 11 M1 na 13 4 ins 116 9 18 Chapter 12 Ba 122 123 24 2s 126 27 28 D9 10 ma Riz iim Condensation inside Horizontal Tubes Dropwise Condensation ‘Condensation in the Presence of Noncondensable Gas Poo! Boling Regimes Pool Boiling Correlations Forced-Convecton Boiling inside Tubes Summary of Equations Problems References Heat Exchangers Classification of Heat Exchangers ‘Temperature Distribution in Heat Exchangers Overall Heat Transfer Coeicient ‘The LMTD Method for Heat Exchanger Analysis Correction for LMTD for Use with Cross-Flow and Multipass Exchangers e-NTU Method for Heat Exchanger Analysis ‘Compact Heat Exchangers Heat Exchanger Optimization Problems References Radiation among Surfaces in a Nonparticipating Medium Nature of Thermal Radiation Blackbody Radiation Radiation from Real Surfaces Radiation Incident on a Surface Radiation Properties of Surfaces Solar Radiation Concept of View Factor Methods of Determining View Factors Network Method for Radiation Exchang in an Enclosure ‘Radiosity Matrix Method for Radiation Exchange in an Enclosure ‘Correction for Radiation Effects in Temperature Measurements ‘Summary of Equations Problems References Radiation in Absorbing, Emitting Media Equation of Radiative Transfer Transmisivity, Absorptivity, and Emissivity Radiation Exchange between a Gas Body and Its Black Enclosure 48 489 0 49 93 505 sie sie is 54 325 536 28 3a 539 sn 582 590 393, 593 585 67 oy I co 657 67 om 68s oss 6904 Chapter 14 m1 42 43 144 Ms “7 Radiation Flux from an Absorbing, Emiting Slab at Uniform Temperature—An Analytic Solution Problems References, Mass Transfer Definitions of Mass Flux Steady State Equimodal Counterdifusion in Gases Steady State Unidirectional Diffusion in Gases Steady-State Diffesion in Liquids Unsteady Diffusion Mass Diffusivity Mass Transfer in Laminar and Turbulent Flow Problems References Appendixes ‘A. Conversion Factors Physical Properties © Radiation Properties D__ Error Function, Root of Transcendental Equations, and Exponential Integral Function E Dimensional Data for Tubes and Stel Pipes Indexes ‘Name Index Subject Index 238 702 73 m 73 16 m9 To ™ nm 6 156 168 169PREFACE ‘The field of heat transfer is so wide and diversified that an orderly presentation of scientific facts is essential for effective teaching of this subject, In our teaching, we should place emphasis not only to the transmission of the knowledge but also to the laying of a strong foundation on which future knowledge can readily be ‘accumulated and the acquired knowledge can be fully utilized for useful purposes. This book, although based on my 1977 book entitled Basic Heat Transfer, is actually completely rewritten and reorganized, first by pedagogical considera- tions, and second by providing a large number of fully worked out illustrative ‘examples, a large number of problems at the end of each chapter, summary tables for ready reference, improved heat transfer charts and correlations, and com- prehensive physical property tables to help to increase its usefulness in practical applications. ‘The role of an introductory text on the subject isto establish the guidelines for the transmission of the knowledge and serve as the catalizer in the interaction between the teaching and learning processes. Therefore, it isnot only the knowledge contained in a book but also its organization which influences the effectiveness of teaching. These principles have been the basic guidelines in the preparation ofthis ‘undergraduate text for the teaching of heat transfer in engineering schools for the mechanical, chemical, and nuclear engineering students. ‘Our primary goal is not only to transmit the knowledge effectively but also to ‘provide a good understanding of the physical aspects of the subject matter and t0 develop the necessary skills and background for the handling of related heat transfer problems to be encountered later in the professional career. To achieve ‘such an objective, the fundamentals are systematically developed, the physical significance of the developments are emphasized, and applications to the solution of practical problems are illustrated with ample examples in each chapter. There is sufficient material in this book to meet individual course objectives at different levels, in both the junior and senior years. The spectrum of its possible ‘uses may range from a one-semester basic heat transfer course to a sequence of oy‘two-quarter or two-semester courses spread over the junior and/or senior years. ‘When used in a sequence course, conduction, finite differences, and radiation should preferably be covered first, followed by convection, boiling and condensa- tion, and heat exchangers. ‘The book can also serve as a source of ready reference for engineering graduates and industry. A background in differential equations at the sophomore level and some familiarity with fluid mechanics are all needed for following this book. Chapter 1 introduces the basic concepts and gives a bird’seye view of the ‘mechanisms of heat transfer. A discussion of units and conversion factors is also presented. Chapter 2 builds up the necessary background for the understanding of the physical significance of the heat conduction equation and its boundary con- ditions. Emphasis is placed on the development of necessary skills needed for the ‘mathematical formulation of practical heat conduction problems. This matter is iMlustrated with numerous representative examples. Chapter 3 provides an intro- duction to the solving of heat conduction problems and developing analytic ‘expressions for the determination of temperature distribution and heat flow in solids. Only the one-dimensional steady-state heat conduction problems are con- sidered. The thermal-resistance concept is introduced for use in the determination Of heat transfer through composite layers. The analysis and application of heat flow through fins are presented. In Chapter 4 the concept of transient heat flow is introduced through the use ofthe lumped system analysis because of its simplicity Todevelopabetter understanding of the significance of temperature transients with time and position, the temperature response and heat transfer charts are introduced before presenting an analysis of transient conduction. The use of these charts for predicting temperature transients in solids having shapes such as a slab, cylinder, and sphere sillustrated with numerous examples. As.a follow-up to this approach, the use of conduction shape factors is also discussed to predict steady-state heat flow in solids having complicated configurations. To give some idea of the analytic ‘methods of solving transient heat conduction problems, the method of separation of variables is considered for the solution of one-dimensional transient heat con- duction in a slab geometry. Chapter 5 presents the fundamentals of finitediffer- ‘ence methods forthe solution of both steady and time-dependent heat conduction problems. The application is illustrated with numerous examples and a computer program is given for solving the resulting finite-difference equations. Chapters 6 through 9 are devoted to heat transfer in forced and free con- vection. Chapter 6 prepares the necessary background for the understanding of the physical significance of various conceptsand fundamental definitionsassociated with the study of convection. The equations of motion and energy are introduced for the case of two-dimensional constant-property incompressible flow. The physical significance of various terms in these equations is discussed, and the use Of these equations in the formulation of forced-convection problems is illustrated with examples. Chapter 7 presents forced-convection inside ducts. To illustrate the use of the equations of motion and energy in the determination of friction factor and heat transfer coeflicients, simple forced-convection problems are solved and temperature and velocity distributions are established. Such elementaryanalysis of forced convection provides a good insight into the role of fluid flow in heat transfer. In addition, it helps toward better understanding of the physical significance of analytic and empirical correlations of friction factor and heat transfer coefficient for complicated situations. Chapter 8 deals with forced con- ‘ection over bodies. To illustrate the use of boundary layer equations in the pr diction of drag and heat transfer coefficients, the integral method of analysis ‘applied to develop analytic expressions for the drag and heat transfer coefficients for laminar flow over a flat plate. Various correlations are then presented for flow lover bodies having other geometries. In Chapter 9 the principles of free convection are discussed. An approximate boundary layer analysis is presented to illustrate the use of the energy equation to predict the heat transfer coefficient for free con- ‘vection from a vertical plate. Correlations of free convection for other configura- tions are then presented and their application is illustrated with examples. ‘Chapter 10 presents the fundamentals of boiling and condensation. Various regimes of boiling and condensation are discussed and the heat transfer correla tions associated with them are presented. ‘Chapter 11 is devoted tothe thermal analysis of heat exchangers. Various types of heat exchangers are discussed, and the use of LMTD and E-NTU methods for the sizing of heat exchangers are illustrated with representative examples. Heat transfer by radiation is covered in Chapters 12 and 13. Chapter 12 deals ‘with radiation exchange among surfaces separated by a nonparticipating medium. ‘The absorption, emission, and reflection of radiation by real surfaces are dis- ‘cussed and the concept of blackbody radiation is developed. The analysis of radiation exchange among surfaces is introduced first by using the network ‘method, because it provides a good insight into the physical nature of the problem. However, the method is not s0 practicable when the system involves more than two arbitrarily oriented surfaces. Therefore, a relatively straightforward radiosity- matrix method, capable of handling radiation problems involving any aumber of ‘surfaces with no additional complexity, is then presented. Chapter 13 considers radiation transfer inside @ semitransparent, absorbing, emitting medium, and radiation from hot gases. Finally, in Chapter 14 the analysis of mass transfer is presented with an analogy of heat transfer by diffusion and forced convection. ‘The SI (Systéme Internationale) system of units is used throughout this book. ‘Comprehensive conversion factors and physical property tables are presented in the Appendix. To provide some feel for the relative magnitude of physical properties in the SI and Btu units, some property tables are also included in both system of units ‘There are more than 170 fully worked-out examples to illustrate the applica tion of the basic theory and concepts. Over 800 problems, arranged in the same order as the material presented in the text, are included at the ends of the chapters. “with answers provided for some of the representative ones. A summary of funda ‘mental equations are tabulated at the end of each chapter for ready reference. This book is the outcome of many years of experience in teaching and writing of textbooks in heat transfer at various levels. Its hoped that it will be helpful toimprove the effectiveness of teaching and learning of the subject of heat transfer. am indebted to many of my colleagues and students for theit valuable suggestions toward achieving the objectives stated previously. I wish to thank Dr. 8. Kakag for providing useful comments on the text and Y, Cengel for thoroughly reading the manuscript. M, NECATI OzISIK\ HEAT TRANSFER | A Basic Approach‘CHAPTER ONE INTRODUCTION AND CONCEPTS. ‘The concept of energy is used in thermodynamics to specify the state ofa system. It is a well-known fact that energy is neither created nor destroyed but only changed from one form to another. The science of thermodynamics deals with the relation between heat and other forms of energy, but the science of heat transfer is concerned with the analysis of the rate of heat transfer taking place in a system. The energy transfer by heat flow cannot be measured directly, but the concept has physical meaning because it is related to the measurable quantity called temperature. It has long been established by observations that when there is temperature diference ina system, heat flows ftom the region of high temp ture to that of low temperature. Since heat flow takes place whenever there is & temperature gradient in a system, a knowledge of the temperature distribution ina system is essential in heat transfer studies. Once the temperature distribution isknown, a quantity of practical interest, the hea flux, which is the amount of heat transfer per unit area per unit time, i readily determined from the law relating the heat flux to the temperature gradient. The problem of determining temperature distribution and heat flow is of interes in many branches of science and engineering. In the design of heat ex- changers such as boilers, condensers, radiators, etc, for example, heat transfer analysis is essential for sizing such equipment. In the design of nuclear-reactor cores, a thorough heat transfer analysis of fuel elements is important for proper sizing of fuel elements to prevent burnout. In aerospace technology, the temper ‘ure distribution and heat transfer problems are crucial because of weight limita- tions and safety considerations. In heating and air conditioning applications for buildings, a proper heat transfer analysis is necessary to estimate the amount of insulation needed to prevent excessive heat losses or guns. In the studies of heat transfer, it is customary to consider thre dstinet modes ‘of heat transfer: conduction, convection, and radiation. In reality, temperature distribution in a medium is controlled by the combined effects of these three modes of heat transfer; therefore, itis not actually possible to isolate entirely ‘one mode from interactions with the other modes. However, for simplicity in2 mar TeansreR the analysis, one can consider, for example, conduction separately whenever heat transfer by convection and radiation is negligible. With this qualification, we present below a brief qualitative description of these three distinct modes of heat transfer; they are studied in great detail inthe following chapters. 1-1 CONDUCTION Conduction is the mode of heat transfer in which energy exchange takes place from the region of high temperature to that of low temperature by the kinetic motion or direct impact of molecules as inthe case of fluid at rest, and by the drift of electrons, as in the case of metals, Ina solid which is a good electric conductor, a large numberof free electrans move about in the lattice; hence materials that are ‘good electric conductors are generally good heat conductors (i.e, copper, silver, etc). ‘The empirical law of heat conduction based on experimental observations. originates: from Biot but is generally named after the French mathematical physicist Joseph Fourier [1]* who used it in his analytic theory of heat. This law states that the rate of heat flow by conduction in a given direction is pro- portional to the area normal to the direction of heat flow and to the gradient of temperature in that direction. For heat flow in the x direction, for example, the Fourier law is given as ar az | ow (ta) ya 2 ‘ ua E x W/m’ (1b) ‘where Q, is the rate of heat flow through area A in the positive x direction and quis called the heat flux in the positive x direction, The proportionality constant ‘kis called the thermal conductivity of the material and is a positive quantity. I temperature decreases inthe positive x direction, then dT dx is negative; hence 4 (oF Q,) becomes a positive quantity because of the presence of the negative sign in Eqs. (I-1a) and (1-16). Therefore, the minus sign is included in Eqs. (1-1a) and (1-1b) to ensure that q, (or Q,) is @ positive quantity when the heat flow is-in the positive x direction. Conversely, when the right-hand side of Eqs. (1-ia) and (1-15) {is negative, the heat flow is in the negative x direction. “The thermal conductivity k in Eqs. (1-1a) and (1-16) must have the dimensions ‘Wym-"C) oF J/(m-s-°C) if the equations are dimensionally correct. There 18 a wide difference in the range of thermal conductivities of various engineering * bracketed mumbers indicate references atthe end of he chapter‘eTRODUCTION AND CONCEPTS 3 i i Figure 1-4 Typicarange of thermal oo. conductivity of various material ‘materials, as illustrated in Fig, 1-1. Between gases and highly conducting metals, such as copper or silver, k varies by a factor of about 10*. Thus, in Fig. 1-1 the highest value i for highly conducting pure metals, and the lowest value is for gases and vapors, excluding the evacuated insulating systems, The nonmetallic solids and liquids have thermal conductivities that lie between them. Metallic single crystals are exceptions, which may have very high thermal conductivities; for example, with copper erystals, values of 8000 W/m -°C) and even higher are possible. ‘Thermal conductivity also varies with temperature. This variation, for some materials over certain temperature ranges, is small enough to be neglected; but for many cases the variation of k with temperature is significant. Especially at very low temperatures k varies rapidly with temperature; for example, the thermal conductivities of copper, aluminum, oF silver reach values $0 to 100 times those that occur at room temperature. Figure 1-2 illustrates how the thermal conductivity of some engineering materials varies with temperature. Actual values of thermal conductivity of various materials are given in App. B, and a compre- hensive compilation of thermal conductivities of materials can be found in Refs. 205. Example 1-1 Determine the heat lux q and the heat transfer rate across an iron platewitharea A = 0.5 m? and thickness L = 002 m[k ~ 70 Wim :°C)] when one ofits surfaces is maintained a T, = °C and the other at T, = 20°C:4 moar reansren 1000 I Sirer'99 92) | cover nary TT T soo Moire] tt I it ro potesi Aluninus ona > 00; tae, “ Mosse ee Fret Wk Qured 10S) — et = et Sori iri oho Figere 1-2 Elect of temperature on thermal conductivity of materia Sovution In this problem the temperature gradient dT/dx is constant; hence the temperature distribution T(x) through the plate is linear, as illus- Then the heat flux qis determined by applying Eq. (1-16) as aT) 20-60 > = 7078 140 kW/m “Thus the heat flow is inthe positive x direction since the result is a positive quantity. The heat flow rate Q through an area A = 0.5 mis computed by applying Eq, (1-10: Q = Aq =05 x 140 = 10kW Example 1-2 The heat flow rate through a wood board L = 2 cm thick for a temperature difference of AT = 25°C between the two surfaces is 150 W/m? Calculate the thermal conductivity of the wood.[ReTRODUCTION AND CONCERTS § Figure 1-3 Heat conduction through aab Souution Equation (1-16) is applied as follows: ar iets 25 150 = kos k = 0.12 Wim °C) 1-2 CONVECTION ‘When fiuid flows over a solid body or inside a channel while temperatures of the fluid and the solid surface are different, heat transfer between the fluid and the solid surface takes place as a consequence of the motion of fluid relative to the surface; this mechanism of heat transfer is called convection. Ifthe fluid mo- tion is artificially induced, say with a pump or a fan that forces the fluid flow over the surface, the heat transler is said to be by forced convection. If the iuid motion is set up by buoyancy effects resulting from density difference caused by tempera ture difference in the fluid, the heat transfer is said to be by free (or natural) con- ‘ection. For example, a hot plate vertically suspended in stagnant cool air causes ‘2 motion in the air layer adjacent to the plate surface because the temperature sradient in the air gives rise to a density gradient, which in tur sets up the ait ‘motion. As the temperature field in the fluid is influenced by the fluid motion, the deterinination of temperature distribution and of heat transfer in convection for most practical situations is a complicated matter. In engineering applications, to simplify the heat transfer calculations between a hot surface at T, and a cold6 wear meaNsean wit wn ht Figure 14 Heat transfer byconvection from aot nr sallat 7, toa cold us, uid flowing over it at a bulk temperature 7, as illustrated in Fig. 1-4, a heat transfer coeflicient his defined as (1-24) where q is the heat flux (in watts per square meter) from the hot wall to the cold fluid. Alternatively, for heat transfer from the hot fuid to the cold wall, Eq. (1-22) is written as a= MT; ~ TD 125) where q represents the heat flux from the hot fluid to the cold wall. Historically, the form given by Eq, (1-2a) was first used as a law of cooling as heat is removed from a body to a liquid flowing over it, and itis generally referred to as “Newton's law of cooling,” Ifthe heat flux in Eqs. (1-2a) and (1-26) is given in watts per square meter and the temperatures are in degrees Celsius (or kelvins). then the heat transfer coefficient hin Eqs. (1-2a)and (I-25) must have the dimensions W/(m? °C) if the equations are dimensionally correct. "The heat transfer coefficient h varies with the type of flow (ie, laminar or turbulent), the geometry of the body and flow passage area, the physical properties Of the fluid, the average temperature, and the position along the surface of the ‘body. It also depends on whether the mechanism of heat transfer is by forced convection (ie, the fluid motion is caused by a pump of a blower) or by natural ‘convection (.e, the fluid motion is caused by the buoyancy). When h varies with the ‘position along the surface of the body, for convenience in engineering applications, its average value h, over the surface is considered instead of its local value fh. Equations (1-2a) and (1-26) are also applicable for such cases by merely replacing A by hy; then q represents the average value of the heat flux over the region con- sidered. The heat transfer coefficient can be determined analytically for How over bodies having a simple geometry such as a flat plate or flow inside a circular tube.[RSTRODUCTION AND CONCEPTS 7 Table 1-1 Typical values of the convective heat transfer coefficient & ‘Type offlow Wile Free convecion AT = 28°C + 025m verti plat in: Aunespherie ait s | Engine ei 7 Water “0 + 002-m-OD* horizontal tinder in [Atmospheric air ‘ Engine oi a Water a + 002-m-diameter sphere in: ‘Atmospheric air 9 Engine of @ wate, 606 Forced conection ‘+ Atmospheric air at 2°C with U. = 10 mis over a tpt: Lroim » L=0sm 0 + Flow at 5 ms across -em-OD cylinder of ‘Aumosphene air as Engine oi! 1300 1+ Water at 1 kg inside 25-cm-4Dt tobe 10.00 Bolling of water at aim i + Poo! boiling in contain 3000 ! + Poo! boing at peak heat fu 350 1 Fil boiling 300 Condensation of seam o | tm «+ Film condensation on horizontal tubes 93000-25000 ‘Film condensation on vertical surfaces 4000-11000 + Dropwisecondensstion a00- 120000 + OD = outer diameter, ‘HID = inner diameter. For flow over bodies having complex configurations, the experimental approach is used to determine h. There is a wide difference in the range of the values of the ‘heat transfer coefficient for various applications. Table I-1 lists typical values of k encountered in some applications. Example 1-3 An electrically heated plate dissipates heat by convection at a zate of q = 8000 W/m? into the ambient air at T; = 25°C. Ifthe surface of the hot plate is at 7, = 125°C, calculate the heat transfer coefficient for con vection between the plate and the air.‘Souurion Heat is being transferred from the plate to the fluid, so Eq. (1-2a) is ‘applied: a=KT.~ 1) 8000 = A125 — 25) = 80 Wim? °C) Example 1-4 Hot air at T; = 150°C flows over a flat plate maintained at T, = 50°C. The forced convection heat transfer coefficient ish = 75 W/(m?-*C). Calculate the heat transfer rate into the plate through an area of A = 2m? Sovumion Forheat transfer from the hot fuid to the plate, Eq(1-2a)is applied: q= MT, ~T,) q = 75(150 ~ 50) Q = 4A = (15 x 102) = 15 kW 15 10° W/m? 1-3 RADIATION All bodies continuously emit energy because of their temperature, and the energy thus emitted is called thermal radiation. The radiation energy emitted by a body is. transmitted in the space in the form of electromagnetic waves according to Maxwell's classic electromagnetic wave theory or in the form of discrete photons according to Planck's hypothesis. Both concepts have been utilized in the in- vestigation of radiative-heat transfer, The emission or absorption of radiation energy by a body is a bulk process; that is, radiation originating from the interior of the body is emitted through the surface. Conversely, radiation incident on the surface of a body penetrates to the depths of the medium where itis attenuated. ‘When a large proportion of the incident radiation is attenuated within a very short distance from the surface, we may speak of radiation as being absorbed or emitted by the surface. For example, thermal radiation incident on a metal surface is attenuated within a distance of a few angstroms from the surface; hence metals are opaque to thermal radiation, ‘The solar radiation incident on a body of water is gradually attenuated by water as the beam penetrates to the depths of water. Similarly the solar radiation incident on a sheet of glass is partially absorbed and partially reflected, and the remaining is transmitted. Therefore, water and glass are considered semitrans- parent to the solar radiation. Tt is only in a vacuum that radiation propagates with no attenuation at all ‘Also the atmospheric air contained in a room is considered transparent to thermal radiation for all practical purposes, because the attenuation of radiation by air is insignificant unless the air layer is several kilometers thick. However, gases such as ‘carbon dioxide, carbon monoxide, water vapor, and ammonia absorb thermalradiation over certain wavelength bands; therefore they are semitransparent to thermal radiation, It is apparent from the previous discussion that a body at a temperature T. emits radiation owing to its temperature; also a body absorbs radiation incident ‘ont, Here we briefly discuss the emission and absorption of radiation by a body. Emission of Radiation ‘The maximum radiation flux emitted by a body at temperature T is given by the ‘Stefan-Boltzmann law Ey=oT* | Wim? (13) ‘where T is the absolute temperature in kelvin, «is the Stefan-Boltzmann constant [eo = 5.6697 x 10°* W/(m* -K*)], and E, is called the blackbody emissive power. Only an ideal radiator oF the so-called blackbody’ can emit radiation fux according to Eq. (1-3). The radiation flux emitted by a real body at an absolute temperature T is always less than that of the blackbody emissive power Ey; itis given by = 08, = eo a4) where the emissivity ¢ lies between zero and unity; forall real bodies its always less than unity. Figure 1-5 shows a plot of the blackbody emissive power E, defined by Eq. (1-3) versus the absolute temperature. The radiation flux emitted rapidly increases with rising temperature. For example, the emissive power increases from E, = 461 W/m? at room temperature T = 300K to E, = 3562 W/m? at T = 500 K and E, = 56,100 W/m? at T = 1000 K. Absorption of Radiation Ifa radiation flux gia is incident on a blackbody. it is completely absorbed by the blackbody. However, if the radiation flux qigc is incident on a real body, then the energy absorbed duu by the body is given by ane = ine as) where the absorptivity 2 lies between zero and unity; for all real bodies it is always less than unity. ‘The absorptivity «ofa body is generally different from itsemissivity«. However, in many practical applications, to simplify the analysis, «is assumed to equal « Such matters are discussed in depth and actual values of emissivity of various surfaces given in Chap. 12,10 wear rmanseen ‘Buackbodyemine power EC?) 102 § 10 5 4 2 vol to? Se Soret nk Fire 15 Blackbody emisve power Radiation Exchange When two bodies at diferent temperatures “see” each other, heat is exchanged between them by radiation If the intervening medium is filed witha substance such as air which is transparent to radiation, the radiation emitted from one body travels through the intervening medium with no atenuation and reaches the other body, and vie versa. Then the hot body experiences a net heat loss and the cold body’a net heat gain, asa result ofthe radiation heat exchange. The analysis of radiation heat exchange among surfaces is generally a complicated matter and is dealt within Chap. 12. Here we examine some very special cases with illustrative examples. igre -6 shows small, hot, opaque plate of surface area A, and emissivity e, thatis maintained atan absolte temperature T, and exposed toa large surrounding area A; (ity Ai/Ay +0) at an absolute temperature 7. The space between them contains ait which is transparent to thermal radiation, The radiation energy emitted by the surface A, is given by ‘AwaTtFigure 16 Radiation exchange between surface Ayand its sutroudings, ‘The large surrounding area can be approximated as a blackbody in relation to the small surface Ay. Then the radiation flux emitted by the surrounding area is o7$ which is also the radiation flux incident on the surface 4. Hence, the radiation energy absorbed by the surface A, is AyayoTy ‘The net radiation loss atthe surface A, is the difference between the energy emitted and the energy absorbed: Qs = AyoT} - AyayoT$ (1-64) 4, this result simplifies to Aio(T$ ~ TD, (1-66) which provides the expression for calculating the radiation heat exchange between ‘small surface element A, and its surroundings at T;, Clearly. the positive value of implies heat loss from the surface A, and the negative value implies heat gain, We now consider two finite surfaces 4, and A; as illustrated in Fig. 1-7. The surfaces are maintained at absolute temperatures 7, and T,, respectively, and have ‘emissivitiesc, and ¢,. The physical situation implies that part ofthe radiation leav- ing surface A, reaches surface 4, while the remaining is lost to the surroundings. ‘Similar considerations apply for the radiation leaving surface A, The analysis of radiation heat exchange between the two surfaces for such a case should include Figure 1-7 Raciaton xchange between surfaces 4, and Athe effects ofthe orientation of the surfaces, the contribution of radiation from the surroundings, and the reflection of radiation at the surfaces. For the arrangement shown in Fig. 1-7, if we assume that tne radiation flux from the surroundings is negligible compared to those from surfaces A, and A;, then the net radiation heat transfer Q, at the surface A, can be expressed in the form Q, = FrAo(T$ - TD an where F, isa factor that includes the effects of the orientation of the surfaces and their emissvitis. The determination of this factor is a complicated matter, and the analysis of radiation problems of this type is the subject of Chap. 12 Radiation Heat Transfer Coefficient ‘To simplify the heat transfer calculations it may be possible, under very restrictive conditions, o define a radiation heat transfer coefficient h,, analogous to the con- veetion heat transfer coefficient, as = MT - (a8) “This concept can be applied to the result given by Eq, (I-6b) as now described. ‘Equation (1-66) is written as Q = Ayyo(T} + TMT + BHT ~ 1) (1-90) If|T, — T;| < T,, this result is linearized as Ayo4T X(T, — Ts) (1-96) (0-10) ‘A comparison of Eqs. (1-8) and (1-10) reveals that for the specific ease given by Eq, (1-66), a radiation heat transfer coefficient h, can be defined as b= 4Tigo aan Example 1-5 A heated plate of D = 0.2.m diameter has one of its surfaces insulated, and the other is maintained at T, = 580 K. Ifthe hot surface has an ‘emissivity ¢, = 09 and is exposed (0 a surrounding area at 7, = 300 K with ‘atmospheric air being the intervening medium, calculate the heat loss 6) radiation from the hot plate to the surroundings.en SOLUTION Assuming e, = a,, we can apply Eq. (1-66): Qy = Avt.o(TS ~ T?) [f02°osxser«0-y055" 99 1 onus Dos A son hot st tape = Kt ‘T, = 400 K. If this radiation transfer process is characterized by a radiation heat transfer coefficient h, calculate the value off, SouvTION For this particular case, the requirement T, — T; < T; is satisfied. ‘Then Eq. (1-11) is applied as follows: h, = 4T}ey0 [(4.3)? x 10°}(0.8X5.67 x 10 144 Wim? -°C) 1-4 COMBINED HEAT TRANSFER MECHANISM So far we have considered the heat transfer mechanism, conduction, convection, and radiation separately. In many practical situations heat transfer from a surface takes place simultaneously by convection to the ambient air and by radiation to the surroundings. Figure 1-8 illustrates a small plate of area A and emissivity ¢ that is ‘maintained at T, and exchanges energy by convection with a fluid at T,, with a hheat transfer coefficient h, and by radiation with the surroundings at T;. The heat Joss per unit area of the plate, by the combined mechanism of convection and radiation, is given by Ge = AAT, ~ Te) + 0o(TS ~ TY (112) wh Convection rounding Figure 1-8 Simultaneous convection and radiation fom a plate14 Hear massren If|T, ~ T,] < Ts the second term can be linearized. We obtain (Te — Te) + he 7) (130) where b= deo 138) Example 1-7 A small, thin metal plate of area A m? is kept insulated on one side and exposed to the sun on the other side. The plate absorbs solar energy at a rate of 500 W/m? and dissipates it by convection into the ambient air at T, = 300 K with a convection heat transfer coefficient h, = 20 W/m? -°C) and by radiation into a surrounding area which may be assumed to be a blackbody at T,, = 280 K. The emissivity of the surface is ¢ = 0.9. Determine the equilibrium temperature of the plate. SOLUTION The energy balance per unit area ofthe exposed surface is written as sox =a, 30) +09 307 « (3) — ont ay m0 0a) — se 1. mass -008(%) Tpeslion fh etuain y iad ors he patents asusK 1-5 UNITS, DIMENSIONS, AND CONVERSION FACTORS In the field of heat transfer, the physical quantities such as specific heat, thermal conductivity, heat transfer coefficient, heat flux, etc. are expressed in terms of 1 few fundamental dimensions which include length, time, mass, and temperature, and each of these dimensions is associated with a unit when it is to be expressed ‘numerically. For example, length is the dimension of a distance, and to express it numerically one may use units of feet or meters or centimeters, ete. Time may be measured in hours or seconds, mass in pounds or kilograms, temperature in degrees Fahrenheit or Celsius, energy in British thermal units or joules, and so on. ‘When the dimensions of a physical quantity are to be expressed numerically, a consistent system of units is generally preferred. In engineering the two most ‘commonly used systems of units are the International System of units (SI) and the English engineering system. The basic units for length, mass, time, and tempera- ture for each system are listed in Table 1-2. Here the symbol “Ibt™ is used forIReTRODUCTION AND CONCERTS 15 Table 1-2 Systems of units English engineering Quantity stem Lent = ®t Mass ks ® Time : ‘ Temperature K R Foree N we Enerey JorNsm Btw or f-Ibe ound-force to distinguish it from the symbol “Ib” commonly used for pound-mass, but there is no such misunderstanding in SI because the kilogram is the unit of ‘mass and the newton is the unit of force. The physical significance of the force units newton and Ibf is better envisioned by considering Newton's second law of ‘motion, written as Foree = mass « acleration (14) where 4, is the gravitational conversionfactor constant. The pound fore It is defined asthe fore that acts onthe mass of one pound at‘ point on the earth Where the magnitude ofthe gravitational aceleration is g = 32.174 fs. Then, in the English engineering system, Eq (1-4) bocomes tbr = 4 1b x 32176 ys? (ts) According to this relation, one pound of force (that is, 11h) will accelerate one pound of mass (that is, 1 1b) 32.174 ft/s*; or 1 Ibf is equal to 32.174 ft - Ib/s*. The Conversion factor gin the English engineering system is obtained from this r- lation as 9¢ = 32.174 Ib ft/(bf-s2) (1-16) Note that the gravitational acceleration g and the gravitational conversion factor 4 are not similar quantities, isconstant, but g depends on the locaton and on the atitode In Si, Eq, (1-14) becomes IN = 4x kg x tis? 417) Clearly, in SI, 1 is a force that will accelerate a t-kg mass 1 m/s?, or a 1-N force is equal to 1 kg -my/s?. The conversion factor g, in SI becomes a= Lkg-m/N-s?) =1 (18)16 wear TRANsrER since IN = kg mis? 19) ‘Therefore, in SI, g, is not needed. Eneray is measured in Bru or ft-lb in the English engineering system whereas {tis measured in joules (J) oF newion-meters (N-m) in SL. Note that 1J = 1N-m and 1J =U kg-m?js* since 1 N= 1 kg: m/s?, Power is measured in Btujh or ft-lbs in the English engineering system and in waits (W) oF kilowatts (KW) or Jjs in SI. Note that KW = 1000W and 1 W=1J/s =1N-mjs kg-m?/s* (120) Pressure is measured in lbfjin? in the English engineering system and in bars or Nim? in SI. Note that 1 bar = 10° N/m? = 10° kg/m: s*) and Satm 98066 bar In SI, when the size of units becomes too large or too small, multiples in powers, of 10 are formed with certain prefixes. The important ones are listed in Table 1-3. For example, 1000 W = 1 KW (kilowatt) 1,000,000 W = 1 MW (megawatt) 1,000,000 N = 1 MN (meganewton) 1000 m = 1 km (kilometer) 10°? m = 1 em (centimeter) and so forth. ‘A comprehensive table of conversion factors useful in heat transfer calculations is presented in App. A. Example 1-8 Convert the heat transfer coefficient h = 20 Btu/(h-ft? °F) to the units of J,(s-m? °C) or Wim? -°C), Table 1-3 Prefixes for multiplying factors 10" =pico(p) 10. = deka) 10°? =pano(a) 10% = hectoc) io (2) 10" = kilo) mill (m) 10" = mega (M) centi(e) 10 = gign ded) 10 ‘era T)INTRODUCTION AND CONCERTS 17 Souunion From Table Al. #9. App. A we have I Btu/(h-f-°F) = 5.677 W/m? -°C), which is written as Wim? -°C) Brun Then the conversion is performed as 5677 Wm? -°O) Bru - hm 20 Blu/(h 2 -F) (Bou i0-0*-Py[ sor © 1136 Wim? -°C) = 113.6 J)(s-m? °C) since IWets 1-6 SUMMARY OF BASIC RELATIONS We summarize in Table 1-4 the basic relations given in this chapter. ‘Table 1-4 Summary of basic relations Equation ‘number Relation Remarks Conduetion: ir ge eT wnt Conduction hea ux inthe x direction Convection: (126) g= T=) Wit ‘Convection beat fu rom the wall surface 1 the aid Radiation 3) eer wim Blackbody emisive power ay a= oT! Wim Radiation fx emitted by areal boxy (1-66) = AocTt— TW Net rai ge medium 4, aT, Joss rom surface Ay at T; toa very for4t 0 (re) ae BATT.) Wm? nergy tranfr by rantion for JT, ~ TT, where h, = dT? (413) gee TT) Eneray transfer by combined convection ad +h. ~T) Wim? radiation for To Te T, where h, = de?18 nearreansren PROBLEMS Conduction 14 A temperature diference of IO°C applied across a fecay brick 10 cm thik with thermal ‘onducty 10 Wim °C) Determine the hea transer rate per square meter are answer: Wn? or Ke) 1.2 A temperature diference of 100°C is applied across a corkboard $m thick with thermal eon Tietag Ube Wien “C), Determine th het transfer rate acrosta bn? area per hour 1.3 A ler hss nclating boetdof thermal conduciity 005 W)(n-“C) isto Be used to imi the Hest one! to 80'W)i? fora temperature diferenoeof 160°C across the board Determine the thickness of ‘the inslaing board. “answer 01m 144 Glas woot of thermal conductivity 0038 W/m °C)isto be wed to inslate an ce box. he ma TW Tou should not enced 45 Wim for’ temperature of 40°C aon the wall of thes bon ‘Getermine the tickbess of he insulation, “Anower: 34.00 15 A brick wall 15 om thick with thermal conductivity 1.2 Wymn-°C) 8 maintained 43°C atone Fee S'c atthe aber face Determine the eat anf rate across the 4m? surface rea of the wall. Amover: 64 1-6 two large plats onc a 80°C andthe oer at 200°C are 8 em apart. I the space betwee them ‘hued by tvnay park rock woo! of thermal conductivity 008 Wifon °C) caleulte the heat transfer ‘rate across the plates pr lem? ae {L7 The heat flow rite across an insulting materi of thickness 3 cm with thermal condctvity 01 ‘Wotan 20) 250 Wr If the hot sue temperature is 175°C, mba is the temperature ofthe cold surface? “Anomer: 100°C 1.8 A -cmthick concrete wal has a srice aa of 40 m?. The jane surface ofthe wal iat 20°; ws fhe outer erase at "10°C. Determine the rate of eat lose through the wal ifthe thermal Conductivity 78 Witm-°C) Mdnawer! LOEW oF fs 1.9 The het flow ate tough a emis wood board fora temperature diferenceof 25°C between the inner outer surfaces 875 Worn?. What the thermal conductivity ofthe wood? vAnomers 012 Wil °C) t.t0 The inde and outsige srace temperatures of window glass ae 20 and — 12°C, respectively Tithe Bases BO cm by 4 ei 16cm tk. and has thermal conductivity 0.78 Wien °C) determine the heat los through the las ver 3h Anoner: S391 KS 1-11 Two plates, one at «uniform temperature of 300°C and the other at 10°C, ae separated by 3 {eck asbestonsement board of thermal condociity 070 W)(m-“C), Determine the rate of eat tramfr across the layer pet Im! srtce. tnt? By conduction 2000 W is transferred through 405m? section ofa -em-tick insulating materi ‘Burin the temperate diference actors the insulating layer if the thermal conductivity #8 0.2 Wa"0). ‘Anse: 800°C Coarction 1413 Watecat mean temperate of 20°C Rows over at pate at 80°: Ifthe heat transfer coefcent {ato Wyte, determine the bea transer per square meter ofthe plate over Sh “Ansner: 216 MIINTRODUCTION AND CONCEPTS 19 1-44 large surface st SO‘Cis exposed to air at 20°C the heat transfer coeficent between the surface and the ars 15 Wim -°C) determine the heat tanfrred rom 3 mo the surge over 7 148 Airat 190°C flows over a at plate whichis maintained at 50°C: Ifthe het transfer coefkent for forced convection is 300 Wi(m? "°C, determine the hea anser tthe plate though 2 one Wh Anower: 48 MI 146 A 25.cm-diameter sphere at 120°C is suspended in at at °C Ifthe natural convection heat ‘transfer between the sphere and tears 15 Wjtm?-C) determine the rate of het lows fom the sphere, Answer: 2945 W 4447 A fd at 10°C Rows over a2.5-cm-OD and 2-4 long tube whose surfaces maintained a 10. the hea transfer coetcient between the tube andthe sir is 300 Wim? C), determine the rte af heat transfer rom the abet the ai 1-48 Pressried water at SO'Cfows inside a S.c-diamete, Im Jong tube wih surface temperature Iaintained at 130°C: Ifthe heat transfer coeticient between the water ard the tube he 30) Wil °C), determine the beat aster rate from the tbe othe water, ‘Anoner:2513KW 1-19 Heats supplied toa plate fom its back surface a a rate of $00 Win ands removed from its front surface by ait ow at 20°. Ihe beat transfer cocffcent between the ar and te plat race a ‘r= 20 Wm"), what i the emperatue ofthe front surface ofthe plate? 1-20 The inside surface ofan insulating layer is at 20°C and the outside surface is disipatig beat by ‘onvection into air at 20°C. The insulation layer i 4 em thick and has thermal conductivity 12 ‘Woe-°C), What i the minimum value ofthe heat tense ceficent atthe ovside sua the ‘outs surtice temperature should not exceed 10°C? ‘Answer: 120 Wim “C). 121 A tO-cniameter sphere is heated internally wih s100-W clectic heater The sphere dissipates ‘eat by convection fom is outer surface to the ambient ai Cakalate the heat tramsor coeticiet for convection fom the sphere if the temperature difeence between the sphere surface andthe abit rie sO: 1.22 A thin meal plate is insulated a the back surface and is exposed othe sun at the ron surface. ‘The front surfoce absorbs the ol radiation ef 10 W/m and dissptesit mainly by convection tthe ambient air at 25°C the heat trnser coefficient between the plate nd the irs 13 Wim’ °C) what is the temperature ofthe plate? Answer: 85°C 1.25 A thin meal plate0.1 m by 0.1 mis placed ina large evacuated conainet whose wal are Kept at 300 K. The bottom surface of he plat insulated, and the top surface is mamtined at SOD Kas ‘esut of electric heating. I the emissivity ofthe surface ofthe plate ise = O8 whats the rate of heat ‘xchange between the plate and the walls ofthe container? Take = 387 10"® Witar KSy “anouer 24) W 424 Two larg paratel platesone at 2 uniform temperature SOO K andthe other at 1000 K, ae separ sted byw nonpartcipating gas Assuming thatthe surfaces ofthe plates are prec emitrs ad that th ‘convection i negligible, determine the eate of heat exchange between the srfaces pot square met 1428 A spre 10cm in diameter is suspended inside a large evacuted chamber whose walls are kept at 300K. If the surtace ofthe sphere as emissivity = OM und Is maintained at 300 K determine the "ate of heat los trom the sphere othe walls ofthe chamber. ‘Ansmer: 77.52 W 1-36 Two very larg, petty back parallel plates. one maiatined at 120 K and the other at 60 K, exchange heat by radiation (1, convection is negipible. Determine the heat transfer rate per 1m? Surface20 ear mmansren 127 One surface of hin plate is exposed a uniflorm heat ux of $00 W/m andthe ober side dis. ‘iptes heat by radiation toa environment at ~ 10°C. Determine the temperature ofthe pits ASRumE blackbody conditions fr eaiation Aner Ty = 313K ai A thin metal shet separates two large parallel plates, on at a uniform temperature of 100 K tin the other a 00 K. Blackbody conditions can be asumat for al surfaces, and heat transfer can be “hsumed tobe by redinton only, Determine the temperature ofthe separating shot. ‘Combined beat trmeter mechanism 129 A fat plate has one surface insulated andthe othe surface exposed othe sun. The exposed surface beets the solar radiation at ate of 800 Wm? and spats it by both convection and radiation {te ambient aia 300 KI the emissivity ofthe surface is = 09 and the convertion heat transfer often betwen he plate and ai 12 Win? °C). determine the temperature ofthe plate, ‘Anomer! 23 K 1-30 A thin plat is expose to an infrared radiation flux of 100 W ® on one surface while he other farface it Lop inslated. The exposed surface absorbs 90 percent of the incident radiation fx and ‘Tepe by comestion and radiation ito the amen ar at 300 KIC the heat ansereoefcent {Re cbmvccion heewcen the surface andthe ambien a 15 W/(a?-°C) determine te temperature of the plate Take the emisvity of pat as ¢ ~ 08. FBT A tia plate 30cm by 50cm is subjected 1 400 W of heating on on surface and dissipates the teat by combined convection and radiation from th other surface into the ambient ait at 290 K. IF the sac of he pate hasan emissivity e-~ 09 and the beat transfer ceficent between the surface td the ambien arin 15 Wi(m “Cp, determine the temperature of the plats “Anomer: 3622 K 1.32 The volar radiation cident onthe tse surface of an aluminum shading vie is 1000 W/m ‘Atainum absorb 12 percent oF the incident oar energy and dispatesit by convection rom the back ‘Aitae and by combined convection and radiation fyom the outside surface The emissivity ofthe “Tuminom 10 the convection heat tamer coefficient i 15 Wim: "C) for both surface, and ae abbleat temperture ean be taken 20°C for both convection and radiation. Determine the temperature ofthe shade, 1133 One surface of a thin metal sheet reeves radiation from a lage plate at 7O0C, while the otber Satay desputes eat by convection toa colant id at APC. The surfaces can be considered as 1 pafet absorber and a perfect emiter for radiation. The heat transiercoefiint for convection ‘ween the surface and the fd i 120 Wim! °C). Deterne the temperature ofthe plate anower: 638 K 1-4 Heat is ot by both conection and radiation from a 2-nv4ong uninsulted portion of S 0. 224 A solid bar of rads r = bis ial ata wniform temperature T. Fortes > ts cooled by convection Irom its surface str ~ B into the ambient at at temperature with a heat transfer nefceat h- Writs the mathemati formulation ofthis eat conduction problem forthe determina- {on of one-dimensional, time-dependent temperature dsribuon Tr, within the rod. 1225 A plane wall of thickness Lis exposed toa uniform beat fx of gp Wim? on one side and disi- pats hen by conection with eat transfer coefiiet he ato the ambient air temperature Ton the Diet side. Wine the mathematical formulation ofthis problem forthe determination of one-dimensional, Steady temperature disrbuion T(x) with the wall. 226 A thick-walled rela tube has inside rade r, and outside adi. Aol gas at temperature 7; flo inside the tbe, anda cold asa lempratare T;Nows outside The thermal conductivity k of the tube is considered constant, The Bet tufrcoaicient for fow inside and outside the tobe ar ‘Specited a and hy sespectively, Write the mathematical formulation fthisheat conduction probes im order to determine the one-dimerstonal,steadsare temperature distribution T() trough the tube wall 2.27 Consider a soli cinder of ads r ~ and the height = ~ H. Heats generated in the soldat fate of gy Win’ The boundary surface at: = O is Kept insulated: the boundary surface at == # ‘isipates heat by convection into a medium a temperature with heat transfer coefficient Te tylinrcal boundary surface atr = bis maintained at» ifm temperature T,. Write the mathe ‘atc formlation ofthis prob for the determination af wo-dimesional stead)state temperature ‘rbution Te, 2) within the eins. 2-28 Consider the two-dimensional, steady-state heat conduction problem fr a rectangular region 0s x = 4,0 << blorthe following boundary conditions: a) The boundary surface at x ~ Os electrically ated at a at of gy Wim? (by The boundary surfece atx = ois kept at constant temperature Ty, {e) The boundary surface at y= Oi insulated (a) The boundary surface aty ~ b spats heat by convection into a medium a temperature 7. with eat transfer eoefeest{CONDUCTION BASIC EQUATIONS 4 ‘The thermal conductivity ofthe solid is constant and there i 0 heat generation i the medio, Write the mathematical formulation ofthis problem forthe determination of twmenion, steady stare temperature distribution Tx 3) wih the teion 229 A copper bar of adi bi initially ta uniform temperture Ty The heating of the rod begs A time ¢ = 0 by the pasage of electri current which generates beat throughout the rod a a eonsant rate of gg W/m. The rod disipates heat by convection rom is surface tr ~b, witha beat ansfer ‘oeficient into the asbient a at temperature T,. Assuming thatthe thermal conduct kof the ‘ods constant and thatthe problem canbe rete a one-dimensional ransen heat conduction ia he "variable, write the mathematical formulation forthe determination of the one-dimensional tine ‘dependent temperate ditbution Te, 1} within the sod fort > 0 1230 Consider a holow sphere of inside adi r~ a and outside radi r= 5, The inside surface is tniformly heated electrical ata ato g W/m? andthe outside surface dasipats heat by convection with heat rander coeticient hint an ambien ta constant temperate T. Wht the mathematical {ormulatin a this heat conduction problem forthe determination othe oneimensiona stealy-sate temperature dsebution T() i the sphere 231 Consider long te ofinside radius r= cand outside radissy = b Heats pneratedin the tube ata constant tate of ge W/m? by the passage of elec current. The inside surtace kept insulted, and the outside surace i disiating heat by convection into an ambient at temperature T. with heat Uanser coefficient h. Wee the mathemattca formation a ths heat sondacion problem fr th determination ofthe one-dimensional. steadysate temperature dibution TU) the soli REFERENCES 1, Schneider, P.J.: Conduction Heat Transfer, Addison. Wesley, Reading, Mae, 1955 2 Carsiaw, HS, and J.C Jaeger: Condition of Heat Slide, 24 ed, Oxford University res, London, 1959, 3. Gnik, M.N.: Boundary Value Proboms of Heat Conduction, Inerational Textbook, Scranton, Pa, 1868 4. Myers, GM: Anayical Methods in Conduction Hoot Transfer, McGraw-Hill, New York, 1971 5. Ont, M.N.: Hear Conduction, Wiley, New York, 1980‘CHAPTER THREE ONE-DIMENSIONAL, STEADY-STATE HEAT CONDUCTION In this chapter we present the application of one-dimensional, steady-state heat conduction problems to determine temperature distribution and heat flow in solids having geometries inthe form of a slab (i. plane wall, cylinder, and sphere. ‘The one-dimensional, steady-state condition implies thatthe temperature gradients exist along one coordinate direction only and that temperature within the solid does not vary with time. Situations involving different types of boundary con- ditions, with and without internal energy generation within the solid, are examined. ‘The concept of thermal resistance, analogous to electric resistance, is introduced to solve one-dimensional, steady-state heat conduction with no internal energy generation. The thermal resistance concept is also utilized to determine one~ dimensional, steady-state heat flow through layers of slabs, cylinders, or spheres. Heat flow problems involving the determination of critical thickness of insulation and the use of extended surfaces (ie. fins) are examined, 3:1 THE SLAB Consider a slab (ie. plane wal) of thickness L as illustrated in Fig. 3-1. The plate is sufficiently large in the y and z directions in comparison to its thickness L to ‘ensure that the temperature gradients in the y and z directions are negligible ‘compared with that in the x direction. The temperature within the solid does not ‘vary with time. Then the temperature distribution T(x) within the solid is governed by the one-dimensional, steady-state heat conduction equation. For the ease of constant thermal conductivity k and with an energy generation at a rate of g(x) ‘W/m?, the heat conduction equation is given by (Ti) a ange SE + ta =0 6+)‘ONE-DIMENSIONAL, STEADY-STATE HEAT CONDUCTION 43 rs Ono woe 7, * Figure 31 One-dimensional steady-state et flow through a Sin slab nd the equivalent thermal revstance concep which is valid over the domain of the slab confined to the region 0 7, the right-hand side of this expression is positive, hence the hheat flow isin the positive x direction, ‘The heat flow rate through an area A ofthe slab normal tothe direction of heat flow is On ag= ae BB w This result is now rearranged inthe form G50) where awd 5b) Here R is called the thermal resistance of the slab for heat flow through an area A across a temperature potential T; — T;. The concept is analogous to ‘electric resistance in Ohm's law and is illustrated in Fig. 3+. Example 3-2 Consider a slab of thickness L as illustrated in Fig. 3-2. A fluid at a temperature 7, with a heat transfer coefficient h, flows over the surface atx = 0,and another fluid ata temperature Tz witha heat transfer coefficient ‘hz flows over the surface at x = L of the plate. Develop an expression for the heat flow Q through an area 4 of the plate. Calculate the heat transfer rate through A= I m? ofthe slab for Ta, 130°C, Tay = 30°C, hey = 250 Wim? -"C), haz = 500 Wim? °O), L = 4om, k = 20 Wi(m °C), SOLUTION Since there is no energy generation in the medium and only the heat flow rate through the plate is required, it is more convenient to utilize Oo co ote g Figure 2 Thermal resistence concept for v heat Row through hie fc v ‘with convection ax ia im, at both srftces46 near mansren the thermal resistance concept for the solution of this problem. Refer to the ‘nomenclature shown in Fig. 3-2. The heat flow Q is by convection from fluid 1 to the surface of the plate at x = 0, by conduction through the plate, and by convection from the surface at x = L to fluid 2. With this consideration we write h-% Q= Ah To — T) = Ahg(Ts ~ Tas) ‘This result is rearranged in the form =T_T-h_B-Trs ai) ~ LAR) ~ “17CAh) This is analogous to Ohm's law, with each term in the denominater representing the thermal resistance to heat flow ofthat particular layer. By adding the numerators and the denominators of this equality, we obtain e= Tes = Tor 1h) + LAR + TAR) “The het transfer rte Q through the area ofthe sab can be eaeuatd from thi expresion, since all the quantities nthe Hight-hand side are know, ‘This esl can be writen more compact inthe orn Teo, ~ Taz Re hte the total thermal resistance heat flow Ris defined as tyby a Ah,” Ak he Ah, Here (Ah, ithe thermal resistance for convection through fd 1, L(42) isthe thermal resstance for conduction through the lab, and /(4h) i the thermal esstancs fo convection hough id 2 ‘The numerical part of hs examples computed as Tate toon Tx 2507 1x 20 Q Q Ru 8x 107 Example 3:3 Consideraslab of thickness Land constant thermal conductivity kin which energy is generated at a constant rate of gp W/m. The boundary surface at x = O is insulated (adiabatic) and that atx = L dissipates heat by convection with @ heat transfer coefiienth into a fluid ata temperature T,(ONE-DIMENSIONAL, STEADY-STATE HEAT CONDUCTION 47 Develop expressions for the temperature T(x) and the heat flux q(x) in the slab, Calculate the temperatures at the surfaces following conditions: L = tem, k= 20W/m = 4000 W/(m? °C), and T,, = 100°C. O and x = L under the . go = 8 x 10" W/m’, Souution There is energy generation in the medium, hence the thermal resistance concept cannot be utilized for this problem. Therefore, the heat ‘conduction equation should be solved to determine the temperature distri- bution. "The mathematical formulation of this heat conduction problem is given » £T 009 ocxe 279,09 nocect ax=0 ‘The first integration of the differential equation gives aT) 7. ‘and the application of the boundary condition at x = 0 yields c=0 A second integration with C, = 0 results in =-%24¢, Te) = - S238 4 Ce and the application of the boundary condition at x = L. gives gol’, o.\_ or ‘Then the temperature distribution in the slab becomes 19 821 -(9) eee48 ear mansion ‘The physical significance of each term in this solution is as follows: The first term on the right-hand side is due to the energy generation in the solid ‘The second term is due to the presence ofa finite heat transfer coefficient at the surface. For hh» co, this term vanishes, and the boundary surface at x = L. is at the temperature T,, ‘The expression for the heat flux anywhere in the medium is determined from its definition as aro dx a) 90x Finally, by using the numerical values given above, the temperatures at the boundary surfaces x = O and x = L are, respectively, 8 = 107 x (01)? | 8 x 107 x 001 2x0 4000 8 x 10” x 001 4000 TO + 100 = 500°C 4+ 100 = 300°C Example 3-4 An iron plate of thickness 1. with thermal conductivity k is subjected to a constant, uniform heat flux qq W/m? at the boundary surface at x = 0, From the other boundary surface at x = L., heat is dissipated by convection into a fluid at temperature T,, with a heat transfer coefficient h Figure 3-3 shows the geometry and the nomenclature Develop expressions for the determination of the surface temperatures T, and T; at the surfaces x = O and x = L, respectively. Caleulate the surface temperatures T, and T; for L = 2em, k= 20 Wom -C), qo = 10° W/m?, T,, = 50°C, and h = $00 Wi(ta? -°C), SOLUTION Since there is no energy generation in the medium and only the temperatures of the boundary surfaces are required, itis more convenient to utilize the thermal resistance concept as illustrated in Fig. 3-3 to solve this problem, igre 3:3 Nomenclature and thermal resistence com cep for Example 34“ONE-DIMENSIONAL, STEADY-STATE HEAT.CONDUCTION 49) ‘Using the nomenclature in this figure, we immediately apply the thermal resistance concept: ete cea) Ts ~__N\- Te __ M40 = Lak) ~1itahy ~ Tak) + (AR) Here the last expression is obtained by adding the numerators and the de- ‘nominators of the second and third expressions. Clearly, in these equations the area 4 cancels, and the surface temperatures T, and T; become the two ‘unknowns. By equating the fist and the last expression, Tis found: Lt i= (beter and by equating the first and the third expressions, T; is found: 4% n-# na P+ To For the numerical part of this example, the temperatures T, and T; are calculated by introducing the numerical values of various quantities in the above results. We find ee (+g) +o-re W | gC 3.2 THE CYLINDER -The problems of one-dimensional, radial heat flow in solids having a geometry in the form of a cylinder have numerous engineering applications. Heat removal fom a cylindrical fuel element of a nuclear reactor by the coolant fluid, heat Gieeipation from a current-carrying wire, and heat flow across # thick-walled circular tube are typical examples. ‘Suppose there is an energy generation in the solid ata rate of ar) Wim and the thermal conductivity is considered constant. The temperature distribution Torin the solid is governed by the following heat conduction equation: 1a (é ral) 1 70 ooWhich is valid over the cylindrical region. Once the temperature distributic in the cylinder is established from the solution of this equation, the radia, flux q(r) anywhere in the solid is determined from the definition aT) Wim o a) Todcterminhetempereditcbaon Tin theregon (sine aultterutngnerton contac dceranedhon he eo bound sod ere seb Tole eel rss, we cose By (46 for sonst nergy mneatongadwie cae (2) =a, 28 a(Z)--% ¢ ete y= conan The fat and tesco inteaton of Fa (4) ge omy, a7), ar RTT G9) bo T(r) ae tCiinr +c, (3-10) Carly, two boundary conditions are needed to determine the two integration $Ertants Ci and C,. In the case of a hollow cylinder illustrated in Fig. 3-4, the boundary conditions at the inner and outer surfaces can be a prescribed Figure 3-4 Coordinate simensiona beat con ‘inder (a) Sabi « hollow cylinder,ONE-DIMENSIONAL, STEADY STATE HEAT CONDUCTION $1 insulteds ag 20,27 €0 e8ape from the medium because the boundaries are insulated sa result the problem has no steady-state eeloign £5 he case ofa solid cylinder, lustrated in Fig dene boundary condition an be specified for the outer surface, but another boundary condition is needed siuen. center of the cylinder. Such a boundary condition ig specified from the Physical consideration of the temperature distribution tong cylinder. Physically condiven lution fequies thatthe temperature not be info ae 0; this Condition is satisfied if Te) = finite atr=o @-tta) onedimensional, radial heat flow ina soli cylinder, the temperature is sym- boundary att the center of the cylinder. With this considersue alternative boundary condition at r = atr=0 G11) Example 3-5 Consider a sold cylinder of radius r= b in which energy is Sanat @coastant rate of gy W/m’, while the boundary sense ee 6 ‘8 maintained at a constant temperature 7, Revelop an expression for the one-dimensional, radial steady-state Kemperature distribution T(r) and the heat ux g(r), Calculate the center temperature T(0) and the heat fx atthe boundary Toone” ~ PFOrb = Lem, 95 = 2 x 10° Wim? = 90 Wes *Oyand 7 = 100°C, SOLUTION The mathematical formulation of this problem i siven as raparoys : Pt 0 mocres 0 . M9 a TO=T, arab its and the second integrations of this diferential equation give, respec- tively, are) r+ erates EP tC inr +c, TT]!2 meat ANSE “The application ofthe boundary condition atr = Oo the Bist equation gives c.=0 “The same result also could be obtained by the application of the alternative The sare boundary condition given by T() = faite at r= 7 the second Ghution. The constant Cis determined by the appicah of the boundary eauittgon at 7 = bt the second of the above equations, We find gob Cr +h -Then the temperature distribution in the cylinder becomes ro -G)]es and the heat ux (7) anywhere in theasedium is determined from definition as aT) _ 90r ao kG “The center temperature 7(0) is caleulated from the above exDre=ion for TO) by setting r = 0: 2x 10° x OO? , 199 = ot GOW 4-100 = 350°C -thesurace heat floxiscaleulated from the expression for 4) By SinBT = b: gob _ 2.x 108 x 001 Oy r TO) 108 Wim? Example 36 The inner surface at r = a and the outer surface at r= bof t ea der are maintained at uniformtemperatures 7 and 7, respectively. ‘The thermal conductivity k ofthe solid is constant Develop an expression forthe one-dimensional, steady-state temperature distribution T() in the cylinder. Develop an expression forthe radial heat how rate Q through the cylinder over a length H. ene an expression forthe thermal resistance of hollow cylinder of length H. SSouvTi0n The mathematical formulation ofthis problem is giv=n 2 a a0 s[p]-0 eeeretONE-DIMENSIONAL, STEADY-STATE HEAT CONDUCTION 53, The first and the second integrations ofthe differential equation give, respec- tively, are) a TO) = The boundary conditions at r = a and r = b are applied to the equation for Toto give Pyne + Cy T=CnatC; T=C,Inb+C A simultaneous solution of these two equations gives Ba In@a) Ina C= T-h- Win tay Introducing these coefiients into the above equation for T(), we obtain TO)~T, _ In(ola) T—T ~ In(bja) The heat flow rate Q over a length H of the cylinder is determined from iM = a¢)- area = 47 rept = -knHtC, since dT(rjdr = (1/r)C,, Now, introducing C; into the expression Q, we find 2k = ey 8 Fada) ~ ‘This expression for Q is now rearranged in the form o=-5eh G-120) where 1 (/a) 2k omS54 wear reansre [Equation (3-125) can also be rearranged in the form In(b/a) _ (b ~ a)In [2xbH\(2naHt)) Dak ~ (6 — ane Rae G29 where ae Ar= Ao 3-124) in (Ai/A0) om when area of inner surface of cylinder 2nbH = area of outer surface of cylinder logarithmic mean area — a= thickness of eylinder Here, R, as defined above, is called the thermal resistance for a hollow cylinder. Example $7 A hollow cylinder with inner radius r= a and outer radius is heated at the inner surface at a rate of qo W/m? and dissipates heat by convection from the outer surface into a fluid at temperature T., with a heat transfer coefficient h. There is no energy generation, and the thermal ‘conductivity of the solid is assumed to be constant. Develop expressions for the determination of the temperatures T, and 7, of the inner and outer surfaces of the cylinder. Calculate the surface temperatures T, and T; for a= 3cm, b = Sen, 1h = 400 Wyn? °C), Ty, = 100°C, k = 15 W/m: *C), and gg = 10* W/m? SOLUTION Since there is no energy generation in the medium, it is more ‘convenient to solve this problem by utilizing the thermal resistance approach, Figure 35 illustrates various thermal resistances in the path of heat flow. We therefore immediately write a Fev ee Tiesto aldo ty (bjay(QakH) ~ T)CRbEN) T, * in @jayQnki) + 1OnbEh) Here H is the length of the cylinder, and the last expression is obtained by adding the numerators and the denominators of the second and third ex- pressions. Clearly, in these expressions 2H cancels, and the surface tempera tures T, and T, are the two unknown quantities.ve tam Figure 35 Nomenclature for thermal resistance con- cept for Example 1 eee (tnd «geen ‘and equating the first and third expressions, we find 1, pate t Te By introducing the numerical values given above into these solutions, we find t= (B08 + ggg) «10 4100-322¢ aol x 10° + 100 = 250°C T 0.05 x 400 3:3 THE SPHERE ‘The one-dimensional, steady-state temperature distribution T(r) in a sphere in which energy is generated at a rate of g(r) W/m? is governed by the heat con- duction equation 1a(.dr) 1 ba (0G) +po0-0 ou)56 war meansren This equation must be solved over the domain of the sphere subject to appropriate boundary conditions. Once the temperature distribution T() is known, the heat flux (7) anywhere in the medium is determined from the definition aT) ar a) Wim? G14 To illustrate the procedure, we consider Eq, (3-13) for the ease of constant ‘energy generation gp and write it in the form 0,2 G19, 4 (41) a" a k where go = constant. The first and second integrations of Eq. (3-15) give, re spectively, 470), : Fem Hr 616) ap _ 1-17) Te) =~ Br Sc, 7 Two boundary conditions are needed to determine the two integration constants C, and C,. In the case of a hollow sphere, illustrated in Fig. 3-6b, the boundary conditions at the inner and outer surfaces can be a prescribed temperature, @ prescribed heat flux, o a convection boundary condition. The case in which both boundary surfaces are subjected to prescribed heat flux is not considered for the reasons stated previously in connection with the slab and the cylinder, In the case ofa solid sphere, illustrated in Fig. 3-6a, a boundary condition can ’be specified for the outer surface, but another boundary condition is needed at the center ofthe sphere. As discussed in connection with the solid cylinder, the boundary ‘condition at the center of the sphere can be taken as Te) = finite atr=0 G18) aT), ar Both houndary conditions lead to the same result G-18b) Figure 36 Coordinates for oe imensionsl beat condection in & ‘phere (e) Sobd sphere, (6) hollow sphereONE-DIMENSIONAL, STEADY-STATE HEAT CONDUCTION 57 ‘We now illustrate with examples the application of heat conduction in a sphere. Example 3.8 The inner surface at r = a and the outer surface at r = b of a hollow sphere are maintained at uniform temperatures T, and T;, respectively. The thermal conductivity k ofthe solid is constant. Develop an expression for the one-dimensional, steady-state tempera- ture distribution 7() inthe sphere. Develop an expression for the radial heat flow rate Q through the hollow sphere. Develop an expression for the thermal resistance of the hollow sphere SoLuTIoN The mathematical formulation of this problem is given as df dTy PefO]-0 acres TH =T, atr=a TO=T, atr=b ‘The first and second integrations of the differential equation give, respectively, aT) dr T= - Fae, ‘The boundary conditions at r = a and r = b are applied to the equation for T() to give and T= -S4, PD “Pr bna tr bma The heat ow rate Q through the hollow sphere is determined from af _,aT@) = rr [-«52 r58 Hear TRANSFER ‘When C, is substituted, we obtain ab Q = 4ank - - T) ‘This result is now rearranged in the form 3-194) b-a ‘ankab where (3-19) and R is called the thermal resistance for a hollow sphere. Example 39 A hollow sphere of inside radius r = a and outside radius is electrically heated at the inner surface at a constant rate of qo W/m?. At the outer surface it dissipates heat by convection into a fluid at temperature T., with a heat transfer coefficient h, The thermal conductivity k of the solid is constant. ‘Develop expressions for the determination of the inner and outer surface temperatures T, and T, ofthe sphere. ‘Calculate the inner and outer surface temperatures fora = 3 cm,b = Sem, fh = 400 W/(m?-°C), Tz = 100°C, k = 15 Wi(m-°C), and gg = 10° W/m? ‘SOLUTION This problem can be readily solved by the thermal resistance ‘concept since there is no energy generated in the medium. Figure 3-7 illustrates the thermal resistance network for this problem. We therefore write tag, « i=. k= To nado = aj(ankab) ~ 1/(xb°h) -_—_hi-Te __ ~ © ayiarkaby + 1AGab™h) In these equations, 7; and T; are the only unknowns; the last expression is ‘obtained by adding the numerators and denominators of the second and third expressions. n t 7 sx0'gg eo — On ow v Ge Fine 37 Thermal resistance network for Example 34By equating the first and last expressions, T, is found: v-[Pe*s Gib and by equating the first and third expressions, T; is found: fa\? 1 (tae. By introducing the numerical values given above, we find the surface eee) = PBEM (Y ge] on 2c T, = @abs * 10° + 100 = 190°C T 3.4 COMPOSITE MEDIUM In many engineering applications, heat transfer takes place through a medium ‘composed of several different layers, each having different thermal conductivity. Consider, for example, a hot fluid flowing inside a tube covered with a uniform layer of thermal insulation. The thermal conductivities of the tube metal and of insulation are different; hence the heat transfer problem from the hot fluid to the colder, outer environment involves conduction through a composite medium consisting of two parallel concentric cylinders. The thermal resistance concept discussed earlier is applied now to the prediction of one-dimensional, steady-state heat transfer rate by conduction through a composite structure. Composite Slab Consider a composite wall consisting of three parallel layers in perfect thermal contact as illustrated in Fig. 38. Consider the heat flow rate Q through an area 4 of the slab. The equivalent thermal resistance network is also shown in this figure. By applying the thermal resistance concept, we immediately write L-h_h-h_Ti-h_h-h_B-h @ R R RB ‘where various thermal resistances are defined as Ret A gn RK, Ah, 1 Ay ake© xr RANGER Figure 34 compote of thre walle in series paths and the equivalent thermal esitance network By summing the numerators and the denominators of the individual ratios in Eq, (3-20), we obtain w 3-220) where R=R4+R +R +R +R 6.22) Here R is the total thermal resistance in the path of heat flow through an area A from temperature 7, to Tj, and various individual resistances are defined by Eq. (G-21), The composite wall arrangement shown in Fig. 3-8 is a series arrangement because the individual thermal resistances are connected in series. t—-—4 Invuted Figere 3.9 A composite of to material it wh parallel paths snd the equivalent thera) ak resistance network‘ONE-DIMENSIONAL, STEADY-STATE HEAT CONDUCTION 61 oe 2 0 Figue 3:10 Equivalent thermal resistance network approximating heat ow through ‘composite wall a one-dimensional heat fw. Figure 3-9 shows a composite of two materials combined in parallel paths with the ends maintained at uniform temperatures T, and ,. The equivalent thermal resistance network is also shown in this figure. The total heat transfer rate Q through this parallel arrangement is given by o- 230) where the equivalent parallel resistance is ett ear Ccares 320) R°R RL tL xed) Figure 310 illustrates a composite of several different materials arranged in parallel and series paths. An equivalent thermal resistance network shown in this figure is based on the assumption that the heat flow path can be approximated as one-dimensional. In reality, the heat flow through such a system is two-di- ‘mensional. But if all the materials have the same thermal conductivity, the heat flow is truly one-dimensional, and the analysis becomes exact. Therefore, ifthe thermal conductivities ofthe materials in the composite do not differ significantly, the approximation ofthe problem as one-dimensional may be reasonable. Other- wise, two-dimensional analysis is necessary. Example 3-10 Consider the composite of two materials combined in parallel paths with the ends maintained at uniform temperatures as illustrated in Fig.3.9, Various quantities are specified: A, = 02m?, ky = 20W/(m-°C), Az = 04 m?, ky = 15 Wilm-°C), L= 05 m, T, = 150°C, and T= 30°C. Caleulate the rate of heat transfer Q across the composite medium. SoLUTION The equivalent parallel resistance is computed according to Eq, G.230) as Aaky _02 x 20, 04 15 _ 5p rarer ge aes Then the heat transfer rate @ is calculated by Eq, (3-234) as 7 . (150 — 30x20) = 2400 W ‘Composite Coaxial Cylinders Consider a composite cylinder structure consisting of two coaxial layers in perfect thermal contact, as illustrated in Fig. 3.11a. A hot fluid at a temperature T, flows inside the tube, and heat is transferred to the tube wall with a heat transfer coefficient h,.On the outside, heat transfer takes place from the exterior surface of the tube (o a cold fluid at temperature 7, with a heat transfer coefficient hy, ‘The total heat transfer rate Q from the hot to the cold fluid over the length Hf of the cylindrical structure is the same through each layer and is given by T_h-h b-h -524 Be 2) here various thermal sistance are defined a, 7 Be Saat 75 Lon 1 ay 2 036 Wi(m -°C) T- 500 ~ 300 Q = Ak OF = (1 x 036) Se We note that for the linear variation of k with temperature the heat flow rate through the slab can be calculated by the simple relation given by Eq. (3-70) in which the thermal conductivity kis evaluated at the arithmetic mean of the boundary surface temperatures, that is, (Ty + T;)/2. 400 W Hollow Cylinder Consider a hollow cylinder in the region a aero Wi est anf coetient hy = 15 Wit! -°C)- Determine the eat rane rate rom the rd nto the surounding air at T, = 37°C “anoner: 2016 W ‘3.84 A tan rod ofwiform crossection 4, length L, and thermal conductivity kit thermal) ata 84 A tin oe (wo wall which are maintained at tmperatures T, and T. The rod i asipating hes Krom i eral surac by convection into the ambient air at T witha heat ans coefcient he Decie the difeaual equation governing the one-dimensional steady-state temperature itibation TO) in the od. Ta) Bysohag thaequationdeelopanexpresson forthe emperatare dsributon T(sintherod (G) Buseiop tn expresion forthe Beat fos fom the od ito the ambient i, “3.88 Consider a ste cod of length L = 50 cm, diameter D = 2em and thermal conductivity & = $$ Wien) Ove ond ots rode thermal tached to « ot soface maaan at Th 154° Pere Cred i attached to « old rule maintained at T, = SPC. The rod dspaes Heat by 208 tation ito tbe ambien air at temperature, = 20°C with a heat transfer oeficient he = 13 Sina °C), Determine te beat loss fom the ro at the ambien i. What fraction ofthis eat Loss isfrom the sorface maintained at T, = 190°C? MAniner: 194 W, 3487 ‘3.86 Avery long slender bas ro of diameter D = 2om and thermal conduciviyk = 60 W/m °C) TRAIT, Minted atone end to a lag, bot surlae maintained a T, = 200°C. The fod dissipate, Herma con wha heat ane coeticeat he = 25 Wis -°C) into an ambient at Ta = 3: 1G) Calelate the heat lose trough the rod info the ambveat i. (0 arcopper rad [k= 386 W°C)] of intl size were usc in pce of the brass ro nat would be the revaling change in the heat los through the rod? Jar Asin od oflength L = Wem, diameerD ~ 1m,and thermal conduetivy & = 65 Wim °C) ae aot ewontaly to large tenk at temperature Ty = 200°C. The od is dsipating eat by a aces vos une ambent at at T — 20°C with a heat tater cocticient he = 15 WYte"- °C) ‘Whar the temperatre ofthe rod 10 aod 20cm from the tank? “Anoner, AC, 01°C ‘3.88 Two very long. slender rods of the same dizmeter are given. One rod is of aluminum and has 2 ae rreeaaiay ky = 00 Wim. €).but the thermal conductivity kote ober rodisnarknown aaa toe te thermal conductivity of the ok od, on end of each rod is thermal tached Fe dee chic is maintained al «constant trperatre T,, Both rods are losing heat by ns at eat anf coeffovent hint he amen ai at 7, The surface temperature of ech Tegan ar vaiousdixances fom the bot base surface. The temperature ofthe aluminum rod(ONE-DIMENSIONAL, STEADY-STATE HEAT CONDUCTION 99 il — 9+ (hdd tl — cosh m+ him) nha sng thie deeop an xeon fr ht ander theaph tha Temperate deendet 7) ad vara cet an 2a Be neal srice of «rod i pec nated while tend atx = O and x thermal coer T and 7, tespecively, The cross-sectional are 4 cake esas ‘ermal condvcity ofthe matcil varie mthteoperienn Mn are kept a and the il +27) Whee ty and ware constants (@) Develop an expresion forth het fw rte Q through the cod (©) Devclop an expression forthe thermal resistance of tes so ‘391 Thermal condciviy of plane wal aes wt temperature scoring to he elation HT) = kya + 6) Where and fare constants, {@) Develop an expresion forthe heat fow throu snd + Lare maintained at worm temperturs 7, (0 an tion fr the thermal reise othe wall fhe et rnd suri Oak com aT trough A= Ol mathe pele aoe cc, 1 = 04 my = 60 Wim "Chand f= 028 5 10-8 Spec ale x Oa = Lave tpt wir tempers 7, end Tywe ‘pesinely and the thermal conductive of the material depends on tcopeener na ee We = kar — 79, where ky and Ty ae constants boundary surfaces at r= a and r= 6 siataned at tan [reggie The thermal conductivity ofthe material varies wth ereraune Mel SBT, Develop an expression forthe heat flow tough aah ena ot activity varies wth tempera: 0:40 + 10"*7), where Pi in depres Caen ‘Determine the heat fow rate per Im length of ind Sr shair=an ‘worm temperatures 7 and, respectively The thermal cont ay A) = be + ar-4 prt, 7 = bare maintained es with emperatre 8100 ear TRANSFER (a) Dewlop an esgrenion rhe ll eat How rte @ hh oop rn forte teal restr of the hollow ee spor ema i of hicks wich ea i gee a ST of go Wm?. The Bo come dant = Loti aba manne temas and T,respectivey aoa conductivity ofthe ib vases with emperare a8 ary = kl + TY erent vecontans Develo an exreion rhs a. 8h 8 sateen conto ne alo hikes Lair heme a= kl +97) where kan Bae constants, The boundary soracesatx = O30 = aremintaned a terperatures and Ty respetively. enn forthe heat ow ata anatee HEE one he beat wane reer An OM LOAN. TS Chg = wie Stand fe 035 x 10-C 7280 compare ar ea Hh Souaned in Problem 3-9 REFERENCES 1, Ceuta, TN, and M. Fenn: Terma Coase MESS in Contact. Gener Cana TN a rar, Confer of sono Mecanzl ESEOS fends) and ASME, pp. 271-215, 1981 2. Banwiay'M. E, KN. Tong and G.F, Holloway sexect of Pressure on Thermal Condocance aay teins” NACA Tech Note 3295, May 1955 seer i perne Termal Cont Condes Genera His 6482, May toed aaa ARS Paper No. 190-61, AD 196 eee ee Spin eros Therma Contact Resistance Pome Space Weis.” RS rma, 32:237-283 196, ees oman WH. Rolucoow:“Predtion of Therma) Comsat of Meta Surfaces in ene ear Trager, Tra ASME. 8: 15-26 (98). i ees ernie nee Condos + Yawn? AIAA Paper No. (4-253, 1964 ee gan 87. Chao: “Tm Conta Resta 0 ASR Environment,” She Franger,81:249-265 (1969) er rate Tara Conact Conducta: Geel FASE Co, 658495, March 1965. a ese. HA, Bum. an Hin: Sabet CS? pibiography for Thro Moor gaice Std” ASME Pope OE WALHT-K Dante, Fae BG BB Miki and M.M. Yowanoih: "Thermal Cont ‘Conduetances," It. J. ‘eat Mass Transfer, 12:279-30 (1969) "aM Tit Teanser at the In stage ofDissimiar Metals The tauene of Thermal ped Hat Mass Transfer, 9:791-$01 1960) Panna t= Conetion of Temal Conta! Consumes experimental Results,” Pro. yea so 2, Academe Pe, 1c. New York 195 ene A nd. Re Brown Matheateal Equations or He Conduction nthe Finsof Homrrcled Bagnes” NACA Rep. 158, 1922 ee Re ene of ended Sree.) Tne AON 21-631 (1949) 3 Gard and AD Ks Extended Sra Het Tene McGraw-Hil, New York, 1572‘CHAPTER FOUR TRANSIENT CONDUCTION AND USE OF TEMPERATURE CHARTS I the surface temperature of a solid body is suddenly altered, the temperature within the body begins to change over time. It will take some time before the Steady-state temperature distribution is reached. The determination of the temperature distribution within the solid during temperature transients is a more ‘complicated matter because temperature varies with both position and time [1-9], In many practical applications, the variation of temperature with position is negligible during the transients, hence the temperature is considered to vary with time only. The analysis of heat transfer under such an assumption is called the lumped system analysis, and since the temperature is a function of time only, the analysis becomes very simple. Therefore, in this chapter we begin with the lumped system analysis of transient heat conduction problems. ‘The use of transient-temperature charts isillustrated for solving simple transient heat conduction in a slab, cylinder, and sphere in which temperature varies with both time and position, The method of product solution is described for the analysis of multidimensional transient heat conduction. The use of conduction shape factors is discussed for predicting the steady-state heat flow in two-dimensional haat flow systems. Finally, the analytic solution of transient heat conduction by the method of separation of variables is presented for the case of the slab geometry, and the use of tabulated solutions is introduced. 41 LUMPED SYSTEM ANALYSIS ‘Consider that a solid of arbitrary shape, volume V, total surfaog area 4, thermal conductivity k,, density p, specific heat c,, at a uniform temperature Ty is suddenly immersed at the time t = 0 in a well-stired Quid which is kept at a uniform temperature T,,, Figure 4-1 illustrates the considered heat transfer system. Heat transfer between the solid and liquid takes place by convection with a heat transfer 10Figure 41 Nomenclature for Tumped system analyse of tra sen eat flow. coefficient h, It is assumed thatthe temperature distribution within the solid at any instant is suficiently uniform that the temperature of the solid can be considered to be a function of time only, that is, T(0). The energy equation for heat transfer in the solid may be stated as ‘Rate of heat flow into the \\ [rate of increase of solid of volume V through} = { internal energy ofthe solid) (4-1) [boundary surfaces A of volume V By writing the appropriate mathematical expressions for each of these terms, Eq, (4-1) becomes = pe,v TO ANLT,, ~ T(Q] = pe,V aa 40), Ab ry . fh pv TO fort > 0 (420) subject tothe intial condition TW)= Ty fort =0 (42) For convenience in the analysis, anew temperature O()is defined as Ht) = TW) - T. (43) Then Eqs. (42) become 20 4 mit) =0 fort > 0 (440)“TRANSIENT CONDUCTION AND USE OF TOMPERATURE CHARTS 103 Figere 42 Dimensionless temperature 0/65 a8 8 function of time and MQ) = TT. =O fort 0 (4-4) wee we have defined Ak pe aod Equation (44a) is an ordinary differential equation for the temperature 6(), and its general solution is given as 4) = Cem 6) ‘The application of the initial condition (4-46) gives the integration constant as C = 65, Then the temperature of the solid as a function of time is given as «7 Figure 4-2 shows a plot of the dimensionless temperature given by Eq. (4-7) as a function of time. The temperature decays with time exponentially, and the shape of the curve is determined by the value of the exponent m. Here m has the dimension of (ime)! Clearly, the curves in Fig. 4-2 become steeper as the value of ‘m increases. That is, any increase in m will cause the solid to respond more quickly to a change in the ambient temperature, An examination of the parameters in the definition of m reveals that increasing the surface area for a given volume and the heat transfer coefficient increases m. Increasing the density, specific heat, or volume decreases m. To establish some criteria under which the temperature distribution can be considered uniform within the solid, and hence the lumped system analysis becomes. applicable, we define a characteristic length L, as L (48) and the Biot number Bi as 9)104 ear rmanseen wer ih tema conti fhe ol For sods inte hae ofa, ionnndr and spose temprarcdstnbusonderaguaere oases so any hans nfo isan cori ha sb paene vi co 0 This matter is discussed further and becomes clear later in this chapter. Here we assume that the lumped system analysis is applicable for situations in which Bi < 01, ‘The physica significance ofthe Biot number is better envisioned ifit is arranged in the form h kL, which is the ratio of the heat transfer coefficient for convection at the surface of the solid to the specific conductance ofthe solid, Hence, the assumption of uniform. {temperature within the solid is valid if the specific conductance of the solid is much larger than the heat transfer coefficient for convection. Bi au) Example 41. An aluminum plate [k = 160 W/(m °C), p = 2790 kglm?,c, (0.88 KJ((kg-°C)] of thickness L = 3 em and ata uniform temperature of Ty 225°C is suddenly immersed at time t = 0 in a well-stired fluid maintained at a constant temperature T,, = 25°C, The heat transfer coeficient between the plate and the fluid is k = 320 Wi(m? - °C). Determine the time required for the center ofthe plate to reach 50°C. ‘SOLUTION This problem can be solved by the lumped system analysis with sufficient accuracy if Bi <0.1, Therefore, we need to check the magnitude of the Biot number. The characteristic dimension L, is determined as volume LAL 0 area 242 Som ‘Then the Biot number becomes 300 x 15 x 10-7 2 aja Me WOK LS 1 5 1g ‘which is less than 0.1, hence the lumped system analysis is applicable. From Eq. (4-7) we have where WC T= 28°C T= DSC and h 320 BepL, ~ 2790 % 880 x[TRANSIENT CONDUCTION AND USE OF TEMPERATURE CHARTS 105 50 = 25 0.00869 25-25 (0.008691 = 2.079 63230524 min Example 42 The temperature of a gas stream is measured with a thermo- ‘couple. The junction may be approximated as a sphere of diameter D = 1 mm, k= 25 Wi(m=°C), p= 8400 kg/m’, and cy = 04ky/(kg-°C). The heat transfer coefficient between the junction and the gas stream is h = 560 W/(mn? -°C). How long will it take for the thermocouple to record 99 percent of the applied iemperature difference? Souution The characteristic dimension L i v_ te Lae ae “The Biot number becomes 56010" AL. a ee ag EST X10 hence the lumped system analysis is applicable. From Eq. (4-7) we have To -T, T= To When the temperature reaches 99 percent of the applied temperature dif- ference, we have The value of mis ‘Then ‘Mixed Boundary Condition Inthe previous discussion we considered a situation in which all boundary surfaces of the region were subjected to convection. The method is also applicable when part of the boundary surface is subjected to convection and the remainder is subjected to prescribed heat flux as now illustrated,‘oats Heat sup Win? Figure 43 Nomenclature for lumped analysis of transient heat flow inal Consider a slab of thickness L, initially at a uniform temperature Ty. For times ¢ > 0, heat is supplied to the slab from one of its boundary surfaces at a constant rate of q W/m? while heat is dissipated by convection from the other boundary surface into a medium at a uniform temperature T., with a heat transfer coefficient A Figure 4-3 shows the geometry and the boundary conditions for the problem, We assume an equal arca A for heat transfer on both sides of the plate. The application of the energy balance equation (4-1) for this particular case gives Aq + AMT, ~ T()] = peyat 7 ath TO)=p LTO forr>o 4120) ‘with the initial condition TW=T for For convenience in the analysis, a new temperature O() is defined as 0 (4:12) avert “9 Then Eqs. (4-12) become 9 5 payee r1>0 (ate) and aoe Te2h bee at) where we hive dened h = 4 ce om 0=5 as The solution of Eq, (4-142) is written asa sum of the solution of the homogeneous part of Eq, (4-14a) and a particular solution in the form HD) = Cem +0, 4-16) ‘where Cis the integration constant. The particular solution 6, is given by (4-17){TRANSIENT CONDUCTION AND USE OF TEMPERATURE CHARTS. 107 By combining Eqs. (4-16) and (4-17) we obtain yaad ea = is determined by the application ofthe initial condition, ‘The integration constant (4-14b) as a=c+e 19) Substitution of Eq, (4-19) into (4-18) gives the solution ofthis heat transfer problem (0) = Bye"™ + 1 = 0) = Bem + (1 420) Fort -+ «this soution simplifies to Qa 4a) = 2-8 ay ‘which is the steady-state temperature in the slab. Example 43 A household electric iron has a steel base [p = 7840 kg/m’, ¢, = 450 J/kg -°C), and k = 70 W)(m-°C)} which weighs M = 1 kg. The base hasanironingsurface of A = 0.025 m? and is heated from the other surface with @ 250-W heating element. Initially the iron is at a uniform temperature of T, = 20°C. Suddenly the heating starts, and the iron dissipates heat by convection from the ironing surface into an ambient at T,, = 20°C with a heat transfer coefficient h = 50 W/(m? - °C). CCaleulate the temperature ofthe iron ¢ = Smin after the start of heating. ‘What would the equilibrium temperature of the iron be if the control did ‘ot switch off the current? SOLUTION The thickness L of the base is determined to be M 1 Ap ~ W025 x 840 O51 x 107? m ‘The Biot number becomes 50 x Ot x 10°? 70 = 036 x 107?108 ear reansrex hhence the lumped system analysis is applicable, From Eq. (4-20) we have (0) = Oye™ + (1 my 1 as where 0) = T-T, = 20-20=0 250 ‘i 4 = ggg = 10000 Wii fh = 50 Wim? -°C) h 50 egh ~ TRAD ABO x OSI x 10? ~ 0278 * 10° Then the temperature at the end of t = 5 min is determined as mt = 0.278 x 10°? x 5 x 60 = 0834 04 = 1 = 0994) 2 re T() = Ht) + Te. = 113 + 20 = 133°C ‘The equilibrium temperature becomes 10,000 (20) = 2 — 200° (co) = Woo) + T, = 200 + 20 = 220°C 42 SLAB—Use of Transient-Temperature Charts In many situations the temperature gradients within the solid are no longer negligible, hence the lumped system analysis is no longer applicable. Then the analysis of heat conduction problems involves the determination of the tempera ture distribution within the solid asa function of both time and position, and itis a ‘complicated matter. Various methods of analysis for solving such problems are discussed in several texts [1-5] on advanced treatment of heat conduction ‘Simple problems, such as one-dimensional, time-dependent heat conduction in a slab with no internal energy generation, can be solved readily by the method of separation of variables, as described later in this chapter. In addition, the tempera- ‘ure distribution for such situations has been calculated, and the results presented in the form of transient-temperature charts in several places [6-8]. We now resent transient-temperature and heat fow charts and discuss their physical significance and use. Consider a slab (ie, a plane wall) of thickness 21 that is confined to the region =L < x < L. Initially the slab is at a uniform temperature 7,. Suddenly, at ¢ = 0, both boundary surfaces ofthe slab are subjected to convection with a heat transferss To) Figure 4-4 Geometry, coordinates, and boundary condition for ransint het conduction in slab coefficient h into ambients at temperature T., and are maintained so for > 0. Figure 4-4a shows the geometry, coordinates, and boundary conditions for this particular problem. However, in this problem there is both geometrical and thermal symmetry about the x =0 plane, so we need to consider the heat con- duction problem for only half the region, say, 0 < x < L. With this consideration the heat conduction problem for the slab of thickness 2L confined to the region =L 0 (4-220) Feo ax=oforr>o (4-228) rT enreht, atx=Lfore>0 (4220) T=T — fort=Oind0 (4240) atX =0,fort>0 (4-240) aX = I,forr>0 (424c) Oat in0sXiforr=0 (249 ‘The physical significance of the dimensionless time + or the Fourier number is better envisioned if Eq (4-23d is rearranged in the form rate of heat conduction across L in volume L?, at _ K/L)? _ WFC PeyLit~ rate of eat storage in volume LPC ‘Thus, the Fourier number is a measure of the rate of heat conduction in com- parison with the rate of heat storage in a given volume element. Therefore, the larger the Fourier number, the deeper the penetration of heat into a solid over a siven time. ‘The physical significance of the Biot number is better understood if Eq. (4-23e) is rearranged in the form (4-250) heat transfer coefficient at nL, __h__ the surface of solid “K ~ W/L ~ internat conductance of solid across length L That is, the Biot number is the ratio of the heat transfer coefficient to the unit conductance of a solid over the characteristic dimension. ‘Comparing the heat conduction problems given by Eqs. (4-22) and (4424), we conclude that the number of independent parameters that affect the tempera- ture distribution in the solid is reduced significantly when the problem is expressed in the dimensionless form. In the problem given by Eqs. (4-22), the temperature {depends on the following eight physical parameters: Xt Lyk fy Ths To However, in the dimensionless problem given by Eqs. (4-24), the temperature depends on only the following three dimensionless parameters: X,Bi,and (426) Clearly, by expressing the problem in the dimensionless form, the number of parameters affecting the temperature distribution is significantly reduced, There- fore, it becomes feasible to solve such a problem once and for all and present the results in the form of charts for ready reference. (4-256)i Haitian t H 1 58 12-161 20124128140 40 [0 to) Tite BorPoosto20s [235 1020 90 | or 1 10 0 Figure 45 Transientemperature chart for «slab of thickness 2L subjected to convection at both boundary surfaces. (a) Temperature Tp atthe center pase. x = 0; (0 poston correction for use with pat a) (Frm Heuer (6)112 wear TRaNsru ‘Transient-Temperature Chart for Slab The problem defined by Eqs. (4-28) has been solved, and the results for the d- mensionless temperature are presented in Fig. 4-Sa and b, Figure 4-5a gives the midplane temperature Ty or (0, 2) at X = 0 a8 a function of the dimensionless time + for several diferent values ofthe parameter 1/Bi. The curve for 1/Bi = 0 corresponds tothe case + 0, or the surfaces of the plate are maintained at the ambient temperature 7, For large values of 1/Bi, the Biot number is small or the internal conductance ofthe sold is large in comparison with the heat transfer coefcient at the surface. This, in turn, implies thatthe temperature distribution within the solid is sufficiently uniform, hence the lumped system analysis becomes applicable. Figure 4-5b relates the temperatures at diferent locations within the slab to the midplane temperature Ty. If we know Tp, temperatures at diferent Ications Within the slab can be determined. ‘An examination of Fig. 4-5b reveals that for values of 1/Bi larger than 10, or Bi <0, the temperature distribution within the slab may be considered uniform with an error less than about 5 percent. We recall thatthe eriterion Bi < 0.1 was used forthe lumped system analysis to be applicable. Figure 446 shows the dimensionless heat transferred Q/0, as a function of mensionless time for several different values ofthe Biot number for a slabof thick ness 21. Here Q represents the total amount of energy which s lost by the plate up to anytime ¢ during the transient heat transfer. The quantity Qo defined as Qo = pepMT—T.) Wes 27 represents the initial internal energy of the stab relative to the ambient temperature, Example 4-4 A 5-cm-thickiron plate [k = 60 WAm -°C),c, = 460J/(kg °C), 850 kg/m, and a= 1.6 x 10-* m?/s] is initially at 7; ~ 225°C. oo ea agree Figure 46 Dimensionless heat transfered Q/Qg fra sab of thickness 2L. (From Ref 8)[TRANSENT CONDUCTION AND USE OF TEMPERATURE CHARTS. 113, ‘Suddenly, both surfaces are exposed to an ambient at T,, = 25°C with a heat transfer coefficient h = 500 W/m? -°C). ‘Calculate the center temperature at = 2min ater the start ofthe cooling. ‘Calculate the temperature at a depth 1.0 cm from the surface at ¢ = 2 min after the start of the cooling. ‘Calculate the energy removed from the plate per square meter during this time. ‘Souumon ‘The transient-temperature charts may be used to solve this problem since the lumped system analysis is not applicable. We have wesxi0?m or L=25x10?m oat Ty = Te + (T~ TeXO58) = 25 + 200(0.58) = 141°C “The temperature 1.0 em from the surface is determined as x_ 28-1 Las 7° For 1/Bi = 48 and x/L = 06, from Fig. 4-5b we have T-T. T = T, + (To ~ TaX095) = 25 + (141 ~ 25095) = 1352°C “The heat loss from the plate per square meter (including both sides) during the transients uptot = 2 minis determined as follows: From Fig. 46,for Bi = 0.21 and Bi?-+ = 0.217 « 3-1 = 0.137 we find aL Q 4s114 wear reansren where Qp, by Eq. (4-27), is Qo = pe(2LXAXT, — Ty) (7850)(460X5 x 1021225 = 25) = 3533 x 10°) Then the heat loss from the slab per square meter in 2 min becomes Q = 0.459, = 159 x 10°) 43 LONG CYLINDER AND SPHERE—Use of ‘Transient-Temperature Charts The dimensionless transient-temperature distribution and the heat transfer results similar to those given by Figs. 4-5 and 4-6 also can be developed for a long cylinder and sphere. ‘Transient-Temperature Chart for Long Cylinder Consider one-dimensional, transient heat conduction in a long cylinder of radius », which i initially at a uniform temperature T,. Suddenly, at time t = 0, the bound- ary surface at r = b is subjected to convection with a heat transfer coefficient h {nto an ambient at temperature T., and maintained so for t > 0, The mathematical {formulation of this heat conduction problem is given in the dimensionless form as 12 (, 30) 2 LA (RB) mocrcnorr>o aay a Bag unenierso at a ee emen ea (e289 tat wostsimreo zag we vow cmesone ume dete los vi =! = atau (6389 72 = dimensionless time, or Fourier number (4-296) dimensionless temperature (4-290) = dimensionless radial coordinate (4298)HOH : eee 19 o| | of o4| TT. Bor o0s 0210235 10 20. 100 or 01 0s 0 Figere 47 Transienttemperatar chat for # long sid cinder of radius r = 6 subjected to con ‘retin atthe boundary srfaer = . (2) Temperature Ty athe ais ofthe cylinder; (8) positon cor- eton fo se with put). (Frm Heiser (6).) us116 wear reansren Gos a Be gy Figure 48 Dimensionless eat trasfered fora long clindr of ada (From Rf 8) ‘The problem given by Eq. (4-28) has been solved, and the results for the center temperature 7, or (0, 1)are shown in Fig. 4-7a asa function of the dimensionless. time + for several different values of the parameter 1/Bi, Figure 4-7b relates the temperatures at diflerent locations within the cylinder to the center temperature T,. Therefore, given Ty, temperatures at different locations within the cylinder can bbe determined from Fig. 4-7b Figure 48 shows the dimensionless heat transferred Q/Q, as a function of dimensionless time for several different values of the Biot number for the cylinder problem given by Eqs. (4-28). Here, Qo is as defined by Eq, (4-27), and Q represents the total amount of energy which is ost by the cylinder up to any time # during the transient heat transfer. ‘Transient-Temperature Chart for Sphere For a sphere of radius b, initially at a uniform temperature T;, and for ¢ > 0 subjected to convection at the boundary r = b, with a heat transfer coefficient h into an ambient at temperature T.., the transient heat conduction problem is «given in the dimensionless form as 12 fg.) LE (wa) mocn curso aon R 0 aR = 0,forr>0 (4-306) Be pie-0 R= 1forr>0 4-300 R = atR = fort « ‘) o= inO0 0. The applied temperature 7; may bbe higher or lower than the initial temperature 7; of the medium. 2. In Fig 4-116, at = 0a constant heat flux qo W/m? is imposed on the boundary surface at x = O and maintained for t > 0. 3. InFig. 4-1c, at = Othe boundary surface at x = (is subjected to convection with a fuid at temperature T, and with a heat transfer coefficient h. This con- dition is maintained for all't > 0. Here, the fluid temperature T, may be higher or lower than the initial temperature 7; of the soli. ‘Clearly, the temperature response within the solid will be different for each of these three cases ‘The semi-infinite medium transient heat conduction problems have numerous ‘practical applications in engineering. Consider, for example, temperature transients in a slab of finite thickness, initiated by a sudden change in the thermal condition at the boundary surface. At very early times, the temperature transients near the boundary surface behave similar to those of the semi-infinite medium, because some time is required for the heat to penetrate the slab before the other boundary condition begins to influence the transients. ‘The transient heat conduction problems for the three cases illustrated in Fi 411 have been solved, and analytical expressions are available forthe temperature oS Fos Bo So ‘Srreaon a = sity = fad + : igo Tah o ” © Figure 4.11 Tare diferent boundary condition for ransieat het conduction in sem nfit soi[TRANSIENT CONDUCTION AND USE OF TEMPERATURE CHARTS 121 distribution in the solid as a function of time and position. We now examine the results obtained from such solutions for each case. (Case 1 The solid is initially ata uniform temperature T,,and fort > Othe boundary furface at x = Ois maintained a a constant temperature Tyas illustrated in Fig {Eta For this particular case the dimensionless temperature TD) = To He, = PEO is plotted against the dimensionless parameter ~ x(a shown in Fig4-12. {The pipe sgnieance of this graph i as follows: Fora given value of x the Toph represents the vraton in temperatre with time t that particular location SE etc fora given value of the graph represent he vaFaton of tempera ture with postion within the solids that particular ime t in engineering applications, the heat fux at the boundary surface x = 0 is aso of interest For the case ilusrated in Fg. 41a, the analytic expression for the heat uz a the boundary surface x Ois given by Te — af = =D | Wm a3 Vast ‘0 | a x 1 a 02 oe On ia maren a Ona) ‘igure 412 Temperature distribution Tx, in semi-infinite solid which intially at Ti for > 0 the surface atx = Ois maintained a To122 ear Tmansren Clearly, when 4, is a positive quantity, the heat flow is into the medium, and Case 2 The solid isinitially at a uniform temperature T,,and or t > Othe boundary surface at x = 0 is subjected to a constant heat flux qo W/m, as illustrated in Fig. 4-11b, For this case, the temperature distribution T(x, ) within the solid is given by (4320) Text) = 7, + con H+ Fete) ~ ] where ~ (4-32b) a me Here, the function er) iscalled the error function of arguments &,andits value are tabulated in App. D, Table D-L Case 3 Thesolidisinitially ata uniform temperature T;,and fort > Othe boundary surface at x = 0 is subjected to convection with a fluid at temperature T,, with a heat transfer coefficient h, as illustrated in Fig. 4-1 1c. For this case, the dimension. less temperature is ploted as a function of the dimensionless parameter x/(2/) for several diferent values of k/ aif as given in Fig 413, Note thatthe case» ao is equivtent to the boundary surface at x = 0 smaitained at a consant temperature 7 Example 446 A very thick concrete wall (a =7 x 10-7 m/s) is initially at a uniform temperature T; = 25°C. Suddenly one of its surfaces is raised to Tp = 125°C and maintained at that temperature. By treating the wall as a semi-infinite soli, calculate the temperatures at 5, 10, and 15 em from the hot surface 30 min after the raising ofthe surface temperature. SOLUTION This problem is the same as case 1, and the temperature distribu- tion is plotted in Fig. 4-12. Fora = 7 x 10~? m?/sandt = 30 x 60 = 1800s, the parameter & becomes 4x[TRANSIENT CONDUCTION AND USE OF TEMPERATURE CHARTS 123, aT 06.0 Tiss) To % Na ‘igure 4-13 Transient temperature Tx) in smn slid subjected to convection tthe bows dary sari. (From Ref. 9.) For x = 005m: = 07, and Tet) = To T-To Tex) = $8°C Forx = 010m: € = L4and Te = 8 _ 995 Text) = 30°C Forx = 015m:€ = 24,and Tx) = 125 25 = 125 Tox.) = 28°C It is apparent from these calculations that the effect of heating has not pene tated to a depth x = 0.15 m in 30 min124 ear TRANSTER Example 4-7 A water pipe is to be buried in soil at suficient depth from the surface to prevent freezing in winter. When the soils at a uniform temperature T, = 10°C, the surface is subjected to a uniform temperature of Ty = —15°C continuously for 50 days. What minimum burial depth is needed to prevent the freezing of the pipe? Assume that « = 0.2 x 10~* m/s forthe soil and that the pipe surface temperature should not fall below 0°C. SovuTion Figure 4-12 may be used to determine the temperature distribu- tion in the soil. For « = 0.2 x 10-® m?/s and t= 50 x 24 x 3600s, the parameter § becomes Bat ~ AOD RIO x 50% A x HO Taking T; = 10°C, Ty = ~15*C,and T(x 1) 2 0°C, we obtain Ts) Ty_ 0415 “T=% 10415 From Fig. 4-12, for (x, t) = 0.6, we determine ¢ = 0.6; hence 0538 = 06 06 0338 ‘hati, the pipe shouldbe buried at last toa depth of x= 1.12 m 0538 x, 12m 45 PRODUCT SOLUTION—Use of Transient-Temperature Charts ‘When the temperature gradients are important in not one, but, say, two dif ferent directions within the solid, then the problem is one of two-dimensional ‘transient heat conduction in a solid. When there is no internal energy generation in the medium, it is possible to combine the solutions obtained from one-di- ‘mensional transient-temperature charts and to construct the solution for a two- dimensional transient heat conduction problem. Such an approach, called the ‘method of product solution, is applicable ifthe solution of a two-dimensional, time- ‘dependent heat conduction problem can be shown to be equivalent to the product of the solutions of two one-dimensional, transient heat conduction problems. Demonstration of the Concept of Product Solution ‘The basis of the product solution is better envisioned with the following example. Consider a rectangular bar of sides 2L, and 2L,, confined to the region Ly SxS Ly and ~Ly < y < Lz, as illustrated in Fig. 4-14, Initially the slab{TRANSIENT CONDUCTION AND USE OF TEMPERATURE CHARTS 125 i 1Learonta een ren Te hyT hy rigore 414 Product station fr taint heat condeetion in a rectangular ba is at a uniform temperature 7. Suddenly at ¢ = 0 all boundary surfaces are sub- jected to convection to an ambient at @ constant temperature Ty. The mathe- ‘matical formulation ofthis heat conduction problem, in terms of the dimensionless temperature Ou = is given by 20 FO 1 tex chy ty 0 (43%) ax * ay can atx (4335) a reine (4339 By hoo aty (43%) ay re hon0 tye ls (4-33¢) ay o=1 fort = 0. (4339) It.can be shown that the solution ofthis two-dimensional problem can be expressed a product of the solutions of two one-dimensional problems 0,(x, 1) and 840, ) in the form (5, p. 54] Oe, 9.) = 0,08, 84 9126 sear mansren ‘where 0,(x,t)is the solution of the one-dimensional problem 270, _ 100, 1M im-tyexcbyt>0 4340) (4348) axeL, (434) fort =0 (434d) and €;(, 1s the solution ofthe following one dimensional problem #8, _ 100; Gera m-y0 3%) aty= =Ly (4-356) Ly (4359 fort = 0 (435d) ‘The validity of the above decomposition can be verified by substituting 0 = 0,0, in the original two-dimensional problem (Eqs. (4-33)] and utilizing the above problems, defined by Eqs. (4-34) and (4-35). Clearly, the above one-dimensional problems for the functions (x,t) and 0,0, 1) are exactly the same as that whose solution is given by the transient- temperature chart in Fig. 4-5. From the previous illustration we conclude that the solution of the two- dimensional heat conduction problem defined by Eqs. (4-33) for a rectangular region ~L, 0. Figure 4-16 shows the geometry, co- ordinates, and boundary conditions for the problem. The mathematical formula tion of this heat conduction problem is #TR.D _ 1TH Te) LTO ing exo (4370) subject to the boundary conditions Te) =0 atx=0,1>0 (4376) Te)= 0 atx=Lr>0 (4376) and the initiat condition Tex.) FG) fort =0,in0sx 0.[TRANSIENT CONDUCTION AND USE OF TEMPERATURE CHARTS 139 ‘Develop an expression for the temperature distribution in the slab. ‘Develop an expression for the heat flux at the boundary surface x = 0. Souurion This problem is exactly the same as that defined by Eqs. (4-37), ‘except the intial temperature distribution is specified as given above. Therefore, the solution (4-56) is applicable, by setting in that equation F(x) Ta sink (n/L)x). We find 113.1) = To 5 Herm sin ux [sin Ex’ in x he (4570) were foes (es) ws 1a 4a2 (est) ‘To evaluate the integral, we note that according to the definition of 4, we have nis ‘Then the integral term is written as [lana naar rt 9-13 Tied site ake Sa peta urna cto at to . (Oo for Ay # Ay [[inapednsese =f RAR Introducing this result into Eq. (4-572), we find all the terms of the sum- mation vanish except the term for n = 1, and the solution of the problem reduces to ‘This integral is similar to that given by Eq. (4-50a). ind for this 1 (4-58) Thx. 0) = Toe sin Ay (4-590) where L Figure 417 illustrates the transient-temperature distribution in the slab as siven by Eq. (459). ‘Once the temperature distribution T(x, t) in the slab is known, the heat fiux a(, t), anywhere in the medium, is determined from its definition: oT) ox (4-596) a.) (4-60)140 ear rRANSER temperature distibution ‘nal for an inal emperature Tesi) ‘Then the heat flux at the boundary surface x = Ois determined by introducing the solution (4-59) into Eq. (4-60) and evaluating it at x = 0. We find (0,1) = —KTge°4,[008 Ayx]ea0 (0,1) = —KTy ens (61a) where (461b) Since k, Tp, and 4, are all positive quantities, the right-hand side of Eq, (#614) is negative. The negative value implies that the heat flow at the boundary surface x = Oisin the negative x direction, or outward. This result is consistent with the physical reality for the problem. Example 4-13 A slab of thickness L is initially at a uniform temperature T, Suddenly at time f= 0 the temperatures of both boundary surfaces are lowered to and maintained at T,, forall : > 0. Develop an expression for the ‘temperature distribution T(x, in the slab Sowvtion The boundary condition for this problem is not homogeneous; thatis, the temperature at the boundaries isnot zero. Therefore, a new tempera ture 6(, 1) is defined as x) = TU, 1) — Te Then the mathematica formulation ofthis problem becomes 260.0) 10,9) Sa.) _ 125.9) ind0 (4620) He) =o ax-0 (4-626) He) = 0 atx=L (462) 2,0 =T,-T.= 8 forr=0 4624){RANSENT CONDUCTION AND USE OF TEMPERATURE CHARTS. 141 ‘This problem is now in exactly the same form as (4-37) with Fo) ‘Therefore, the solution (4-56) is applicable and we obtain t= mf hee ints finite The integrals evaluated as ae [sin yx ax’ 1 eel = ple anh = toe by = La = cos ma) i qu -Cy forn = even 0 2 2 torn = odd i ‘Then (x,t) becomes Noting that ‘we see that the solution is written as eanetn § +63) which gives the temperature distribution within the slab, in excess of the ambient temperature T., as a function of position and time. “48 TRANSIENT HEAT CONDUCTION IN A SLAB—Use of Tabulated Solutions In the last section, the analytic solution of transient heat conduction in a slab by separation of variables was illustrated for a sample problem given by Eqs. (437). In that example, for simplicity, the boundary conditions for both surfaces are taken to be ofthe frst kind. However, in engineering applications, other combina tions of boundary conditions are encountered. For the slab problem, there are142 mar TRANGHER nine different combinations of boundary conditions, and each such case requires separate analysis. ‘Therefore, in this section we develop one general formal solution for the transient heat conduction problem in a slab. Then this solution is used in con- junction with a table from which the eigenfunctions, eigenvalues, and normaliza- tion integral appropriate for any given combination of the boundary conditions ccan be obtained. To develop such a solution, we consider the following heat conduction problem for a slab with convection at both boundar AA slab of thickness L, confined to the region 0 < x < L, intially (ie, at t = 0) ‘has a specified temperature distribution f(x). Suddenly, at time t = 0, both bound- ary surfaces are subjected to convection with a heat transfer coefficient h into an ambient at temperature T,,. We now solve this problem. Let T(x, 1) be the solution of this problem. We define a new temperature (6(, 1), measured in excess of the ambient temperature T.., a8, Ms.) = Tes.) ~ Ty (464) This choice of a new temperature is made to obtain homogeneous boundary ‘conditions in the mathematical formulation of the problem. Then the mathematical formulation in terms of the temperature O(x, 1) is given by 2700.0) _ 1.0068, ind0 465a) ox? a Ot me 2 + hO(x,) = atx=0 (4-65b) Me 2 5 nx, =0 atx=L (4-65) 0 =/G)-Te= FO) fort =0 (4654) ‘We note that the boundary conditions (4-656) and (4-65c) are now homogeneous. Furthermore, the heat conduction problem defined by Eqs. (4-65) is sufficiently general to include numerous other combinations of boundary condition as special cases. ‘By separating the variables as defined by Eq, (4-38), it can be shown that the appropriate eigenvalue problem is given by PHO) 5 20) =0 index ck (4-660) KO Liye) a0 atx (4-660) then KS eye) 0 atx (4-660)“TRANSIENT CONDUCTION AND USE OF TEMPERATURE CHARTS. 143 and the solution forthe separated function (again an exponential inthe form ‘The heat conduction problem (4-65) can be solved by following a procedure 0, the boundary surface at x = 0 is kept insulated and the boundary surface at x = L is maintained at constant144 ear Tansee Table 4-2 Eigenfunctions yA, x), eigenvalues 4, and the normalization integral ‘V of the eigen- value problem PHO 4 y= 0 nosest subject to boundary conditions at x = 0 and x = L as shown Boundary condition a Eeanbet ne : won toot mn a eo? Li veo sinks sinit = A “ 2 2 veo Heo sine A . at +) | fare sin deol" Ss A dea eo ce ? , Besa “ % 32 #0 con ix a deci 04 aa +m) ‘ cour esate peeeeeoLt cm Tats H « aa) 1 Hep ao sin = rece e | ee fay u-» We yeh “ # L an at +17) s -ttsweo Hao cout htt eet aye) cont Hsindx uni= mine demas inde walla ESRF 1H defined ax H = Wik is an eigenvalue or this case ‘temperature T.,. Develop an expression for the transient-temperature distri- bution in the slab SoLuTion The boundary condition at x = L also can be made homogeneous by defining a new temperature O(x,¢) as Ox, 1) = TEx, 1) = To. "TRANSIENT CONDUCTION AND USE OF TEMPERATURE CHARTS 145 ‘Then the mathematical formulation ofthis heat conduction problem becomes 20,0) _ 1.20(e.9) 5 1 ts in00 e700) This problem is a special case of the general problem (4-65). Therefore, its solution can be obtained immediately from Eq, (4-67) by utilizing the results in Table 4-2. The problem (4-70) corresponds to case 4 in this table, because it has boundary condition of the second kind at x = O and of the first kind at B x = L Therefore, we obtain Wess) = conde wie, where the 4, are the roots of cos AL _Qr= in eee am and the normalization integral seven by 12 R7L (A7le) These results are introduced into Eq, (467), and the inital condition is taken as FG) = 0, We find 2 Fe cos Ax feos ax axe x0 ‘The integral is evaluated as 1m [loon Ax ay = Fin sin 4b = Lt sin [2 -1) 5 oe 5 ayia cos Ax (470) Qx— ie where 42S (72)146 ear TRANSFER Determination of Eigenvalues {In the previous examples we chose problems for which the determination of the ‘eigenvalues 4, was a simple matter. However, in many other situations the de- termination ofthe eigenvalues is not so straightforward, Consider, for example, case 6 in Table 4-2. The eigenvalues A, are given as the roots of the transcendental equation Atan AL =H an h wi Het here i Clearly, itis not possible to obtain an explicit solution for the eigenvalues A, for thiscase. However, the transcendental equation (4-73) can be solved by a numerical scheme, as discussed in Refs. 12 and 13, and the A, can be determined, The results, ‘ko can be presented in tabular form if Eq, (4-73) i rearranged as Bran =e 474) where f= AL and c= HL = hL/k. The first six roots of this transcendental equation are tabulated in Table D-2 from c = Oto ¢ = ee. Case 4 in Table 4-2 gives the eigenvalues 2, as the roots of the transcendental equation Acot AL = -H (475) ‘This equation also can be rearranged in the form foot p= -¢ (4-76) where HL = hL/k. The first six roots of this transcendental ‘equation are tabulated in Table D-3. PROBLEMS amped system anayss 41 A hot metal block initally at a uniform temperature 7, is suddenly immersed in a welstired. old liquid bath which is maintained at a uniform temperature T, The bet iransfercoeficent between the block and th qui ish Wi(s?-°C) The metal block has a weight A, surface area 4 mand specie heat c,J(kp-°C). Assuming thatthe lumped system analysis applicable develop an ex- ‘resson forthe determination of he temperature T() ofthe book as. fasion of ime 142 AtolidcoppersphereofiO 0 two of the boundary suraces dre Kept insulated, two are subjected to wiform eating at rate gy Wie, and the remaining two 4 {toes dissipate beat by convection ito an ambient at T= 20°C witha heat transfer coefficient h~ 30 ‘Wik! "C).Assuing the lumped system analysis applicable, develop an expression forthe temper: ture T()ofthe aluminum block at faction otime. Caeat the equibrivm temperatre ofthe Bock for a = 10000 Wim? (412 Consider large plate of thickness L niialy a a uniform temperature T,, Suddenly one of surtace is exposed to convection wih an ambient a temperature, witha heat rans coefcen‘The other surface exchanges heat by cadiation with an ambient at temperature T; which regarded as A blackbody. Applying the limped system analysis develop anexpesion forthe temperature Tin the Slab a function of te, 1418 A thermocouple junction may be approximated asa sphere of diameter D = 2 mm, with k = 30 Wil: °O), p = 8600 kg/m’, and cy ~ 08 Khe “C) The heat tanfer coefiiem between the gas stream andthe junction ih 280 W/(m*-“C). How log wil take fr the thermocouple o record ‘98 percent ofthe applied temperature diference? “anoner: Bs 414A thermocouple isto be used to measure the temperatut na gsream, The junction may be approximated asa sphere having thermal conductivity k~ 25 Wkm=“C). p~ S400 kei and cy = (04 Wg"). The heat transfer coefcient between the junction ad the gas sven is b= 360 ‘W/m ©) Caleulatethe diameter ofthe junction the thermocouple should measure9S percent ofthe applied temperature dilerence i 3 418 A 2cm-thick copper plae[p = 895¢kg/m?.cy = O.3830KI (kg “C)-andk = 6Whin-°C)is IntalyatT, = 25°C Fort > Oitisubjeted oa heat fxg = S000 Ws tone o ts stlace andi ‘cooled by convection fom the other surface wth a eat tanslrcoeficienth = 200 Wim?) into ‘an ambient at 7, = 25°C. Using the lumped sstem analysis, develop an expresion forthe transient temperature T() inthe plate 416 A S.cmdiameter aluminum sphere [k = 208 Wim-°C), p= 2700 im", and <, ~ 0896 °C) is iiialy at Ty = 175°: Is uddenly immersed ina mellstired Buid at T= 25°C. The temperature ofthe sphere i lowered to T() = 100°C in ¢ = 42s, Calelate the het iransler coon “Answers = 2003 Wie? °C) ‘Use of eran compete charts forthe sla, cylinder, and phere 417 AO.O.mthick brick wall [2'= 05 x 10-* mis, = 0.69 Witm-"C), and p = 2300kgm?) is ‘nally tT, = 230°C. The wall suddenly expe toa convective environment at Ts = 30°C with heat transfer eoefcient b= 60 Wim? °C) By using the ransientsemperatre chars, determine: (@) The center temperature at 2 and 4h after exporare to the cooler ambiat (6) The surface temperature at | and 2k (6) Energy emoved trom the plate per square meer daring }h 4418 A Jong $-cm-iameterchromestel rod [2= 1.1 10° mand k= 40 WAm- "Cs iially ‘tauniform temperature 7, = 25°C Iissuddenly expose toa convectineenvironment at T= 25°C witha surface heat transfer coefficient h = 50 Wim’). By using the rasin! temperature cher, 10." mifeandk-- 068 Wim "C)]}. Determine thetimerequiredfor the center temperatore ofthe potato to reach 98°C and the energy translred tothe potato during this time “dnowor: 33 min: 318K ‘424 A solid brass aphere[2 = 18 « 10°? mifsandk = 61W)(m-°C)] ofdiameter D = 25emisinitlly Sr, = Tao Wis cooked with an arream at. = 15°C, The heat ransfer coefficient between the Iacam and the solace f= 500 Wim“). How lng wl i take forthe centr of he sphere © ‘lotto 30°C? What Faction ofthe inl energy wil be removed during this ime? (625 Aster pate[e = 1.2» 10°? and k = 42 Wie. °C)] of thickness 2L = 10cm, nity at wiform Temperature of 20°C is suddenly immerse nan oi bath at T, ~ 40°C. The convertion heat transfer ‘eficent between the id andthe surface ih = 600 Wi? °C). Hox long wilt ake forthe center ‘an to cool to 100°C? What action of then energy removed during his time? {$26 A solid aluminum sphere [2 ~ 84 10"S?)sand k = 204W/on-°C)] of diameter D = Demis intatly at T= 250°C Suddenly immersed a wells bath at T= BOC, The heat tranler ‘Coefcnt between he Bud andthe slack = 1000 Wm °C). How long wilt take forthe center ‘ofthe apere to coo to 100°C? “Annwer: 8045 4.27 Consider ala of thickness 10cm aylnde of diameter 10.cm, and a sphere of diameter 10cm, ‘Ek mace of sect [2 = 16-2 10? m'lsand k-= 61 Wii") a inal at uniform temperature 2 Soe Suddeny they ae all immersed into a welltred bath at = SO°C. The heat raster Cooicient between the surface and the id ish ~ 1000 Wis °C). Cau the time required forthe ‘eters of ab cyiner and sphere to colt 80°C, “Anower 547,266 and 18 3 Use of tramient-temperatre charts fora semb-nfiite medlam 428 A thick concrete sab (2 = 7» 10-7 minty at a unilorn temperature J, = 60°C. One its surtaces is suddenly lowered to 10°C By treating this asa oneimensional transient het con ‘duction problem in asem-infnite medium. determine the temperataes at depts Sand 10cm from the ‘surface 0 min after he trace temperature lowered, 4429 A thick stainless slab [2 = U6 5 10°? ms and k = 61 Win CI) inaly a ior empetature T= 150°C I sua rodent Lowered to 20°C: By treating this as a one-dimensional ttatsen eat condection problem maser-infinite medium, determine the temperatoreatadepth2em fom the surface and the heat fat hesurace | min afer the surface temperature islowered Anomers SC; — 144 Wie 430 A thick copper slab [2 = [1 = 10°4m/and = 380W)en °C) intially at a uifrm tem erature 10°C Suddenly the surtice sand 1 100°C. Calculate the het ux at the sae Sand 10 Ii afer he raising of the surface emperatare How log wll take the temperature a depth Sem from the srface to reach 90°C? “£31 A thick bronze (2 = 086 x 10°? mfsand k= 26 Wim-°C)] is intl at a ifom temperature 350°C Suenl the suace exposed toa coolant 25°C. Assuming thatthe heat transfer coficient for convection between the id tod the eine i 150 W/m? °C), determine te temperature Sem from the surlace 10 min alter the exposure ‘Answers 205°C190 wear TRANSHER 432 A thick wood wall [2 = 082 x 10°" ms and & = 0.15 WAim:°C)] i intially ata uniform {emperaturof 20°C. The wood may igiteat 40°C I the prac sexposed toot aver at T= SOC 'and the heat transfer oeficien! between the gasand the surface is = 4 W.-C), bow long wil it {ake far the surface ofthe mood to reac 40°C? ‘£39 A thick wood wall [x = 08 10° m/sand k ~ 0,15 W/im-°C)] is inal a wnitorm tm. petatureT; = 20°C Suddenly the surfaces ase 120°C. Calculate the temperature 2em fom the Surface at 2 and 20min afer the exposure Answer 20°C: 36°C 4434 A thick concrete slab 4 = 7 < 10-7 mp and k = 137 Wm-“C)] is initially a a nif temperature T, = 40°C. Sudenly its surfaces subjected to convective cooling with & heat teenfet coeicient h = 100 W)(m?-°C) into an ambient at T= 40°C. Caleulte te temperature 10cm from Thesurface 1 after he star of cooling Product soltion 435 A rectangular aluminum bar 6m by 3cm [k = 200 Wim °C) cy = 890,/g-°C), p = 2700 Kain’ and 2 = 84 x 10-* m5} is tially ata uniform temperature T= 175%: Sualy the surfaces are subjected vo convective cooling with a hea ransercoricent h = 250 Whim? =C) ints fan ambient at T_ = 25°C as shown in Fig. P35, Determine the centr temperature T, of the bar (= T min ater the start ofthe cooling Gomecton Figure PAS Avene: 075°C 4.6 A shor, cylindrical aluminum bar U D = Gemand height L aresubjcted to comect aT, (on-°C) and 2 = 84% 10°? m')) of diameter msn at nif temperature T= 175°C. Suddenly the surfaces cooing witha heat transfer copthevet f= 250 Wim? °C) into an ambient 28°C Calevlatethecenteremperature Ty the eylnder ~ I min fer the wart ofthe cooling Anower BPC 4437 A brick column wih crossscton 1Dcmby IOem[k = 0.69 Wij(mn"C)anda = 05 x 10-+m)s] is inal a «uniform temperature T, = 725°C. Suddenly the sutacs are subjected to convectig 0 the boundary surface SA ciskept insulated and that at x ~ Liskeptat constant temperature ,.The thermal properties tte slabarecontan and thereisno heat generation within the medio. “@) Develop an expression forthe one-dimensional time-dependent temperate dstibution Tea) inthe slab (@) Develop an expresio forthe temperature ofthe inated srfce, {Develop an expression forthe hea flux at the boundary surface x ~ L “451 A brick wall confined tothe region = 1. 0, Both boundary suraces are maintained at «uniform tempera- ture Ty i Figure PAST (Develop an expression forthe time-dependent temperature dsribution T(x) inthe sab (Unt; Compare this problem with Problem 430) () Compute the temperature atthe center plane for = Sem, k= 1 Wiim-°C), 2 o-tm BMPC Ty = WCatt jh and 3 afer he exposure to cooling 452 Develop an expresion for the onedimensonal, time-dependent temperature distibation ‘Tex OforasabinO = + 2 £,whichisiniallyat temperature T = Fls)andlfort > Oboth boundaries are maintained a 22 temperature, sn Fig. PAS? osx Figure Past (6) Calculate the center temperature forthe seca case of ntl dsributon given by FC) = 100 sin (ex/L) by taking L = 10em,a = 03 % 10-* mand k= 1 W.km-°C)at = 35 Land 2h after the exposure to cooling (6) Calelate the het xcs athe boundary sutaces x = forthe case considered i (0) and mate = fh feand 28154 ear Teansren 459 A slabof thickness Lhasa intial temperature distribution 7, ~ Ty os [es(2L)) For > Othe boundary surface atx = Oiskeptinslated nd tht atx Lis kept at Sere temperal we (@) Develop an expression forthe onedieasionl time-dependent temperature distribution Te inthe sab, (@) Calculate the temperature ofthe insulted houndary att = jy, 1, and 2 for ress L = 10m, and thermal ditsiviy @ ~ 10-" m/s by taking = 200°C 454 A slab in 0.< x < L has an initial temperature distribution T = F(x) For > 0, the boundary x Oinkept at zero temperature and tht xx = Liskep insulted, a alustatedin Fig Pes plate of thick Figure Past {«) Develop a0 expression forthe one-dimensional, time-dependent temperature distribution Tex ia the sab, (@) Obtain solution forthe speci cse of i temperature distibution given by sin (6) Caleulte the temperature of he insulated surface forthe casein (bby sting Ty = 20°C {= 1Ocmand a = 10-7 m?/sats = fy, and 2 hater the start of colin Answer () 1982, 183 and 167°C 45 A large plate of thickness is intlly at a uniform temperatre 7, At = 0, the temperatures a both surfaces are suddenly lowered toa temperature. () Develo an expression fr the one-dimensional, time-dependent temperature distribution Tea, in the slab (©) Catelat the comer temperature at ¢ T= 250, and 7, = PC. ‘Answer: (9) 250 and 211°C 456 A large plat of thickness Lisintally at a prescribed temperature Fla). Fort > 0 the boundary surface at x= 0 suddenly lowered to zero temperature, andthe boundary suttace atx = Lis Rep insulated (a) Develop an expression forthe one-dimensional, time-dependent temperature dstibution Tex) in the sab, (0) Develop an expression forthe heat fox a the boundary surface atx = 0. “457 A lage plate of thickness Lis iniily at a uniform temperature 7, Fort > 0 the boundary at kept insulated, and the boundary surace atx ~ Li disipating beat by convection ithe Ambient ar at 7, with aheat transfer coeticient h The thermal properties ofthe aubarcconant, (@) Develop an expression for the one-dimensional, time-dependent temperature distribution Tex )in the sab. (0) Develop an expression forthe temperature ofthe insulated surface atx = 0 (6) Develop an expression fo the het fa a the boundary surface Fe) fh and 1h for L-= 20¢m, «= 05 x 10°* mip, REFERENCES 1. Canlaw, H.S. and J.C. Jaeger: Conduction of Heat in Solids, 24 e., Oxford University Pres London, 1959.be Pa, 1958 ‘Arpac V,S.: Conduction Heat Troyer, Addison-Wesley, Reading, Mas, 1966. Mubnaios, M..Deand M,N. Oui United anaes and Soluions of Heat and Mass Dison, ‘Wiley, New York, 1984, Oaink, M.N.: Heat Conduction, Wiley, New Vork, 1980. Howley M.P Temperature Chats for Induction and Constant Temperate Heating.” Tran ASME, 6227-25 (194. Schneider, . J: Temperature Response Chris, Wiley, New York, 1963 Goober HS BAL aod U. Grigll: Findamentale of Heat Transfer, McGraw-Hill, New York, 161. ‘Schneider, P.1: Conduction Heat Transfer, Aason- Wesley, Reading, Mass. 1955, ‘Sonderand J. and K_R, Johnson: "Shape Factors for Heat Conduction trough Bodies with Itothermal ot Conveetie Boundary Conditions,” Tans. ASHIRAE, 70:237-241 (1960 Tene, sad U, Origa. "Farfakior und Foomwiderstand der Stationdren Mehrdimen- ‘Sonalen Warmeltung. nt. J. Heat Mass Transfer, 18751-76197) Jamar M.LG.M. Smith, and J. C, Wolford: Applied Numerical Methods fr Digital Comput tation wth Fortron and CSMP, 24 ed, TEP, New York, 1977 Groove, W.E Bil Numerical Mothds, Prentice Hall, Englewood Clif, NJ. 1966.CHAPTER ~ FIVE FINITE-DIFFERENCE METHODS FOR SOLVING HEAT CONDUCTION PROBLEMS. The numerical method of solution is used extensively in practical applications to determine the temperature distribution and heat flow in solids having com- plicated geometries, boundary conditions, and temperature-dependent thermal Properties. A commonly used numerical scheme is the finite-diference method, ‘described in Refs. 1 to 10. In this approach, the partial differential equation of heat conduction is approximated by a set of algebraic equations for temperature at a ‘number of nodal points over the region. Therefore, the first step in the analysis is the finite-diflerence representation, or the transformation into a set of algebraic ‘equations, ofthe differential equation of heat conduction. This matter is discussed in this chapter, fist, with emphasis on the finite-difference representation of one= dimensional, steady-state heat conduction problems in rectangular, cylindrical, and spherical coordinates; and, second, by the finite-difference representation of the ‘two-dimensional, steady-state heat conduction problems in rectangular coordin« ates, Since the method transforms the analysis of the heat conduction problem to the solution ofa set of coupled algebraic equations, we briefly discuss the methods of solving simultaneous algebraic equations. Finally, the solution of one-di ‘mensional, time-dependent heat conduction problems by finite differences is described. In recent years, another numerical scheme called the finite-element method, developed originally for the solution of structural problems, has been applied to the solution of heat conduction problems. For problems with complex geo- ‘metres, the finite-clement method offers some advantage over the finite-difference ‘method in the solution of heat conduction problems. The application of this method is described in Refs. 11 to 16. In this approach, either the variational principles or, preferably, the Galerkin method is used to transform the heat conduction problem to a set of algebraic equations; hence some background is needed in the use of these methods as well as the selection of finite elements, Therefore, in the limited space available here, this method is not discussed. 186IMITE-DIFFERENCE METHODS FOR SOLVING HEAT CONDUCTION PROBLEMS. 157 5:1 ONE-DIMENSIONAL, STEADY-STATE HEAT CONDUCTION When a heat conduction problem is solved exactly by an analytical method, the ‘resulting solution satisfies the governing diflerential equation at every point in the region as well as at the boundaries. However, when the problem is solved by a ‘numerical method, such as finite differences, the differential equation is satisfied only ata selected number of diserete nodes within the region. Therefore, the starting point in an analysis by the fnite-diflerence method is the finite-ifference repre- Sentation of the heat conduction equation and its boundary conditions. “The finite-difference form of the heat conduction equation can be developed by replacing the partial derivatives of temperature in the heat conduction equation fwith their equivalent finite difference forms or by writing an energy halance for a Aifferential volume element. In this book, we develop the finite-difference equations by writing an energy balance. However, its instructive to use the former approach to illustrate the relation between the heat conduction equation and its finite- difference form. Therefore, we also illustrate the use of this approach in order to provide a better understanding of the finite-difference formulations. Finite-Difference Forms from the Differential Equation Consider the following one-dimensional, steady-state heat conduction equation with energy generation FO 4 hae ind 40: In addition othe change made above, epic in ines 24nd 3 the vale of 40 by the new ‘alu of Nao ine 2 replace the vale of 160 by vale at leat equal to N°‘Table 51 Computer program for solving a system of N simaltancous equations applied to the solution of Example 5-1 108 ' IMPLICIT REAL#8 (A.02) 2 REAL#t A(400).(40,1).WRAREA(1600) 3 NDIMEN= 0 ‘ wer-0 © READ THE MATRICES (A) AND [B] BY ROWS 5 6 3 . DOW 1=1.N 0 WRITEG,70(AG.1)=1,N),8(10) M70 FORMATAX/TIO,13F93) 12 @ CONTINUE a (CALL LEQTIF(A.LNNDIMEN. c IDGT.WKAREAIER) WHEN RETURNED, THE MATRIX [B] CONTAINS THE SOLUTION. © THE SOLUTION IS NOW PRINTED. c WRITED.80) 1880 FORMAT(IX//T30/NODE #”, 5X,TEMPERATUREY) 6 O90 "7 ‘WRITEG. 100, BAL.) W100 FORMATUX.T28,5.7X.F9.3) 19% CONTINUE » Stop 21 END =200 1.000 0000 0.000.000 66.000 1.000 2000 1.000 0000.00 — 16.000, 000 1000-2000 1.000.090 — 16.000 0000 0.000 1000-2000 1.000 — 16.000 0000 om oom) 2000-2044 20.440 NODE# TEMPERATURE 1 119.085 | 2 i700 3 209135 4 20.180 5 ass “The exact analytic soltion ofthe heat conduction problem in Example tia | straightforward and yields the following expression forthe temperature distibu- ro)=02s£20(0-3)()166 next manseen ‘We compare the finite-difleence solution using five nodes and the exact solution ofthe problem; the results are very close as shown below. sik Tydiference Exact 02 7 11905 1900 04 7 20 © 1200 0s T2818 290 of Bois 23000 oT, 2522 23s00 In this particular problem, we have the exact solution to check the accuracy of the numerical solution. When no exact solution is available, the convergence of the numerical solution should be checked by performing further compu- tations using finer subdivisions, until the solutions from subsequent compu- tations satisly a specified convergence criterion. Example 5-2 An iron rod L = $ cm long of diameter D = 2 cm with thermal conductivity k = 50 W/(m °C) protrudes from a wall and is exposed to an ambient at T,, = 20°C and h = 100 W/(m? -°C). The base of the rod is at To = 320°C, and its tip is insulated. Assuming one-dimensional steady-state heat flow, calculate the temperature distribution along the rod and the rate of heat loss into the ambient by using finite differences. Compare the finite- difference solution with the exact analytical solution ofthis problem. SOLUTION This problem is exactly like the fin problem considered in Chap. 3, ‘and its mathematical formulation is given by [see Eq. (3-54) for m = N}] AO — N2H3)=0 ind 2hb) | Bg, 2hb | 2b or a+r (14 Bnet re Pe Example 5-4 Consider steady-state heat conduction in a rectangular region 0s x < 36,0 < y < 2, subjected to the boundary conditions shown in Fi Figure $12 The geomeiryand houndaryon- Aitons for Example 5174 ear mance 5-12. Calculate the temperatures Tym = 1 to 6, at the six nodes shown in this figure, and compare the finite-diffetence solution with the exact results SOLUTION The finite-difference equations for each node are Node 1 27; +21; ~ 47, =0 Node 2 i +27, + 100-47, = 0 Node 3 Ti 42+ G44 =0 Node 4 T+ Ty + Ty + 8666 — 47, =0 Node 5: T+ 2% ~4T=0 Node 6: T+ T+ 0-47 =0 ‘These equations can be arranged in matrix form as follows: 4 202 0 0 off, 0 1-4 0 2 0 off] |-100 1 0-4 2 1 off}_]| o 0 1 1-4 0 aff r=] 86066 o 0 1 0 ~4 aflr 0 o 0 o 4 4 ~4lfn} | -s0 ‘The solution of this matrix equation gives the six node temperatures T., m= 1106, The exact analytic solution of this heat conduction problem for the temperature distribution within the region is, Spent = (s*) ‘We now compare the finite-difference solution with the exact results: T(x, ») = 100) Fine Te iference Exact T 86 22 rosa m @s % 38 ae)[FINITE DIFFERENCE METHODS FOR SOLVING HEAT CONDUCTION PROBLEMS 175 oe Fire 5.13 Boundary conditions a and the finteiflerence network Acinanctountay. 300 for Example 55 The accuracy of the finite-diflerence solution can be improved if finer mesh is used. Example 55 Consider a rectangular region subdivided with a square mesh of side and subjected to the boundary conditions shown in Fig. -13. There is ‘energy generation in the medium at a constant rate of g W/m’. Write the finite-diference equations for the 10 node temperatures Ty, m= 1 to 10. Express these equations in matrix form, SoLUTION The finite-dfference equation for the internal nodes is immediately ‘written by utilizing Eq, (5-25), The same equation also can be used for a node ‘on the insulated boundary if proper adjustment is made for the symmetry requirement. An energy balance equation is readily written for a node on the convection boundary. We summarize the resulting 10 equations: Node ‘umber Fntediference equation 2, + 21 — 47, + =o i Mea eetee re ear gees my (alte, 8 a eany(40 Mn Mee wane nen (6+ ws (A176 ext reanseen ‘These 10 equations can be expressed in matrix form as a A) a2) | | -64 || To -G-HT., ~G-HT.- fy where we have defined 2ht E ‘Once the numerical values of the parameters G and H are specified, these equations can be solved for the node temperatures. G wo, 7 and It is now apparent that in the finite-dfference approach, the heat conduction ‘equation is replaced by a set of simultaneous algebraic equations for temperatures at the nodal points of a network constructed over the region. These equations cean be solved readily with a high-speed digital computer. The larger the number of nodes, the closer the finite-difference approximation to the exact results; but larger number of nodes requires more equations, hence more computer time and may introduce some roundoff errors. However, in most practical applications, it ‘may not be necessary to choose a large number of nodes. The results obtained with 1 small number of nodes may be sufficiently accurate in view of the uncertainties in the determination of thermal properties of the material and the heat transfer coefficient at the boundaries.[ANITE-DIFFERENCE METHODS FOR SOLVING HEAT CONDUCTION PROBLEMS 177 5-4 METHODS OF SOLVING SIMULTANEOUS ALGEBRAIC EQUATIONS ‘The analytic solution of simultaneous algebraic equations of order greater than 3 is quite tedious, and itis quite prohibitive for orders such as those encountered in the finite-difference analysis of heat conduction problems, Therefore, various ‘numerical schemes have been developed to solve simultaneous algebraic equations by a digital computer. Table 5-1 shows a computer program calling a computer subroutine that utilizes the gaussian elimination method: Here we discuss the basic features of the gaussian climination method as well as other methods utilizing an iterative scheme and the matrix inversion. Gaussian Elimination Method ‘The finite-diflerence formulation of heat conduction problems, as discussed previously, leads to a system of algebraic equations that form a banded matrix. ‘The gaussian elimination method is very efficient for solving such systems of algebraic equations. In this approach, the matrix is transformed to an upper diagonal form. To illustrate the basis of the gaussian elimination process, we consider a system of algebraic equations forming a banded matrix as shown in Eqs. (5-30). fain a2 a 0 0 OTT) [Cy Jas, 22 ay ae 0 te} cs 0 ayy 4s aa Gel = o 0 0 Gyn 2 Onn t Oma) LTr ‘The matrix in this system can be transformed to upper diagonal form. The first equation is used to eliminate the nonzero elements a;, and a, in the first column. ‘That is, the first equation is multiplied by a;,/a,, and the resulting equation is subtracted from the second equation in order to eliminate a,,; then the frst equation is utilized in a similar manner to eliminate a,,. Next the second equation is used to eliminaze a, and a,2. The third equation is used to eliminate ay, and so forth. When this process is carried out to the last equation, the resulting system {forms an upper diagonal matrix, as shown in Eqs. (5-31). at at at, 0 ae oyfn ct 0 af as at . olla ct 0 0 at, ate q a 0 0 0 ak + : «IP n]=| ct}os ° ° chy (00) 0) 0 co178 war manne Once the upper diagonal form is obtained, the solution follows immediatly. That is, the last equation in the system (5-31) immediately gives T,. Given 7, the (n — 1)st equation gives T,_ , ;and given T,_,, the(n — 2)d equation gives T,_ 2; and so forth. The calculations are carried out unl 7, is determined from the first, equation. ‘When the numberof equations tobe solved is very lage, the roundoff errors can accumulate and sometimes cause a deterioration in accuracy. Rearranging the equations so that the coefficients which are largest in magnitude are put onthe ‘main diagonal may improve the accuracy. ‘Various computer subroutines are available for solving simultaneous linear algebraic equations by the gaussian elimination process or by the methods which are some variation of it. The International Mathematical and Statistical Libraries [18] gives numerous such subroutines. For example, LEQTIF and LEQTIB are typical subroutines for solving systems of algebraic equations, while LEQT2F and LEQT2B are the high-accuracy solution versions of these subroutines, re- spectively. Gauss-Seidel Iteration ‘When the number of equations is very large, the matrix is not sparse, and the computer storage is critical, an iterative technique frequently is preferred for the solution of such a system, The Gauss-Seidel iteration (often called the Liebmann, iteration) is one of the most efficient procedures for solving large systems of equations. The procedure involves the following steps: (1) Solve each equation for one of the unknowns. (2) Make initial guesses for all unknowns, and compute the unknowns from the equations developed in step 1 (use the most recently ‘computed values for the unknowns in each equation). (3) Repeat the procedure ntl a specified convergence criterion is satisfied ‘To illusteate the application ofthe Gauss-Seidel iteration proces, we consider the following three simultaneous algebraic equations for the three unknowns Ty Ty,and Ty: 4yT tanh bash =h (S-32a) 42, T, + G22 Ty + 4357s = fy (5-32b) 47; + an2Ts + nT = fy (220) “The first equation is solved for T, the second for T, andthe thied for T: = 43Ts (5-33a) tr (6338) fy ansT, ~ ay 7, = Soon — anh (6-330)‘INIT. DIFFERENCE METHODS FOR SOLVING HEAT CONDUCTION PROBLEMS. 179 Let T?, TY, and T" be the initial guesses for the temperatures. These values are introduced into Eq. (5-33a) to obtain a first approximation for 7, as =a TP ~ aT? nm (5-34a) ‘This most recent value of TY? and the initial guesses are introduced into Eq (5-336) to obtain 7)": Spa anTi ~ aT? re (5-340) By using the initial guess or the most recent values of temperature in Eq. (S-33e), a first approximation TS? is obtained for temperature 7, The above procedute is repeated until a specified convergence criterion is satisfied, The convergence criterion may be specified as Tr) TM ce (5:35) which should be satisfied for all 7. ‘Another convergence criterion may be specified in the form re 7p a0 ¥ layl 631) shies where N is the total number of equations in the system. Weillustrated the application ofthe iterative method of solution with a specific ‘example given by Eqs. (5-32). We now illustrate its application to the solution of amore general system of finite-difference equations (5-27). We solve Eqs. (5-27) for the temperature T, AV, dp + Ys (T/R yp) LGR, Where p = 1,2,..., N and N isthe total number of nodes in the region. “The general procedure for iterative method of solving Eq, (5-38) i as follows: (5-38) 1. An initial guess is made for the node temperature Tp = 1,2,...,N. 2. Afirst approximation for the node temperatures T?iscalculated, by introducing. the intial guess values o,ifavailable, the most recent values of the node tempera- tures on the right-hand side of Eq, (5-38) 3. The procedure is repeated until a convergence criterion, such as that given by Eq, (5-35) or (5-36, is satisfied. Matrix Inversion ‘The matrix inversion is another method that can be used for solving simultaneous algebraic equations. This method, like the gaussian elimination method, provides fan exact solution to the unknowns, instead of approaching the exact results ‘gradually by iteration, as in the Gauss-Seidel iteration. Consider, for example, the ‘matrix equations (5-30) written compactly in the form [antr] = (8) (539) where [A] is the coefficient matrix, (7']is the vector for the unknown temperatures, and [B] is the known coefficient vector whose elements involve contributions due to energy generation and/or the boundary conditions forthe problem. The solution of the problem by matrix inversion is given as (r= (ay) (5-40) where [A] i the inverse of matrix [4], The matrix inversion can be performed readily by using the standard computer subroutine for matrix inversion, Accuracy of Finite-Difference Solution In the numerical examples in this chapter, we assessed the accuracy of the finit difference solution by comparing it with the exact analytic solutions. Therefore, is available, the numerical solution should be compared with an analytical solution {or a similar problem. We stated that performing the calculations with smaller values of mesh size would improve the accuracy of the finite-ifference solution, However, there isa limit to the use of very small mesh size, because an increase in the number of nodes increases the number of machine calculations, which in turn increases the computational roundoff errors. Therefore, one should start the finite-diference calculations with a coarse mesh, gradually refining the mesh size and observing the convergence of the solution before choosing the mesh size to perform the final computations.[INITE-DIFFERENCE METHO0S FOR SOLVING HEAT CONDUCTION PROBLEMS 181 Finer mesh size is needed in the regions where the boundary conditions exhibit steep temperature variations and where heat flow rates are to be determined. ‘An accurate determination of the heat flow rate requires the use of finer mesh size. 55 CYLINDRICAL AND SPHERICAL SYMMETRY ‘The formal representation of the finite-difference formulation given by Eq. (5-27) is now applied to develop the specific relations for the finite difference of the one- ‘dimensional, steady-state heat conduction in cylindrical and spherical coordinates. Cylindrical Symmetry Consider radial heat conduction in a long, solid cylinder of radius r = b in which energy is generated at arate of g(r) W/m?. The region 0 4 results in an unstable solution. ‘There are various mathematical techniques for the determination of the stability criterion associated with the finite-difference representation of the time- dependent heat conduction equation. It is instructive to determine this stability criterion by the following physical argument. ‘Suppose at any time step i the temperatures T',_, and T,,, at nodes m ~ 1 and m + 1 areequal but less than T', at node m between them. Then ifthe value of r ‘exceeds }, the coefficient 1 — 2r becomes negative. Then, according to the finite- difference equation (5-59), for 1 — 2r negative, the temperature Ti"! at node m fat the next time step should be less than that at the neighboring two nodes. This is not possible thermodynamically, since we assumed that T", was higher than that at the neighboring nodes. Therefore, to obtain meaningful solutions from Eq. (5-59), the coefficient 1 — 2r of T', should not be negative; that is, 1-220 of rsh which isthe stability criterion given by Eq. (5-61). ‘Two-Dimensional Heat Conduction ‘We now generalize the finite-diflerence procedure to the finite- 0 the boundary surfaces at x =O and x = L are subjected to convection. The mathematical formulation of this problem ‘can be taken as or er Faate indcx0[FINITE DIFFERENCE LATHODS FOR SOLVING HEAT CONDUCTION PROBLEMS 197 sujet tottus condos or en ro ry hyrehatn ane bio ‘and the initial condition T=FQ) forr=0,in00 ar ao | atx=O1>0 T=0 ax=Lt>0 T=200 forr=0 ‘This problem is a special case ofthat in Example 5-8; hence its finite-difference formulation is immediately obtained from that example by setting hy = 0, hy = 2, a0d Tay = 0: TH) = eT += 2TR + Tey me Le M1 Tete" for node m = 0 Tet for node m= M Ta = 200 fori = 0,m To solve this problem, we divide the region into M Hence,The value of the parameter r is taken as step Ar becomes 45 then the corresponding time 1(04 x 10-7)? 2104s, by ting = he toe ote diene tn toes Te = MTS + Tre) m= 1234 ae ° ms 0 ma O2a4si=0 Clearly, the temperature Ti! at an interior node m at time step i +1 is ‘equal to the arithmetic average of the two adjacent node temperatures at the previous time step i. For illustration purposes, the numerical computations are performed by hand calculations, and the results for 10 consecutive time steps are listed in Table 5-2. A programmable calculator or a digital computer can readily be used to perform such calculations. In Table 5-2, the first row, i = 0, is the initial temperature distribution, ‘The second, third, etc. rows are the temperatures at the end of At 2.At = 165, and so on. The heat flux at the boundary surface x = L (that is, M determined from Wim? For example, the heat flux gat the end of time step 1 = 10 (ie. ¢ = 80 8) ‘becomes 485-0 04 x 10? 24,250 W/m? ‘Table 5-2 Finite-difference calculations for Example 58 m0 30 2S m0 20200 mm 20100 m0 10100 ms 1075 fo 162 8s 191 194 as 91 17e S47 aa 969539 BS 2 8S & eccccoccced200 mar TRANSFER “The exact analytic solution of this problem for the temperature distribution is Te.) 200 where ‘Then the heat flux at the boundary surface x = L becomes aT) 400k Say {L.) = ~k tt The exact value ofthe heat flux atthe boundary surface x = L att = 805 is esas = 24890 W/m? ‘The finite-dfference solution obtained by using a rather coarse mesh size ‘compares well with the exact solution. Example 5-10 Consider two-dimensional, time-dependent heat conduction for the region shown in the accompanying figure. The boundary conditions are ree AT eo Whm-"C)stated in this figure, and a very coarse network of mesh size Ax = Ay = Lem is used for finite-diflerence formulation. Develop the finite-difference equations for each of the five nodes, m = 1 to 5, and establish the stability criterion for the resulting system. Takeh = 60 Wim? -°C),k = 30 Wim -°C), and T, = 0°C. SOLUTION Since the fnite-diference network for this problem is very coarse, ‘there are only a few nodes for the entire region. Therefore. we need to use a transient energy balance to determine the finite-ifference equations for all the nodes on the boundaries, because their thermal capacitance cannot be neglected. The accompanying figure also shows the volume element associated ‘with each node used for writing the energy balance equation. The results are summarized here: = ET = Pes Ar Solving for 71", we obtain Ti! = (1 rT} + ry + rT + 100r a) kP_aP where ret Ele ey At AF IT-T 4! og BTM = TE ks PhS Te = T= yp Setting T,, = 0, h = 60 W)(m? -°C), k= 30 W)(m °C), and = 001 m and then solving for 7", we obtain Ty! = Ty += 40M )TS + 27S o Node m = 3: Tb + hi, = 73) = wey Substituting the appropriate values for T..,h, and k and then solving for T$**, we obtain aaaaran Tit = UT + (1 - 4087S + TS © Node m = 4: 1100 = T kb i eran ea my + IMT = T= pep 3P2a wear aren Substituting the appropriate values for T.,, h, and k and then solving for Ti, we obtain Ti'= 13337} + 06MTS + (I — 4027)TY + O67FTS +1333 Node m = 5: ITT) Substituting the appropriate values for Th, and k and then solving for 73", we get TH! = Ti + (1 — 4087 + 2007 © Equations (a) to (e) are now expressed in matrix form Te] fina or o Pa o Vr my a Fa ° o fir ry}=] 0% 1-408 o Ir ret] |1333r 0st 1 402% sr | J74 Ts o ° 0 x 140%] [14 100r 0 +| 0 133.37 2000 ‘An examination ofthe elements in the main diagonal of the coefficient matrix reveals that the smallest value for ris associated with 1 ~ 408 = 0 ‘Hence the stability criterion forthe system becomes 1 Oo (578) subject to the boundary conditions or eT mT ef atx O,fore>0 (5-790) iT KE enTafh ax=Lfort>o (5-796) T=F() forr=Oin00 Tot atxver>0 Tat foer=Ginx 20 By using the explicit Onite-dierence scheme, develop a Grite-diferene formulation of this heat ‘conduction problem, and discus the ability ene ‘S41 Conse the following one-dimensional. time-dependent heat conduction problem fora sab of Ahicknes err pat occ ebi>o ar tare rn oro0 sar mir, 0. Tat, atxebeoo T=0 fr=0moexet By using the exp fnte-diference scheme, develop finite ilerece formulation of this hea con- ction problem, and discuss the stability criteria {5442 By using the impli iteiference scheme. develop the ite dference formulation of Probes Sell. and sess the sabiity criteria 5-43 By using the Crank Nicoionfite-ieeace scheme, develop the Bitedierence formulation of Problem $4]. and discus the stability criteria, ‘5.44 By using the explicit Snitediferenceacheme, devel the fnite-diference formulation of the folowing heat conduction problem: indexek are T=h — fort-0"ISITE: DIFFERENCE METHODS FOR SOLVING MEAT CONDUCTION PROBLEMS. 21 5-45 By using the explicit fintedifernce scheme, develop a fatediference formulation for the Following time-dependent heat conduction equation: (7-22) ar rr) 8 Ge Discus the stability teria ‘5-46 Consider one-dimensional, time-dependent heat conduction in sab of thickness L, which is {nally at uniform temperstre Ty For! > Othe boundary surface atx = Ois kept insulated and the ‘boundary surface atx ~ 1-dsipates heat by natural convection and radiation into an ambient which ih ata temperature T, K and can be rparded as blackbody. Hea! wane by radiation and con- ‘ection atthe boundary surface x = [can beclculate rom fan -TS) Ww e-CaT~Toe Ww where eis theemisivityand C, = 192 constant for fice convection. By uinga mesh ze Ax ~ Lit nd the eapict shee develop the fnite-diferene formulation of his heat conduction problem. 5-47 Consider two-dimensional, ime-dependent heat condveton i a region L by L whichis iially ta uniorm temperature Ty and for > Oi subjected tothe boundary conditions shown in Fig PS ‘By using an explicit scheme anda meth Ax = Ay = L/3, develop the fnite-diference formulation ofthis time-dependent heat conduction problem x Radiation nto ambient “amber ne Symnery Figure P57 15.48 Consider two-dimensional tine dependent heat conduction in a region 3cm by Sem, which it {nil a uniform temperature Ty ~ 34°C and for > O's subjected tothe boundary conditions shown in Fig. P5-48, By using am explicit scheme and a mes ie Ax = Ay = 1 em, develop the fnite- Aiference repretentation ofthis hea conduction problem. ot Pore Ye Pea ee Figure PS222 xr mansreR ‘$49 Consider two-dimensional time-ependet heat conduction ina region which i initily at ‘uniform temperature Ty and for ¢ > 0 subjected the boundary oncitionsistated in Fig P59, By using an explicit scheme anda mesh sie Ax = Ay = Lem, develop the fniesiference formation ‘ofthis eat conduction problem, tempest 7, Figure P59 ‘$0 Consider one-dimensional time-dependent heat conduction in a sphere of dius b = 3.cm. in hich heats generated at rate of gy W/m nal, the sphere is ata uniform temperatre Ty and for > Othe boundary surface at r= bis manlained at °C. By using an explict method of the finite Aieence scheme and diving the region int six concentric spheres wth Av = 05 om. develop the fnitedierence formulation of his heat conduction problem. ‘S81 Consider one-dimensional tine dependent heat conduction ins conducting rod of radius r= b, ‘hich inialy tw uniform terapecatreTy-Sudenly het x generated inthe rod at constant ateof 9 Wim by the passage of etre current while the boundary surface tr ~ Bistaintained a tmpers- {re T,.The thermal conductivity ofthe rod remains constant, andthe temperature varcs oly tbe ada) dietion (thats, variation of temperatre inthe = direction i epi), By dividing the od Imo five concetic cylinders with equal increments in radi (that, Ar ~ 19, taking Ae = Tana ‘ing the expt scheme, develop = Brite-diference formulation ofthis one-dimensional, time de Pendent het condaction probes. Unsteady heat conduction solutions ‘$2 A large and very thick brick wall(x = 5 x 10°" m/s) which sini at a uiform temperature T= 125'Cissuddenly exposed o cooling by maintaining tssurlaceat + ~ Oat Ty = 28C.Tocaleulate ‘he temperature transients at depts small iacomparon to thethicknes the wal canbe regarded a 8 ‘semiafnite medium confined to the region x > 0. By using an explicit scheme and a mesh size Ax = (03 cm,cakeulate the temperatore x = 1.2 em rom the wtfacet = tnd Sin afer the exporere ‘$53 Repeat Problem 582 forthe case when the boundary surface atx = O exposed to convection inco an ambien at T, ~ 25°C witha eat wander coticient k= S Wik’ °C). The thermal co ductivity ofthe bik wall ean be taken as = 0.7 Wim-“C). ‘554 A thick concrete wall [x = 8 x 10-7 ms k = 08 Wim °C)] is initialy aa uniform tempera- ‘ure T= SOC. Sudenly it is expooed to a cool airstream at, = $C. The het tranercoeficent between the airstream and the surfaces h-= 10 Wifm= °C) The wal in be regarded asa sm infinite medium confined to the region x > O with the surtace atx ~ O subjected to convection. By ‘sing an explicit nite-dfevence scheme, allt the wal temperature at depth x ~ 10cm from the surface r= 10min and I hater the tar of cooling, by taking Ax ~ 2m, 5 A very thick copper slab [x = 11 « 10°? mij, k= 386 Wi-"C)l initially at 325°C is suddenly exposed to cooling. nd its surface temperatures lowered to Ty ~ 25°C. For shot tines, the ‘copper sla can be ueated sa sem-infte medium confined othe region x > 0, while s boundary at <= Oiskeplat 25°C By usinga meshsie Ax = | cand an expt ite-diference scheme calculate the temperature x = 6 cm from the surface 2 min alte the lowering of the surface temperature $6 A thick copper late[2 = 11 19°? mf & = 386 Wion°C)] intl a uniform temper ‘ure 7) ~ 25°C Suddenly one ofits surfaces is expose o constant heat fax yg = 2 10° Wim ForFANITE-DFFERENCE METHODS FOR SOLVING HEAT CONDUCTION PROBLEMS 23, stort tans the copper pat can be teat 1 2 sembinfite medium confined tothe region x = 0, fected to a constant bet Box a the boundary srface x = O BY usinga mesh size Ax = 2 cmand an ‘aplet intediference scheme, calculate the emperature x ~ 10cm ffom the surface ¢ = $min er the tart of surface heating SST A very thick ickestet pat = 05 10°? mij, =20 Wie °C) inital ata uniform Temperature P= 280°C suddenly exposed toa cool ssream aT, = 30°C. The heat ansfer co: ‘ica berwen the estean andthe surface sk ~ 60 Wm" °C) For shoe tines the plate cam be ttened asa semiinimite medium conned to the region x = 0 with surface at x ~ 0 subjected to Convertive cooling By using» mesh size Ax ~ I cm and an expt ite-diference scheme, calculate the temperature atadepth* = Sematr = 2and 8 minafer thestart of cooling. ‘S58 An hemhick chromestel plate [x = 6 x 10°? mils,k = 61 Wien °C] ily at 2 uni form temperatre T= 125°C suddenly exposed to a col sisream at T= 25°C at both of ts ‘faces Theat anf coeticient between the rand the surface ik = 400 Witm? °C) Byusingan ‘upc Snteciference scheme and mesh sie Ax ~ 1 et, determine the enter plane temperature P'S aa 13 min afer the start cooing ‘S59 A alah ofthicknes L = [2em and thermal difsvty 2= 2 10"* ms has aii emper- ‘ture Gininbution 7, 100 sn evil). Fr ¢> 0, both boundaries are Rep at tero temperate ‘Bysinganeapiet fintedifrence scheme anda mesh ie x 2emcalclae he center temperature PevisSsand 13 mi afer the stare conting ‘40 Aslabofthckness = 6emand thermal difusivity ~ 8 x 10°? m*/shas an nial temperate isributon T, = 100 cos [2 2MaiLy], Fort > 0, the Boundary surface ax = 05 kept insulated, sn that atx = Gm i kept a zero temperature By using an explicit nite-diference scheme and 3 ‘mesh suc Ar ~ I em. calulte the temperature ofthe ineulated boundary = 1 and 10 mie after the start of coon S41 A slab of thickness L = 6 om and thera ifsivty 2 = 8 x 10°S ms hasan initia tempera tare distbuton 7,» 100 sin (a 2MsiL]} Fort >, the boundary surface at + = O's kept at 20 temperature, and tha atx = Sci kept ioslated By using an expt nite-iferene scheme anda Imesh size Av fm calculate the mperature of the ssulated boundary = Land min ae the Sart of ol. ‘$62 An aluminam plat (2 = 8 x 10°? mj) L~ 4m thick i italy ata uiform temperature T= 27 Suenly one fs surfaces is aed 10 T, ~ 220°C white other surface kept insulated By ing ancxplist finite-diference scheme and. sesh size Ax ~ I em. calevlat the temperature ofthe fest surface at and Yin ales the oer surfaces export hightemperature $843 Repeat Problem 556 fra lab of thickness L= Dem for tbe ease when one of the boundary Surfaces s exposed to hot air at T, ~ 220°C witha eat aster coefficient h = SO Wim") while the otter surface hep insulted The vermal conductivity of alurinum ik = 204Wi(a °C) ‘464 carhorstel bar of 4 em by 4 em sros section [2 = 1 « 10-* mis, = 35 Wim °C) ie inal at wniorm temperature 7, ~ 425C. Suddenly alls surfaces ae exposed to cooling by an Figure PS224 ear masta aitsueam at T, = 25°C with a hear wanslereoeticient b= 100 Win °C) By using a0 expt fnitediference scheme and mesh size Ax = &y ~ Lem, calulate the center temperatute = 1 ang Simin aller the slat of cooling, Bocuse of symmetry consider only one-quarter ofthe css section, asilusratedn Fig, P94 ‘5.65 A chrome-tel bar of 6 cm by 3cm cos setion [2 = 16 x 10°? mk = 60 Wien °C) is ‘willy ata uniform temperature T, = 20°C. Suddenly oe ofits surfaces having em widths expored to hot air at T, = 220 with hen transtercoeficient k= 350 Wier °C), wile the remaining surfacesare kept insulated a illustrated in Fig PS-65. By using an exp finite diference scheme and Imeshsize A= Ay = lem, culate how long it wil ake forthe midpoint ofthe insulated surface of ‘with L = Gem to reach 120°C. REFERENCES 1. Fox, L: Numerical Solution of Ordinary and Partial Differential Equations, Addivon-Wesky, Reading. Mass. 1962 2. Smith, 6. D.: Mameical Solon of Portal Different Equations with Exercises and Worked Soluions, Oxford Unversity Press, London, 165. 3, Richtmeer, RD: Diference Methods fr tal Value Problem, Interscience, New York, 1957, Dusinerre,G. M.: Hear Transfer Cautions by Fie Difernces, International Tertbok, Scranton, Pa, 1961, Forsythe, G.E.,and W. R. Wasow: Fae Difrencer Method for Portal Difeenial Equations, Wiley, New York, 1960 6. Larkin, BK. "Some Fiste Diference Methods for Problems in Transient Heat Flow,” Chem. Eng. Prog. Symp. Ser. 9,61 (1963). 1. Macon, > Numerical Analysis, Wey, New York, 1963 Southwell R." Relation Methods in Engineering Science, Oxford University Press, New York, 1940, 9 Ames, WF: Nominear Pasa Diferential Equations in Engincoring, Academic, New York, 1965, Pp. 365-389, 10, Crank, J, and P. Nicolson:~A Pracicl Method for Numerical Evaluation of olution of PDE ofthe Heat Conduction Type,” Proc. Combride Philos. Soe, 43:30-67 (1987, 1H, Nickell, RB. and E. Wilton: “Application ofthe Fine Element Method to Heat Conduction Analysis” Nae Bag. Des, 1276-246 (1968), 12. Oktay, Ura: Five Element Method: Basie Concepts and Applications, internation! Textbook, Seaton, Pa, 1973 13, Zeeniewiez,O-C. andl. K. Cheung: The Finite Bomon Method n gnering Scene, McGraw Hil, New York, 171[FIMITE-DIFFERENCE METHODS FOR SOLVING HEAT CONDUCTION PROBLEMS. 225 14, Marti, H.C. and GF. Cate: Introduction to Finite Element Analyis, McGraw-Hill, New York, wm, 15. Hchner, K. HL: Fine Element Method for Engineers, Wiley, New York, 1975 16. Segevin, LJ: Applied Fine Element Anaiyas, Wiley, New York, 1976 17, Oasit, MN. Heat Conduction, Wiley, New York, 1980 1s, Inernstoael Mathematial and Stata! Libra, th ed. ONB Bid, 7500 Balle Blvd Houston, TX 71036, 1979.Gaarrer SIX CONVECTION—CONCEPTS AND BASIC RELATIONS So far we considered heat transfer by conduction in solids in which no motion of the ‘medium was involved. In conduction problems, the convection entered the analysis. ‘merely as a boundary condition in the form of a heat transfer coefficient ‘Our objective in this and the following chapters on convection isto establish the physical and mathematical basis for the understanding of convective transport and. to reveal various heat transfer correlations In engineering applications, the pressure drop or the drag force associated with flow inside ducts or over bodies is also of interest. Therefore, appropriate cor- relations are presented to predict pressure drop or drag force in flow. ‘The analysis of convection is complicated, because the uid motion affects the pressure drop, the drag force, and the heat transfer. To determine the drag force or the pressure drop, the velocity field in the immediate vicinity of the surface must be known, To determine the heat transler by convection, the velocity distribution in the flow also is needed, because the velocity enters the energy equation: the solution of the energy equation yields the temperature distribution in the flow feld. ‘The literature on convection heat transfer is overhelming and ever-growing. In recent years, with the availability of high-speed, large-capacity digital computers, great advances have been made in the analysis of very complicated heat transfer problems in great detail. Nonetheless, a large number of simpler engineering problems can be handled with the use of standard heat transfer correlations ‘Therefore, we focus our attention on such situations, To achieve this objective, in this chapter we present a coherent view of the subject of convection in order to provide a firm basis for applications. Basic concepts associated with flow over a ‘body, low inside a duct, and turbulence are discussed. The role of temperature and velocity distribution in flow on heat transfer and drag force is illustrated, The velocity and temperature distributions in flow are determined from the solution of the equations of motion and energy. Therefore, such equations are presented for the case of two-dimensional, constant-property, incompressible ow 26“CONVECTION —CONCEFTS AND RASC RELATIONS. 227 in rectangular and cylindrical coordinate systems. The simplification of these ‘equations is illustrated in order to obtain the governing equations for the analysis of simpler heat transfer problems. Finally, the physical significance of dimensionless parameters is discussed, and the boundary-layer equations are presented. 61 FLOW OVER A BODY ‘When a fluid flows over a body, the velocity and temperature distribution at the immediate vicinity ofthe surface strongly influence the heat transfer by convection. ‘The boundary-layer concept frequently is introduced to model the velocity and temperature fields near the solid surface in order to simplify the analysis of con- ‘veetive heat transfer. So we are concerned with two different kinds of boundary layers, the velocity boundary layer and the thermal boundary layer. ‘Velocity Boundary Layer ‘To illustrate the concept of the velocity boundary layer, we consider the flow of a ‘uid over a flat plate, as illustrated in Fig. 6-1. The fuid at the leading edge of the plate (.c, at x = 0) has a velocity u,, which is parallel to the plate surface. As the fluid moves in the x direction along the plate, those fluid particles that make contact with the plate surface assume zero velocity (ie., no slip at the wall). Therefore, starting from the plate surface there will be a retardation in the x direction com- ponent of the velocity u(x, ») =u. That is, at the plate surface y = 0, the axial velocity component is zero, or u = 0, The retardation effect is reduced when the fluid is moving at @ location away from the plate surfaces at distances sufficiently far from the plate the retardation effect is considered zero, that is,u = wu for large y. Therefore, at each location x along the plate, one considers a distance y = 3(x) from the surface of the plate where the axial velocity component w equals 99 percent of the free-stream velocity u...that i, v = 0:99u,,. The locus of such points where w= 0.99u,. is called the velocity boundary layer 5(x). With the boundary-layer ‘concept thus introduced for flow over a flat plate the flow field can be separated Lamina boundary Teasiion “Trent boundary, pier ‘eon — tye Te Boundary | - F : layer oltidile ast, Boundary bree Rey Figure 61 Boundary ayer concept for Bow along a fat plate228 wont rRANsrER into two distinct regions. (1) In the boundary-layer region, the axial velocity com- ponent u(x, y) varies rapidly with the distance y from the plate; hence the velocity ‘gradients and the shear stress are considered large. (2) In the region outside the boundary layer, called the potential-flow region, the velocity gradients and shear stresses are negligible. Referring to the illustration in Fig. 6-1, we now examine the behavior of flow in the boundary layer with the distance x from the leading edge of the plate. The characteristic of the flow is governed by the magnitude of the quantity called the Reynolds number. For flow over a flat plate as illustrated in Fig. 6-1, itis defined as Re, = “2* D where u,, = free-stream velocity x = distance from leading edge v= kinematic viscosity of fuid ‘The boundary layer starts at the leading edge (that is, x = 0) of the plate as a laminar boundary layer, in which the flow remains orderly and fluid particles move along streamlines. This orderly motion continues along the plate until a critical distance is reached or the Reynolds number attains a critical value. After this critical Reynolds number is attained, the small disturbances in the flow begin to be amplified and fluid fluctuations begin to develop, which characterize the end of the laminar boundary layer and the beginning of transition from the laminar to turbulent boundary layer. For flow along a fat plate, the critical Reynolds number at which the transition from laminar to turbulent flow takes place is generally taken, for most analytical purposes, as eX 5x 10° 62) However, this critical value is strongly dependent on the surface roughness and the turbulence level ofthe free stream. For example, with very large disturbances in the free stream, the transition may begin at a Reynolds number as low as 10°, and for flows which are free from disturbances, it may not start until a Reynolds ‘number of 10° or more. But for flow along a flat plate, the boundary layer is always turbulent for Re, > 4 x 10°. In theturbulent boundary layer next tothe wall, there isa very thin layer, called the viscous sublayer, where the flow retains its viscous-flow character. Adjacent to the viscous sublayer is a region called the buffer !ayer in which there is fine-grained turbulence, and the mean axial velocity rapidly i creases with the distance from the wall. The buffer layer is followed by the turbulent layer in which there is larger-scale turbulence, and the velocity changes relatively little with the distance from the wall. Figure 6-2 shows the boundary-layer concept for flow over a curved body. In this case, the x coordinate is measured along the curved surface of the body; byee ent Poa of ito Figure 62 Boundaryayer concept for flow along a curved body and the Now separation. starting from the stagnation point and at each x location, the y coordinate is measured normal to the surface ofthe body. The fre-atream velocity q(x) is not constant, but varies with distance along the curved surface. The boundary-layer concept discussed above also applies to this particular situation. The boundary- layer thickness 6(x) increases with the distance x along the surface. However, ‘because ofthe curvature of the surface, after some distance x, the velocity profile ‘ux, )) exhibits a point of inflection; that is, u/dy becomes ero at the wall surface. ‘Beyond the point of inflection, the flow reversal takes place, and the boundary layer is said to be detached from the wall surface. Beyond the point of flow reversal, the flow patterns are very complicated and the boundary-layer analysis is no longer applicable. Drag Coefficient and Drag Force Suppose the velocity profile u(x, y) in the boundary layer is known. The viscous shear stress t, acting on the wall at any location x is determined from its definition by =p MD) a 0 Here, the constant of proportionality s isthe viscosity ofthe fluid. Thus, knowing the velocity distribution in the boundary layer, one can determine the shear force acting on the wall owing to the flow. The definition of shear stress as given by Eq, (6-3), however, is not practical in engineering applications. In practice, the shear stress or the local drag force t, per unit area is related to the local drag coefficient c, by the relation 63) oF (64) where p isthe density of the Mud and u. isthe freestream velocity. Thus, knowing the drag coefcient, we can calulate the drag force exerted bythe fluid flowingover the lat plate, Equating Eqs. (6-3) and (6-, we obtain 2v aux, ») 6)230 Hear TRANSTER ‘Thus, the local drag coefficient can be determined from Eq, (6-5) if the velocity profile (x, ») in the boundary layer is known, ‘The mean value of the drag coefficient c,, over x = Oto x = L is defined as ws Thee Knowing the mean drag coefficient c,, we can find the drag force F acting on the plate from x = Oto.x = Land for width w from Pl F = wl, oy Example 6-1 The velocity profile ux, ») for boundary-layer flow over a flat plate is given by ene ma ev El @ te ~ 2509] 72 [365] where the boundary-layer thickness 6(x) is ax 66) B80 Bue Develop an expression forthe local drag coeficient cy, Develop an expression for the average drag coefficient c, over a distance x = L from the leading edge of the plate 5) ) SOLUTION The local drag coefficient c, is related to the velocity profile by Eq, (6-5) as 2v dul») For the velocity profile given above, by Eq (a) we have Bie Ino” 2500) aux, ») ‘Then TaD Introducing the above expression for 4) into this relation, we find By fe _ fT = Ge 280 vx 2800, x 0646 ~ Rel? ©[CONVECTION CONCEPTS AND BASIC RELATIONS 231 The average ¢, over 0 < x < Lis defined by Eq, (6-6) as f “o ae Introducing Eq. (c) into this expression, we obtain 1 fit fe ava? [HT _ 0646 280u,L ~ “Rel? Thus, the mean value of the drag coefficient over 0 < x < L is equal to Wwice the value of the local drag coefficient at x = Note that the expression for the drag coefficient developed here is an approximate one, because the velocity profile used in the analysis is approxi- mate, Example 62 The exact expression fo the local drag coefficient c, for laminar flow over a flat plate is given by 0.664 Rel? Air at atmospheric pressure and at T., = 300K flows with a velocity of = 1Smjs along the plate. Determine the distance from the leading edge ofthe plate where transition begins from laminar to turbulent flow. Calculate the drag force F acting per I-m width ofthe plate over the distance from x = to where the transition starts. SOLUTION The physical properties of atmospheric air at 300 K are p= LAT kim? y = 0.168 x 10-* més ‘The transition is assumed to occur at a distance x = L, where Re, = Then the distance L is x 10%. Re, = tek Ls 7 0168 x 1# = 5 * 10 L=56m The average drag coefficient cy over the length x = Oto x = Lisdetermined by Eq, (6-6) as232 wear raAnsees ‘That is, the average value of the drag coefficient over the length x = Otox = L. is twice the value of the local drag coefficient at x = L. By introducing the numerical values, c, becomes aces . ia = 18 «10 Then the drag force acting on the plate is determined by Ea, (6-7) as Fm beg ME = (1y(56).88 x 19-2) ES = 209 x 10-7N ‘Thermal Boundary Layer Analogous to the concept of velocity boundary layer, one can envision the develop- ‘ment ofa thermal boundary layer along the fiat plate associated with the temperature profile in the fluid. To illustrate the concept, we consider that a ud at a uniform temperature T., flows along a fat plate maintained at a constant temperature T,, Let x and y be the coordinate axes along and perpendicular to the plate surface, respectively, as illustrated in Fig. 6-3. We define the dimensionless temperature OCs, yas _ Tey) = Te To =e where T(x, ») is the local temperature in the fluid. At the wall surface, the Quid temperature is equal to the wall temperature, hence Ox, 9) = 0 at y = 0 (wall surface) (69a) x, 9) 68) [At distances sufficiently far from the wall, the fluid temperature remains the same as Ty: then Hx ya1 as yaw ‘Therefore, at each location x along the plate, one envisions a location y the fluid where O(x, y) equals 0.99. The locus of such points where 6(x, is called the thermal boundary layer 64x). i Bae Tena ‘Figure 62 Thermal boundary-layer concep forthe fow ofa hot uid over coi wal(CONVECTION —CONCEPIS AND BASIC RELATIONS 233 “The relative thicknesses ofthe thermal boundary layer (x) and the velocity boundary layer 8(x) depend on the magnitude ofthe Prandtl number forthe fuid. For fluids having a Prandtl number equal to unity, such as gases, (x) = 5(x), ‘The thermal boundary layer is much thicker than the velocity boundary layer for fluids having Pr < 1, such as liquid metals, and is much thinner than the velocity boundary layer for fluids having Pr > 1 Heat Transfer Coefficient Suppose the temperature distribution T(x, y) in the thermal boundary layer is known. Then the heat fux q(x) from the fluid to the wall is determined from gs) = kT) k (6-100) Where k is the thermal conductivity of the fluid. However, in engineering applica- tions itis not practical to use Eq. (6-104) to calculate the heat transfer rate between the fluid and the wall. In practice, a local heat transfer coefficient h(x) is defined to calculate the heat flux between the fluid and the wall: a8) = WO)(T ~ Ts) (6-108) Equating (6-102) and (6-106), we obtain (eT/av),-0 Hy = a TO (11a) This expression is now written in terms of the dimensionless temperature O(x, 9) as ox) = 429) hx) cD 116) ‘Thus, Eqs. (6-11) provide the relation for the determination of the local heat transfer coefficient Mx) from the knowledge of the dimensionless temperature distribution Ox, y) in the thermal boundary layer ‘The mean heat transfer coefficient hy over the distance x = 0 10 x = L along the plate surface is determined from beat [ oer (612) Knowing the mean heat transfer coeficient hy, we can find the heat transfer rate Q from the fluid to the wall from x = Oto x = Land for the width w Q = WLhg(To — To) (6-13) Example 6-3 An approximate expression for the temperature profile in the ‘thermal layer is given by _Te-T 39 ifyp fo ats ~ [ital234 Wear TRANSFER and the thickness ofthe thermal boundary layer 3(x) is given by where Pr is the Prandtl number, Develop an expression forthe local heat transfer coefficient) 80) = 453 SOLUTION The local heat transfer coefficient is related to O(x, y) by Eq. (6-116) (x, ») Wx) a eo For the temperature profile given above, we have Hx)! _ 3 ay jo 25 (x) Then a6 m9 356) Introducing the above relation for 5,(x), we find to) = 0331 * prt Ret ‘This expression is arranged in the dimensionless form as Nu, = 2% 0331 Prt Re? R where Nu, is called the local Nusselt number. We note that the Nusselt number is a function of the Prandtl and Reynolds numbers. Example 6-4 The exact expression for the local Nusselt number for laminar flow along a flat plate is given by Nu, = “OF = 0.332 Pr'? Rel? Develop relation for the average heat transfer coefficient h(x) from x = 010 xe Atmospheric air at Ty = 400 K with a velocity u,, = 1.5 m/s flows over a flat plate L = 2m long maintained at a uniform temperature Ty = 300 K. Caleulate the average heat transfer coefficient h, from x=0 to x = L=2m, Caleulate the heat transfer rate from the airstream to the plate from x =Otox=L =2mforw =05m.(CONVECTION —CONCEPIS AND HASIE RELATIONS 235, Owx=Lis Souumion The average heat transfer coefficient h, from x determined by Eq. (6-12) as need Pinas t oa nanny [oo ys Rei! = fosat ren BL) ax Cheat “Thats, the average heat transfer coefficient hy fromx = 0tox = Listwice the ‘value of the local heat transfer coefficient h(x) evaluated at x = L. ‘The physical properties of atmospheric air evaluated at the arithmetic average of the free-stream and wall temperatures, 350K, are v = 0.21 x 10-* ms, k = 0.03 Wi(m °C), and Pr = 0.697. Then, at L = 2m, ieee sina vy 021 x 10 Rey, = 143 x 10° 2 = 334 Wim? "C) ‘Then the eat transfer rate Qis determined by Ea. (6-13) as Q = whh(Te — Te) = 05 2 x 3.34 (400 ~ 300) = saw Relation between c, and A(x) In Example 61 we showed that if the velocity profile and the boundary-layer thickness are known, then an expression can be developed for the local drag coef ficient ¢, for laminar flow along a fat plate. Similarly, in Example 6-3 we showed that if the temperature profile and the thermal boundary-layer thickness are available, an expression can be developed for the local heat transfer coefficient for laminar flow along a flat plate Now we seek @ relation between the heat transfer and drag coefficients. We consider the exact expressions for the local drag coefficient and the local [Nusselt number for laminar flow along a fat plate given, respectively, by $ = 0332 Rez" (14a) Nu, = 0332 Pr!? Rel? (6148)236 near TRANSrER We define the local Stanton number St, as st, = 2 Pepe which can be rearranged in the form Noxxik __Nuy (ia\uex) ~ Pr Re, ‘Then the expression (6-14b) for the local Nusselt number can be rewritten as St, = 0332 Pro2 Rez"? 14) From Eqs. (6-14a) and (6-14), the following relation is obtained between the ‘Stanton number and the drag coefficient: St Su,Pr? = 3 (6-150) ‘This expression is referred to as the Reynolds-Colburn analogy that relates the local drag coefficient c, to the local Stanton number St, for laminar flow along a flat plate. Thus, by making frictional drag measurements for laminar flow along a flat plate with no heat transfer involved, the corresponding heat transfer coefficient can bbe determined by Eq, (6-15). It is much easier to make drag measurements than heat transfer measurements. Equation (6-154) is also applicable for turbulent low along a flat plate, but it does not apply to laminar flow inside a tube, In the case of the average values, Eq. (6-154) is written as St, Pr? = (6150) where St, and cq are, respectively, the mean Stanton number and the mean drag coefficient Example 65 Atmospheric air at 300 K flows with a velocity of ug = 5 m/s along a flat plate L = 1 m long. The plate has a width w = 0.5 m. The total drag force acting on the plate is determined to be F = 18 x 10~? N. By using the Reynolds-Colburn analogy, estimate the corresponding average heat transfer coefficient h, for flow of air over the plate. Souvtion The total force acting on the plate is related to the mean drag coe ficient by Eq. (6-7) as = whe, M2 P= whe,‘CONVECTION —CONCEPTS AND RASIC RELATIONS 237 ‘and the average heat transfer coefficient is related to the average drag coefficient by the Reynolds-Colburn analogy. given by Eq. (6-15b), as 4, te pra Byte 2 ‘The physical properties of atmospheric air at 300K are taken as p LITT kg/m’, cy = 1.006 x 10° J/(kg-°C), and Pr = 0.708. Then the mean drag coefficient cq is determined from the fist of these equations as 18 x 10° = 5), AS or Gq = 2447 10-9 a» We can find the mean heat transfer coefficient hy from the second equation as hw 13 2447 x 10°? CAT C.006 x 105) 79” z or hag = 9.12 Wilt? 0) Example 66 Atmospheric air at T,, = 400 K flows with a velocity of u, = 4 ‘ms along a flat plate L = 1 mong maintained at a uniform temperature T,, = 300 K. The average heat transfer coefficient is determined 10 be hy = 775 Wj{m? -°C). Using the Reynolds-Colburn analogy, estimate the drag force exerted on the plate per I-m width, SoLUTION The Reynolds-Colburn analogy is given by Eq, (6-155) as h Pele 2 ‘The physical properties of atmospheric air at the mean film temperature T,, = (400 + 300)/2 = 350 K are given by p= 0.998 kg/m? 1009 J/(kg-°C) Pr = 0.697 ‘Then the drag coefficient cy is 115 0.998 x 1009 x 3 Oem “2 or Cu = 3.03 x 10°? ‘The drag force Fis given by Eq. (6-7) 8 = whe Me Fa wt, = cyaya0s 10-0204 =242 x 10 N62 FLOW INSIDE A DUCT The basic concepts discussed in the last section on the development of velocity and thermal boundary layers for flow along.a flat plate also apply to flow at the entrance region of ducts. We illustrate this matter by considering flow inside a circular tube. Velocity Boundary Layer Consider the flow inside a circular tube, as illustrated in Fig. 6-4. The fluid has a ‘uniform velocity up at the tube inlet. As the fluid enters the tube, a velocity boundary layer starts to develop along the wall surface. The velocity of fluid particles at the wall surface becomes zero, and that atthe vicinity ofthe wall is retarded asa result, the velocity in the central portion ofthe tube increases to satisfy the requirement of thecontinuity offlow. The thickness ofthe velocity boundary layer 6(2) continuously grows along the tube surface until it fis the entire tube. The region from the tube Inlet to little beyond the hypothetical location where the boundary layer reaches the tube center iscalled the hydrodynamic entry region. In this region the shape ofthe velocity profile changes in both the axial and radial direction. The region beyond the hydrodynamic entry length is called the hydrodynamically developed region, because in this region the velocity profile is invariant with distance along the tube. Ifthe boundary layer remains laminar until it fils the tube, fully developed laminar flow of parabolic velocity profile prevails in the hydrodynamically developed region. However, if the boundary layer changes to turbulent before its thickness reaches the tube center, fully developed turbulent flow is experienced in the hydrodynamically developed region. When the flow is turbulent, the velocity profile is flatter than the parabolic velocity profile of laminar flow. For flow inside a circular tube, the Reynolds number, defined as Re (616) is used as a criterion for change from laminar to turbulent flow. In this definition, tu, is the mean flow velocity, D isthe tube’s inside diameter, and v is the kinematie yeodymemicaly developed Fare 64 Concept of development of veloity boundary layer at entrance region of a crular tube‘CONVECTION —CONCEPTS AND RASIC RELATIONS 239 viscosity of fuid. For flow inside a cireular tube, the turbulent flow is usually observed for iq Re > 2300 17) However, this critical value is strongly dependent on the surface roughness, the inlet conditions, and the Muctuationsin the flow. In general, the transition may occur in the range 2000 < Re < 4000. Friction Factor and Pressure Drop In engineering applications, the pressure gradient dP/dz associated with the flow isa quantity of interest, because the pressure drop along a given length of tube can bbedetermined by the integration of dP dz over the ength. To develop an expression defining dP dz, we consider a force balance over a differential ength dz of the tube. By equating the pressure force to the shear force atthe wall, we obtain (soe Fig. 6-5) (PA), — (PA)er ge = S82 ty a__s D4 G7 aS Gap oD (6180) where A is the cross-sectional area and S the perimeter. The shear stress «, at the wall is related to the velocity gradient by dul ou ay va a loan oy since r = D/2 ~ y. Then from Eqs. (6-18a) and (6-186), we have dP _ du dul iz De (180) In engineering applications Eq, (6-18c) is not practical for the determination of <4P/dz, because it requires the evaluation of the velocity gradient at the wall. To Poa Figee 65 A force balance on 2 diferential volume cement240 near TRANSFER calculate the pressure drop in engineering applications fiton factor fs defined a ple D here uai8 the mean velocity of flow inside the tube and piste density ofthe uid By equating (6-18) and (6184), the following expression is obtained forthe friction factor (6184) 81 au) Pll OF baa Thus, given the velocity distribution u forthe Row inside the tube, the ton factor an be dacrmine rom Ea, (180 Given the fetion factor, the pressure drop P, — Py AP over the eat Lotthe bes determined bythe fe- (186) ‘ap = pm or the pressure drop AP becomes Lp ON apafeoe § 192) ICM isthe flow rate in cubic meters per second through the pipe, the pumping. power required to get the fluid through the pipe against the pressure drop AP emp pone = (aan) pumingrover= war Sorw | is Example 6-7 The velocity profile for hydrodynamically developed laminar flow inside a circular tube is given by orm G] where R is the inside radius of the tube and uy is the mean flow velocity Develop an expression for the friction factor f for flow inside the tube. SOLUTION Equation (6-18e) gives the friction factor in terms of the velocity _Bradient at the wall. For the velocity profile given above, we have ue) a My nO R 7D‘CONVECTION CONCEPTS AND BASIC RELATIONS 241 ‘where Dis the inside diameter ofthe tube. By introducing this expression into Eq. (6-18), the friction factor f for hydrodynamically developed laminar flow inside a circular tube is determined as Cc pu, ~ Re where the Reynolds number is defined as Re we Pm _ tinD uy v= 075 « 10°4 m?/s) lows with a circular tube having an inside 5 em. Calculate the friction factor and the pressure drop over 100 m of the tube. the length L. SOLUTION First we calculate the Reynolds number in order to establish ‘whether the flow is laminar or turbulent, gD _ O15 x 0025 vy OTS I ‘The flow is laminar, hence the expression developed in Example 6-7 for the friction factor is applicable. We find ‘The pressure drop cleulated by Ea. (6195) a8 Low arn toy 01025, = 50000Nin? _ ul 100 as nos mip) ‘Thermal Boundary Layer In the case of temperature distribution in flow inside a circular tube, it is more Jiffcult to visualize the development of the thermal boundary layer and the ‘existence of a thermally developed region. However, under certain heating or cooling conditions, such as constant heat flux oF uniform temperature at the tube wall, such a concept is possible. ‘Consider a laminar flow inside a circular tube subjected to uniform heat flux242 ext mansren at the wall. Let r and z be the radial and axial coordinates, respectively. A di- ‘mensionless temperature 6(r, 2) is defined as T(r, 2) Ta) = TS mr, (6-20a) where T.(2) = tube wall temperature T,{2) = bulk mean fluid temperature over cross-sectional area of tube at z T(, 2) = local fluid temperature Clearly, (7,2) is zero at the tube wall surface and attains some finite value at the tube center. Then, one envisions the development of a thermal boundary layer along the wall surface. The thickness of the thermal boundary layer 8,2) con- tinuously grows along the tube surface until it ills the entire tube. The region from the tube inlet to the hypothetical location where the thermal boundary-layer thickness reaches the tube center is called the thermal entry region. In this region, the shape ofthe dimensionless temperature profiler, 2) changes in both the axial and. radial direction. The region beyond the thermal entry length iscalled the thermally developed region, because in this region the dimensionless temperature profile remains invariant with the distance along the tube, that is, Tn2) ~ Tee) Tele) = Tae Its difficult to explain qualitatively why 0(r) should be independent of the = variable while the temperatures on the right-hand side of Eq, (6-205) depend on both r and z. However, it can be shown mathematically that for either constant temperature or constant heat flux atthe wall the dimensionless temperature 6(r) depends on only r for sufficiently large values of z. Hn) = (6-208) Heat Transfer Coefficient In engineering applications involving fluid flow in a tube, the heat transfer rate between the fluid and the tube is of interest. Here we discuss the concept of heat ttansfer coefficient that is frequently used in engineering applications for the determination of heat transfer between the fluid and the wall surface. Consider a fluid flowing inside a circular tube of inside radius R. Let T(r, 2) bbe the temperature distribution in the fuid, where r and z are the radial and axial coordinates, respectively. The heat flux from the fluid to the tube wall is determined from its definition oro % (21a) ‘where k is the thermal conductivity of Rud, ‘In engineering applications, itis not practical to use Eq. (6-2a) to determine the heat transfer between the fluid and the tube wall, because it involves the evalua-‘CONVECTION CONCEPTS AND HASIC RELATIONS 243, ation of the derivative of temperature at the wall. To avoid this difficulty, a local heat transfer coefficient h(2) is defined as 2) = ENLTA2) ~ Tye) (6216) Where T(z) = bulk mean fluid temperature over the tube cross-sectional area at location = tube wall temperature at 2 Te) Clearly, ifthe heat transfer coefficient is available, it is a very simple matter 10 determine the heat fux at the wall for a given temperature difference between the ‘mean fluid and the tube wall. Therefore, the use of heat transfer coefficient is very ‘convenient in engineering applications, and its determination under various flow ‘conditions has been the subject of numerous experimental and analytical investiga- tions, Here we treat the relation between the heat transfer coefficient h(z) and the luid temperature T(r, 2) in order to determine h(2) from T(r, 2) ‘By equating (6-21a) and (6-216), we obtain k___ aT tr.2)) "=~ T-TO beam ed where 72) and T,(2).fora circular tube of radius R, are determined from JEuOMTCr 2)2nr dr _ fSule)T(, 22nr dr To) = RG htardr ge (oa) Te) = Te 2 roe (6229) ‘The mean fluid temperature Tz(2) is a definition based on the thermal energy transport withthe bulk motion ofthe fluid as it passes through the eross section, because the quantity “peu” represents energy flux per unit area, For incom= pressible, constant-property fluid. the pe, term is canceled Equation (6-222) can be written in ‘erm ofthe dimensionless temperature A, 2) defined by Ea, (6-204), as amr, 2) er te) (6230) In the thermally developed region, the dimensionless temperature 6(7) is independent of z. Then Eq. (6-232) reduces to - (6-230) where 6(r) is defined by Eq. (6-206). This result implies that in the thermally de- veloped region the heat transfer coefficient does not vary with the distance along. the tube; and itis valid for heat transfer under conditions of constant wall heat flux ‘or constant wall temperature.244 cat TRANSFER ‘The definitions given by Eqs. (6-23) can be used to develop relations for the heat transfer coefficient if the dimensionless temperature distribution in the fluid, defined by Eq, (6-208), is available Example 69 Consider laminar forced convection inside a circular tube of inside radius R and subjected to a uniform heat Bux atthe tube wall. I the region where the velocity and temperature profiles are fully developed, the ensionless temperature 6(7), defined by Eq, (6-20), is given inthe form URS HLA) -1)] Develop an expression for the heat transfer coefficient. Or) SOLUTION The heat transfer coefficient can be determined by Eq. (6-23). In the thermally developed region, the dimensionless temperature does not depend on the axial coordinate. Hence by Eq. (6-236) we write ae) " dr |or ® For the temperature profile given above, we have dy] 96/1 1) _ hat a ar fog TGR” 2R) ~~ R= UD where D is the diameter, Introducing this result in expression (owe Chain the beat transfer coeicet in the thermally developed region laminae flow inside acrular tube subjected to unonm heat Ruy atthe wall 9 84 “TD ‘This expresion i aranged inthe dimesinls formas h ‘where Nu is the Nusselt number. Example 6-10 Engine oil [v = 08 x 10°* m/s, k = 0.14 Wim. °C)] flows with a mean velocity of ty = 0.2 m/s inside a 1,25.cm-diameter tube which is electrically heated at the wall at a uniform rate of q = 2450 Wim?. The ‘heat transfer is taking place in the thermally developed region. Calculate the temperature difference between the tube wall surface and the mean flow temperatures.‘CONVECTION —CONCEPTS AND BASIC RELATIONS 245, Souurion To ensure that the flow is laminar, we check the Reynolds number: aD _ 02 x 00125 Ren 08 x 10% 3125 Thus the flow is laminar. "The heat transfer coefficient for laminar in the thermally developed region subjected to a uniform heat lux at the wall is determined in Example 6-9 as 48 k fait For this problem, h becomes 48 014 aa i= A pias 7 4887 When? °C) ‘Then the temperature difference AT between the wall surface and the mean fluid temperatures is 2450 ar =4= FF = sore 63 CONCEPTS ON TURBULENCE Osborne Reynolds [J in his classic experiment of injecting dye into the water flowing through a transparent pipe, showed that at low flow rates the flow was streamlined; but as the flow rate was increased, the streamlines became unstable ‘and the laminar motion changed to turbulent flow. The term turbulent is used to denote that the motion of the fluid is chaotic in nature and involves crosswise mixing or eddying superimposed on the motion of the mainstream, ‘The eddying or crosswise mixing in turbulent flow is advantageous in that it assists greatly in improving the heat transfer in flow, but it has the disadvantage of increasing the resistance to flow. The flow patterns in turbulent flow are so ‘complex that they consist of a spectrum of coexisting vortices varying in size from the larger ones down to those of microscopic dimension. For example, in turbulent flow inside a circular tube, the size ofthe vortices varies from a significant fraction of tube radius in the central region to microscopic sizes near the wall where the velocity approaches zer0. In turbulent flow, the properties such as velocity, temperature, and pressure are subject to flctuations both with location in the fuid and with time, Therefore, the instantaneous values of these properties can be represented as a sum ofa time- averaged mean part and a fluctuating part in the form aut btu fen PaP246 ear Tansee r Autaos Taner ‘one with ime in turbulent low where u ;. T, and P, are the instantaneous values; u, v, T, and P are the time= averaged values: and w’«, T’,and P’ are the fluctuations. For example, if a thermo- couple with a sufficiently small time constant is placed in a given location in turbulent flow and the instantaneous value of temperature T; is recorded as a function of time, temperature may exhibit fluctuations, as illustrated in Fig. 66. In turbulent flow, the Ructuations are considered superimposed on local average flow. For example, the time-averaged value T of temperature is defined as ee rok (na 25 Where Ar is a very small time interval which is large enough for recording the turbulent fluctuations but sufficiently small for the temperature to be unaffected by the external disturbances on the system. The averaging process similar to that defined by Eq. (6-25) is applicable for the averaging of velocity, pressure, etc. The effects of these fluctuations occurring in velocity and temperature in turbulent flow are to inerease heat transfer and resistance to low (drag). Consider turbulent flow along a flat plate, as illustrated in Fig. 6-1. The y coordinate is measured normal to the plate surface. Let u,, v1, and T, be the in- stantaneous values of the axial and normal components of velocity and tempera ture; and let vo, and T” denote the corresponding fluctuations from the average values u,v, and T, respectively. We now examine the effects of turbulence on the shear stress and heat fiux in the fluid Let rbe the shear stress in the fluid in the x direction at a location y from the wall surface. This shear stress can be considered to be composed of two parts Piscous shear siress resulting from the mean flow velocity w and turbulent shear stress resulting from the velocity fluctuations w' and v' from the mean values, With this consideration we write T= Tyce + Trent (6-260) where (6-268) Surheten = = piTT (6-26)‘CONVECTION —CONCEFTS AND BASIC RELATIONS 247 Here the viscous shear stress term given by Eq. (6-268) is similar to the one en- ‘countered in laminar flow. The physical significance of the turbulent shear stress term given by Ea. (6-26) is not quite apparent, but it can readily be derived by introducing the instantaneous velocities u, and e, as defined by Eqs. (6-24) into the ‘x momentum equations and by applying the rules of averaging to the terms in- volving the cross products of the fluctuations. See Refs. 2 to 7 for a more compre hensive discussion of turbulence. ‘Now, let q be the heat flux in the fluid in the y direction, This heat flux can be considered as composed of two parts: the difsive or conductive heat flux due to the gradient of the mean temperature T in the y direction and the turbulent heat ‘flux resulting from the temperature fluctuations 7” and the velocity fluctuations wv in the y direction, With this consideration, we write Asse + Atte (6270) where ar a 2) eet (6210) Here the diffusive heat flux given by Eq, (6-276) is similar to that encountered in ‘conduction, but the significance of the turbulent heat flux component given by Eq, (6-270) is not quite apparent. It is derived by introducing u, v,, and T, as defined by Eqs. (6-24) into the energy equation for forced convection and manipulat- ing by utilizing the rules of averaging ‘The relations given by Eqs. (6-26c) and (6-27¢) for the turbulent shear stress and the turbulent heat flux are not useful for computational purposes unless they are related to the mean quantities w and T. Because of the complexity of turbulence, attempts have been made to develop such relations by semiempirical hypothesis. For example, Prandtl [3] and von Karman [4] proposed the mixing-length concept that has been used extensively and successively in relating turbulent stress and turbulent heat flux to the gradients of average velocity and average temperature. Mixing-Length Concept ‘The basic idea is analogous to the mean-jree-path concept for molecules in the kinetic theory of gases. The main difference is that in turbulent motion macroscopic lumps are envisioned. That is, for turbulent flow in the x direction along the surface, itis postulated that fluid particles a distance y from the wal surface coalesce into ‘macroscopic lumps and then travel, on the average, a distance ! in the direetion ‘normal to the main flow while retaining their x direction momentum before they are dispersed. Thus, if the slow-moving lumps enter the fast-moving layer. they ‘act asa drag on it, and the momentum is transferred between layers as a result of transverse mixing. Of course, Is an unknown quantity, and in reality there is no such clearly defined distance. Although the concept lacks generality, it has been found useful in the study of turbulent exchange in most engineering applications.248 ear neansten Prandtl postulated that the velocity fluctuations can be related to du/0y by oy "ay where u’ and v' are of opposite sign and I, and ly are the mixing lengths for mo- ‘mentum transport A similar approach can be applied to relate T”to T/ay as ar (6-284, 6) rene (6-28) where [ss the mixing length for energy transport. We now define Lhe® (29) In view of these relations, we write — aul ou iw Bled (6-300) |u| aT ~ 8S lay (6-308) Introducing Eqs. (6-30) into Eqs. (6-26) and (6-27), we obtain the relations for the shear stress and heat fux in turbulent flow: au au St Pe (31a) or or a= ha = Pers (6-316) where & = eddy diffusivity of momentum 4, = eddy diffusivity of heat (or eddy conductivity) Equations (6-31) can be rearranged as (6-320) (6-328) Where v= kinematic viscosity and a = thermal diffusivity. Equations (6-32) clearly demonstrate the effects of turbulent flow in enhancing both the shear stress (drag) and the heat flux. Depending on the level of turbulence, the turbulent transport properties fm and ¢ may be an order of magnitude larger than the diffusive properties v and a‘CONVECTION CONCEPTS AND MASIC RELATIONS. 249) ‘The eddy diffusivity of momentum cy and the eddy diffusivity of heat «5 are not necessarily the same. Their ratio is defined as ‘= Pr, = turbulent Prandtl number (633) ‘This definition is analogous to the definition of the Prandtl number: 5 = Pr = Prandtl number (6-34) ‘The physical significance of Pr, and Pr should be distinguished. The Prandtl ‘number is a physical property of the Quid, and it varies from about unity for gases {0 very large values for ordinary liquids and oils, However, the turbulent Prandtl number is a property of the flow field more than of the fuid. Various models have been developed for the determination of Pr,. The simplest is due to Reynolds ‘who assumed Pr, = 1, which implies that heat and momentum transfer in turbulent flow takes place exactly by the same process, The numerical values of Pr, may well vary between 1 and 2. ‘Although the mixing-length and the eddy diffusivity concepts have been used extensively and successfully to solve many problems in engineering applications, the very basis of the method and the empirical constants associated with it are not “universal. Thus, their scope is very limited with respect to the geometries and flow conditions experimentally investigated. Hence the results cannot be readily extrapolated to other, more complicated flow situations. To overcome these deficiencies, new approaches have been tried in recent years to determine the transport properties for turbulent flow by solving a system of partial differential equations similar in form to the conservation equations; a number of constants associated with such equations are determined experimentally. Although such ‘equations for determining the turbulent flow properties have existed since the ‘middle 1940s and early 1950s, renewed interest in this area has been generated in recent years with the availabilty of high-speed, large-memory digital computers. Launder and Spalding [8, 9] have summarized the governing equations for the calculation of transport properties for turbulent flow. Velocity Distribution in Turbulent Flow Velocity distribution in turbulent flow has been investigated extensively because of its importance in practice, but no fundamental theory is yet available to de- termine this velocity distribution rigorously by purely theoretical approaches. ‘Therefore, empirical and semiempirical relations ae used to correlate the velocity field in turbulent flow. ‘Nikuradse [10,b] was an early investigator who presented careful measure- ment of velocity distribution in turbulent flow through a smooth pipe. Later, experiments were performed by other investigators for turbulent flow along a flat plate [11,12] and inside a pipe [13]. Attempts were then made to develop empirical relations that would fit the velocity distribution in turbulent flow250 ear TRANSHER [5,714.15]. Here we discuss the velocity distribution law based on the concept of separating the flow field into three distinct layers as illustrated in Fig. 6-1. ‘That is, (1) avery thin layer immediately adjacent to the wall in which laminar or viscous shear stress is dominant is called the viscous sublayer: (2) adjacent to this layer is the bufer layer in which viscous and turbulent shear stresses are equally important; and (3) the third layer that follows the bufler layer is called the turbulent layer in which turbulent shear stress is dominant, We now examine the velocity distribution laws for each layer for the steady, turbulent flow of an incompressible, onstant-property fui over a smooth surface. In the study of velocity distribution for turbulent low, the following two dimensionless quantities are introduced less velocity fe = dimensionless distance ? where p is the density, zis the shear stress atthe wall, is the kinematic viscosity, and w is the velocity component parallel to the wall surface. Experiments have shown that the viscous sublayer is maintained in the region y* <5, where the laminar shear stress is dominant and the turbulent shear stress is virtually zero. Therefore, the shear stress is taken in the form to = js du/dy. The integration of this expression for constant to with u = 0 for y= 0 leads to the following result for the velocity distribution in this layer: Y" for viscous sublayer, 0 < y* < $ (635) ‘The buffer layer is considered to extend from y* = 5 to y* = 30, and a logarithmic velocity distribution law in the form u* = A In y* + B is assumed. The constants A and B are determined from the requirement that the velocity u* bbe equal to that of viscous sublayer and of the turbulent layer at y* = 5 and _y* = 30, respectively. The resulting velocity distribution becomes ue =SOIny* ~305 for buffer layer,S 30 is considered to be the turbulent layer where the laminar shear stress is negligible in comparison to the turbulent shear stress. By utilizing the mixing-length concept and assuming that the mixing length varies linearly with the distance from the wall in the form ! = xy, it can be shown that the velocity distribution in the turbulent layer has a logarithmic profile in the form wntnyrec (63 where is called the universal constant Experiments have shown that x ~ 04, and the constant C hasbeen determined by the correlation of Ea. (637) withthe measured elocty profile. For turbulent flow inside a smooth pipe, C= 3‘CONVECTION CONCEPTS AND HASIC18LATIONS 251 ‘Then the velocity distribution in the turbulent layer is given as ut =25Iny" +55 — forturbulent layer, y* > 30 (6-38) We now summarize the foregoing velocity distribution laws for turbulent flow along a smooth surface y for viscous sublayer,0< y* <$ | (6-39a) ut =S0Iny* —305 for buffer layer, < y" = 30 | (6-396) ut =25Iny' +55 forturbulent layer, y* > 30 (6-39) where ” and fe (6394) veolp p and 9 is the total wall shear stress Figure 6-7 shows a correlation of the velocity distribution law given by Eqs. (639) with Nikuradse’s [102] measured velocity distribution for turbulent flow inside smooth pipes. ‘Although the velocity distribution law obtained by separating the flow field into three distinct layers appears to be in reasonably good agreement with the experimental data, the transition from a viscous (0 a turbulent flow regime in reality takes place gradually. Therefore, the representation of velocity distribution by three different curves having discontinuous slopes at locations where they join is not realistic. A more serious inconsistency of the logarithmic velocity distribution law Eq. (6-39) is that it does not give zero velocity gradient at the tube center. For this reason, the average velocity for flow inside a pipe as determined by using the above equations overestimates the velocity. Despite these short- ‘comings, the velocity distribution laws given by Eqs. (6-39) have been used ex- tensively in the literature to study the relation between momentum and heat transfer. sae Figure 67 Lopaitunic velocity Toros dation law and Nikuradse's {la} experiment data for ture belee flow ise smoot Pies.252 Wear Tmansnan Effects of Surface Roughness on Velocity Distribution ‘The velocity distribution relations discussed earlier are applicable to turbulent flow over surfaces which are hydrodynamically smooth, A surface is considered hydrodynamically smooth if the heights 2 of the protrusions are much smaller than the thickness of the viscous sublayer. Surfaces encountered in engineering applications generally are not perfectly smooth, and since for most cases the viscous sublayer is very thin, the protrusions may penetrate it. Varied geometric forms of roughness and the variety of ways that the protrusions may be distributed over the surface make it difficult to analyze the effects of roughness on velocity distribution. Nikuradse [106] made extensive experiments with turbulent flow inside artificially roughened pipes over a wide range of relative roughness 4/D (ie, protrusion height-to-diameter ratio) from about yghg t0 7. The sand-grain roughness used in these experiments has been adopted as a standard for the effects of roughness. These experiments showed that to study the effects of rough- ness it s desirable to introduce a roughness Reynolds number 2" (.e.,a dimension- less protrusion height) defined as. 7 (6-40) ‘and when 2” is less than about 5, the roughness has no effect on the friction due to flow. With this consideration three distinct situations are envisioned for the effects of roughness: , Hydrodynamically smooth: 0 < 2* <5 Transitional: oss (641) Fully rough: > For the hydrodynamically smooth case, the heights of roughness are so small that all protrusions are covered by the viscous sublayer; hence roughness has no effect. For the transitional case, the protrusions are partly outside the viscous sublayer and cause some additional resistance to flow. For the fully rough case, ‘the heights ofthe protrusions are so large that all protrusions penetrate the viscous sublayer; hence the viscous sublayer no longer exists, and protrusions influence the turbulent mixing. For the fully rough regime, the logarithmic velocity distribution relation given by Eq. (6-37) is applicable ify* is replaced by y/A; then the velocity profile takes the form y misc (6-42) where x = 0.4 as in the case of smooth wall but the constant C is different. A ‘correlation of this relation with Nikuradse’s [106] experiments for the sand- roughened wall surface conditions has shown that C = 85, Then Eq. (6-42) takes, the form 25nd 48s (643)‘CONVECTION —CONCEPTS AND BASIC RELATIONS. 253, where ‘which is called the logarithmic velocity distribution law for turbulent flow in rough pipes in the fully rough region. ‘A significant difference between laminar and turbulent flow is inthe fact that in turbulent flow the velocity profile is affected by the surface roughness, whereas in laminar flow the roughness has no effect. As the velocity distribution affects the shear sires atthe wall, and hence the friction factor, itis expected that in turbulent flow the friction factor will depend not only on the Reynolds number but also on the surface ronghness. This matter is discussed further in the following chapter. 64 EQUATIONS OF MOTION We showed that to determine the drag and the friction coefficients, the velocity distribution in the flow was needed. The velocity distribution can be determined from the solution of the equations of motion subject to appropriate boundary conditions. The problem of velocity distribution in low for general low conditions is a very complicated matte, but there are several very simple, yet physically meaningful and practical situations for which the velocity distribution. can be determined readily. Our understanding of the physical siiicance Bf various parameters in influencing the friction factor and the drag coefficient can be Improved i we have some knowledge ofthe equations governing the fuid motion ‘With this objective, we present here the equations of continuity and momentum for the two-dimensional, steady motion of constant-property, incompressible newton iam fluid in the two-dimensional, rectangular coordinate system for the x and y variables, We also present the equivalent equations in the cylindrical coordinate system forthe, 2) Variables, and we illustrate the simplification ofthese equations for simpler flow conditions. Continuity Equation ‘The continuity equation is essentially the equation for the conservation of mass; it is derived by a mass balance on the fluid entering and leaving a volume element taken in the flow field. Consider a differential volume element Ax Ay 1 about a point (x, 9) in the flow field, as illustrated in Fig 6-8, The equation for the con- servation of mass for two-dimensional steady flow may be stated as fet rate of mass flow entering), (net rate of mass flow entering v volume element in y direction, shun doneat in denn = 649 Letu = u(x, y)and v= o(, ») be the velocity components inthe flow inthe x and y directions, respectively. If M, = pu Ay 1 is the mass flow rate into the element in254 ExT TRANSFER x Figure 68 Nomenclature fr the derivation ofthe com a tiity equation the x direction through the surface at x, then M, + (@M./8x) Av is the mass flow rate leaving the element in the x direction through the surface at x + Ax. The net rate of mass flow into the element in the x direction is the difference between the entering and leaving flow rates, given by (Net rate of mass flow entering) @M, (pu) (Semen estan) ~~ "yeaa = — 0 ay Ca Similarly, for the y direction, we write Net rate of mass low entering) __a{pe) element in y direction ay Substituting Eqs. (6-45) into Eq. (6-44), we obtain AxAyl (6456) 2) Hon a When deny etd cust Be (6 4) empleo (6-466) Equation (6-46b) is the continuity equation in rectangular coordinates for the steady, two-dimensional flow of an incompressible fluid For problems of flow inside a circular tube, the continuity equation in the cylindrical coordinate system is needed. Such an equation can be obtained from Eq, (6-466) by the standard coordinate transformation, Let uw = u(r,z) and v = ofr, 2) denote the velocity components in the axial z and the radial r directions, respectively. Then the continuity equation for steady, two-dimensional flow of incompressible fluid in the (r, 2) cylindrical coordinates becomes 1a pal + oe 4‘CONVECTION CONCEPTS AND BASLE RELATIONS. 255 Momentum Equations ‘The momentum equations are derived from Newton's second law of motion, which states that mass times the acceleration in a given direction is equal to the external forces acting on the body in the same direction. The external forces acting on a volume element in a flow field are considered to consist of the body forces and the surface forces. The body forces may result from such effects as the gravitational, clectric, and magnetic fields acting on the body ofthe fluid, and the surface forces result from the stresses acting on the surface of the volume element. With this ‘consideration Newton's second law may be stated for flow in direction jas, canon) « (optawe et). (stg) a a for a three-dimensional flow, for example, in the rectangular coordinate system i = x,y, and zshence Eq, (6-48) provides three independent momentum equations. In this analysis, we consider two-dimensional, steady, incompressible, constant- property flow in the (x, )) rectangular coordinate system. Therefore, for i and y, Eq. (6-48) will provide two momentum equations, one for the x direction and the other for the y direction Let u = u(x, ») and p = e{x, y) be the velocity components in the x and y directions, respectively. We consider a volume element Ax Ay 1 about a point (x, »)in the flow field. Various terms in Eq. (6-48) are determined as follows: First, if pis the density ofthe fluid, the mass term in this equation is given by (Mass) = (Ax Ay Dp 49) ‘Second, Eq. (6-48) contains a term called acceleration. Commonly, accelera- tion implies time rate of change of velocity, but for the steady flow considered here, there is an acceleration associated with the convective motion of fluid in other directions, because we have a two-dimensional flow. Consider, for example, ‘motion of the fluid in the x ditection. If = t(x,)) is the velocity component in the x direction, there is an acceleration of the fluid in the x direction associated ‘with the motion of the fluid in other directions, given by* (Acceleration) _ du, du jn x direction) ~ “ax * ° ay (6-50) 1 The derivation of Ba, (6-02) a8 lows Consider u =o,» forthe twostimensional steady fw. The total derivative of is au a ates y Dining bot sides by, we have hich the same at Ea, (60a), Smialy, one obtains Ea, (6-50) by considering the total devivat ofe= sy)256 ear TRANSFER Similarly, if = o(x, y) is the velocity component in the y direction, the accelera- tion of the Auid in the y direction associated with the motion of the fuid in other directions is given by (tsseaion ) \in y direction, (6508) Third, Eq. (6-48) contains a term called body forces acting on the uid. Let F, and F, be the body forces acting per unit volume ofthe fluid in the x and y direc tions, respectively (that is, pg denotes the gravitational force acting per unit volume). Then Gemey =) = Fedde yl) (6514) /Body forces acting) _ (ncaa) = Fide ar uy Fourth, Eq (6-48) contains a term called the surface forces acting on the Suid. ‘The surface forces acting per unit area are called stresses. When the stress acts normal to the surface, it is called the normal stress: when it acts along the surface, itis called the shear stress. Figure 6-9 shows various stresses acting on the surfaces ofa differential volume clement. In this figure ¢, and 6, denote the normal stresses in the x and y directions, respectively. The shear stresses are denoted by t,, and t,,, where the first subscript indicates the axis to which the surface is perpendicular and the second subscript indicates the direction of the shear stress. Thu, t,, is the shear stress acting on the surface Ay I (ie, the surface perpendicular to the x axis) at x in the direction . ‘Then the net normal surface force acting on the element in the positive x direction is (6/0) Ay 1) Ax, and the net shear force acting on the element in the positive x direction is (/6yXCyx Ax 1) Ay. Hence, the net surface forces acting on the element in the positive x direction becomes es surface forces cee) . (@ + x) ede direction ox [+# a) Figure 69 Nomenclature for the various : stresses acting on the surfaces ofthe volume a ementee Sin hn tensor tigi tn yi Ftc ane (2 (Stents) (lr away) es When Eqs. (6-49) to (6-52) are introduced into Eq. (6-48) and Ax Ay terms are canceled, the x momentum and the y momentum equations, respectively, become Nowa: — Stra) a gM Be 60 ronan: — oS vet) ty Bt ee = ‘The final stage in the analysis involves the determination ofthe expressions for ‘various stresses appearing in these equations. Such relations depend on the type of fluid considered, and a discussion of this matter can be found in several references {16-18}. For the two-dimensional, incompressible, constant-property flow and the newtonian fluid considered here, various stresses are related to the velocity ‘components by [16] santana) eo an? +24 6) 0 PS (6-540) ‘here P isthe pressure and js the viscosity in the flow fel. ‘When Eqs (6-54) are introduced into Eqs. (6-53), after some manipulation, ‘one obtains du bu oP (Pu | au sanaan: (et) t,o (29) | os CA oP Fv | Hel yonaton: (eit eo) <1, eof 2) | 50 where F, and F, are the body forces per unit yolume acting in the x and y di- rections, respectively. Equations (6-55a) and (6-55) are called, respectively, the x and y momentum equations forthe steady, two-dimensional flow of an incom pressible, constant-property, newtonian fluid in the rectangular coordinate system. ‘The physical significance ofthe various terms in Eqs. (6-55) is as follows: The terms on the left-hand side represent the inertia forces the fist term on the right- hhand side is the body force the second term is the pressure force, and the last term inthe parentheses is the viscous forces acting on the fluid element.258 wear Tansee If the body forces F, and F, are known, the continuity equation (6-46b) and the two momentum equations (6-55) provide three independent equations for the determination of the three unknown quantities u, v, and P for the steady, two- dimensional flow of an incompressible fluid. The analytical solution of these ‘equations is extremely difficult except for very simple situations. For problems of flow inside a circular tube, for example, the momentum ‘equations are needed in the cylindrical coordinate system. Let u = «r, 2) and = 7,2) be the velocity components in the axial 2 and the radial r directions, respectively. Then in the two-dimensional, (r, 2) cylindrical coordinate system, the equivalent of the momentum equations (6-55) and (6-56) is given by ‘Momeatum: (0% + an 2 de Momentum: (02! + u omentum: oo (6-58) Here F, and F, are, respectively, the body forces acting on the fluid in the r and z directions. For simplicity in the nomenclature we used the symbols u and v to 1, hence the ‘momentum diffusion is much greater than the energy diffusion; but for liquid metals, Pr < 1 and the situation is reversed. We recall that in discussing the development of velocity and thermal boundary layers for flow along a flat plate, the relative thickness of velocity and thermal boundary layers depended on the magnitude of the Prandtl number. ‘Consider the Nusselt number, based on a characteristic length L, rearranged in the form AL _ hAT *& TRATE where AT is the reference temperature difference between the wall surface and fluid temperatures. Then the Nusselt number may be interpreted as the ratio of heat transfer by convection to conduction across the fluid layer of thickness L. Based on this interpretation, the value of the Nusselt number equal to unity implies that there is no convection—the heat transfer is by pure conduction, A larger value of the Nusselt number implies enhanced heat transfer by convection, ‘The Stanton number can be rearranged as Nu (67a) ho har (6-736) where AT is reference temperature difference between the wall surface and the fluid, The numerator represents heat flux to the fluid, and the denominator repre- sents the heat transfer capacity of the fluid flow. ‘The dimensionless parameter, the Eckert number, defined as E = u2,(c, AT), frequently arises in high-speed, hat transfer problems. The Eckert number can be rearranged as ‘dynamic temperature due to fluid motion ‘temperature difference oa Here u2,(2c,) represents an ideal temperature rise if an ideal gas with a velocity tug is slowed down adiabatically to zero velocity. This definition implies that ifthe Eckert number is small, the viscous-energy generation effects due to the motion of the fuid can be neglected in comparison with the temperature differences in- volved in the heat transfer process. We recall that the viscous-energy dissipation term appeared in the energy equation, and the magnitude of the Eckert number ‘becomes the criterion in deciding whether the viscous-energy-dissipation effects should be considered inthe heat transfer analysis. Our understanding ofthe physical significance of the dimensionless parameters is further improved if we examine how these dimensionless parameters enter the equations of motion and energy. Ifthe equations of momentum and energy Are expressed in the dimensionless form, the dimensionless parameters Re, Pr, andE appear in these equations. To illustrate this matter we consider the x momentum, ‘equation (6-55) by neglecting the body forces: ofe 22% 624) 67 or, ‘To nondimensionalize this equation, we select a characteristic length L and a reference velocity 1, and we introduce the following dimensionless variables: P 4 ° oe (616) Here the quantity pu2, represents the double of the dynamic head (that is, 3pu2, is called the dynamic head). By introducing these dimensionless variables into the ‘above momentum equation, the dimensionless form of the x momentum equation becomes ou, /2U_ _ ape ay x 1 (eu | eu z Ge + a) om ‘The left-hand side of this equation represents the inertial forces, and on the right hand side the terms in the parentheses represent the viscous forces. Increasing the Reynolds number decreases the relative importance of the viscous forces in the system. ‘We now consider the energy equation (6-66) written as moles) =HGae +52) ER) 6) + G3) | (6-78) and we define a dimensionless temperature 6 as T-T aT 67) where Ty in ference temperature an AT sa refeenc temperature diftrence, Tne nnonss vibe (ot) and (679) ae intodued to Ba (678) We cha yO 1 (0, ‘GY ~ Re Pr \ax* 2, E (au\? fav\? | fav , au\* +BDG) Gs) + Geta) ] oo Here the Eckert number appears 35a multiplet the vscou-nery-dspation term. Thus, the viscous-energy dissipation can be neglected if the Eckert number is small,2700 ear TRANSTER 6-7 BOUNDARY-LAYER EQUATIONS The analysis of convection heat transfer is complicated by the fact that it requires the solution of equations of motion and energy. For example, the two-dimensional, steady-state formulation considered in Sec. 6-6 involves one continuity, two ‘momentum, and one energy equation, which should be solved over the entire flow field. Ifthe flow field is separated into a boundary-layer region where the velocity and temperature gradients are steep and the potential-flow region, significant simplifications can be achieved in the analysis of convection heat transfer. The reason is that the analysis of flow for the potential-flow region is ‘relatively easy and simplifications occur in the momentum and energy equations when they are restricted to only the boundary-layer region, For the two-dimensional, steady flow, the boundary-layer equations are ‘obtainable from: the continuity, momentum, and energy equations given previously, with the simplification of these equations by an order-of-magnitude study. The basic assumption made in the boundary-layer simplification is that the velocity and the thermal boundary-layer thicknesses are small compared with a characte istic dimension of the body. The reader should consult Ref. 16 for a detailed dis- cussion of such an order-of-magnitude study. Here we present only the results of the simplification of the continuity equation (6-546), the momentum equations (6-55) and (6-56), and the energy equation (6-66). If we assume no body forces are acting on the fluid, the resulting boundary-layer equations are given by Comin oan Monet: oe os ar cme publ okE oo U__ We note that the continuity equation remains unchanged. The y momentum equation is no longer needed, because since the boundary layer is thin, it reduces to dP/dy = 0, which merely implies that the pressure across the boundary layer is practically constant. The x momentum equation no longer contains the term 11 0%e/0x°. In the energy equation, the axial conduction term k @7/éx is dropped, and the viscous-energy-dissipation term is drastically reduced. Clearly, ths is a significant simplification ofthe governing equations, ‘The momentum equation (6-82) contains the pressure-gradient term dP/dx. Eq, (6-82) is evaluated at the edge of the velocity boundary layer where u(x, y) becomes equal to the external stream velocity u..(x), then Eq. (6-82) reduces to Pde 2 pute St a(CONVECTION CONCEPTS AND BASIC RELATIONS. 271 ‘This equation relates the pressure-gradient term to the external stream velocity tug(X), which is assumed to be available from the solution of the potential flow problem, Thus dP/dx term in Eq, (6-82) is a known quantity In the case of flow over a flat plate, the external flow velocity u., is constant; then ap qn (685) and the pressure-gradient term in Eq. (6-82) vanishes. Example 6-17 The drag coefficient for laminar flow along a flat plate is 10 be determined. Write the governing diferential equations needed to solve this problem, SOLUTION If the velocity distribution u(x, ») in the flow is known, the shear stress atthe wall and hence the drag coefficient can be established. The govern ing equations for the determination of the velocity distribution in laminar flow along a flat plate are obtained from Eqs. (6-81) and (6-82) as au, a ax * ay ‘These equations should be solved subject to appropriate boundary conditions for the flow. PROBLEMS Flow oer aby {61 Compare the distance from the leading ede of ist plate at which the tanstion curs rom lamina otrbulet fw fr attoupheri si at 37°C with) 2 (8) 1, and (c) 20 ms. (Assume trans tion takes place at Re, = $ » 10°) “Answer: (a) 421 (8) 0882 me) 0421 m ow-spect wind tunel aif at atm and 27°C with velocity of m/s, How longo he Mapa test section be produce a transition Reynolds mum ofS = 10"? 163 Determine the distances from the leading edge ofa fat plate at which transition oocur from lamina otrbuleat forthe ow fair at TPC witha velsity of, = 20 m/s at presure of, land 4 am (Assume Re, = 510) ‘Anomer O19 mor the ase of atm 64 Assoming th ranston fom laminar to turbulent flow takes place at a Reynolds number $x 10, direction ie neplibe in comparison with heat conduction in the iveton and convection i the: dtsction. (645 Further simplify the energy equation obtaincd in Problem 6-4 by assuing thal the viscous ‘nergy dissipation i nelpble, Discus the physclsigaieanc ofeach em ithe simplified equation, ‘What kind of equation results inthe fow veloity becomes zero? Diseuss under which conditions the scousenegy-dsipation er is negligible {646 Consider steady fow ofa Aud between two parallel plates with wand» being the velocity com- ‘ponents i the axial x andthe normaly directions, respectively. Simply the energy equation inthe {ectangular coordinates given in thi chapter for fow betwee two paral plates under the following sumptions (@) The Now i aly develope, (©) The tomperatre gradients inthe ail x diretion are negligible ‘Discus the physica sipiicane of ach term inthe resulting energy equation. (647 The steady-state eneray equation for ow between two parallel pats ie taken at a 27, (%) ae hae Mos) where wisthe Row velocity in thea ‘uation to this form. (648 The steady-state energy equation for fow betwee two parallel plate is given inthe form re terest fw weoiyin oi etn Discs hsp made singly heey Trstoe totus (0 The seedy sate ener suatonr fo been to pr pats ven efor et 2 lear at” ay‘CONVECTION—CONCEPTS AND BASIC RELATIONS. 277 where wsthe flow velocity in the ail x diestion, Discuss the assumptions made to simplify the energy ‘auation to this form, {€50 The steady-state energy equation for Bow inde a crear tubes gven in the form won at) of) Dimensionless parameters {651 Consider the x ection momentum equation given by +3) where Lis a reference length andi is a reference velocity, express this momentum equation inthe Aimensiones form (652 Consider the ydzeon momentum equation given inthe form (neem ee) Pant Pay)” ~ ay * Ma * ap) Expres this equation inthe dimensionless form by uiizag the dimensiones variables defined in Problem 651 (653 Consider the direction momentum equation inthe cylindrical coordinates given by omen) Zee) oa] By intoscing the following dimensionless variables 7 P . “ pat. es Re cP me PU, where Lisa reference length and igi reerence velocity, express this momentum equation inthe Simensiones form. (6-54 The energy equation for flow between two pale! plates is aT) _ eT) aa and the following dimensionless variables are defined: eet oz Py) Oye eS oe fees where Lup. Ty. and AT are, respectively a reference length (., the spacing between the plates, ‘coc temperatere and temperature difrence, and isthe thermal dfs BY uliling thee dimensiones variabis,exprest the above energy equation i dimensionless form. {655 The energy equation for Now inside acca ubeis ed 18 (2 poy TED a4 2 22) Tey) = Te ar278 wear TRANsrER and the following dimensiones variables ae defined W)opet gg Thad h * tae ar v where Lu. Tov and AT are, respectively a tference length, velocity, temperature, and temperate iflerence and 2 the thermal diosvy By uilling these densionless variables. express the above energy equation ithe dimension- les form. {6-56 The energy equation fr ow between tno parle platsis By uilzing the dimensionless variables defined in Problem 654, express this nergy equation in ‘dimensiones form nd show thatthe flowing wo dimensionless groups will appear inthe equalion = = tele number ear Pete Prato pnd ner Discus the physical significance ofthe Eekert umber in relation to the relative importance of the ‘viscous enery-dssipation term in the energy exuation. (657 The energy equation for flow inside a cca tube is given inthe form 2 (,27) , fan a) Ce By oiling the dimensionless variables defined in Problem 6:5, expres this energy equation dimensiones form and show thatthe following tw dimensionless groupe will appear the equation: are.2) yo) Ee = th = Ester number Pr = St =F Prandl number t Discuss the physical significance ofthe Eckert number. oundary-laer equations {658 Discuss the phi significance ofthe velocity and thermal boundary layers. {659 Discuss the physical signicance ofthe dimensionless groupe Prandtl. Reyaoks, and Eckert, umber {6-60 Discus the approximations and the simplifying assumptions made in obtaining the boundary layer equations rom the general momentum and energy equations. {61 Discuss the physical ignicance ofeach erm in the boundary layer x momentum equation given in thi chapter, {662 Discuss the physical significance ofeach erm in he boundary ayer ener eatin given in his chapter {643 Explain why the presure gradient term dpi wll equation for fow along Rt plate {664 Consider the boundary-layer energy equation without the visous- Te—depends on the magnitude of the parameter Pr. E. For Pr-E > 2, the right-hand side of Eq. (7-13) is positive; hence the heat flows in the positive y direction, or from the liquid into the wall, even though the "upper wall is ata higher temperature than the lower wall For Pr-E <2, the right-hand side of Eq. (7-13) is negatives hence the heat flows isin the negative y direction, or from the upper wall into the fluid. For Pr-E =2, the right-hand side of Eq, (7-13) vanishes; hence there is no heat flow at the upper wall ‘Second, when Ty = T,, both plates are at the same temperature, and (7-7) reduces to 1) ~ =H? (1-2) ca) ‘The maximum temperature in the fluid occurs at the midpoint between the plates; by setting y = L/2, Eq, (7-14) becomes ris) By combining Eqs. (7-14) and (7-15) the temperature distribution in the fuid is expressed as TO) = To ar (7-16) Example 7-1 A heavy lubricating oil [4=0.25 kgim-s), k=0.125 Wim-°C)) at room temperature flows in the clearance between a journal and its bearing, Assuming both the bearing and the journal are at the same temperature, determine the maximum temperature rise in the fluid for a velocity of uy ms.286 wet TaNeren Sovutiow The maximum temperature rise ATya for Tp = Ti is obtained from Ea, (7-15) as pa ORE Introducing the numerical values, we obtain ke V/gem)( 1 m-% (0253 1G = (ois W ) c Nem=1Wis. since 1 kg: m?/s? Flow Inside a Circular Tube ‘The problems of steady-state heat transfer and pressure drop in laminar forced ‘convection inside a circular tube in regions away from the inlet where the velocity and temperature profiles are fully developed are of interest in numerous engineer- ing applications. The frition factor and the heat transfer coefficient for the flow are determined, respectively, from the knowledge of the velocity and temperature distributions in the uid, Fiction factor Consider an incompressible, constant-property fui in laminar foreed convection inside a circular tube of adi in the region where the ow i hydrodynamically developed. The fiction factor for flow inside a circular tube istlated tothe velocity gradient athe wally Ea (618) 3s -* Fal TOC Cau The velocity distribution u(r) can be determined from the solutions ofthe equations of motion. In Example 6-12, itis shown that for hydrodynamically developed laminar flow inside a circular tube, the equations of motion reduce to a simple ‘equation given by Eq. (6-60), which is written in the form a(t) 1a 14 (4) UP nacre ow cnt te ound cents ‘i or atr=0 (7-190) u atreR (7-196) The first boundary condition is the symmetry of the velocity profile about the tube axis, and the second is that the velocity vanishes at the wall.FORCED CONVECTION FOR FLOW INSIDE DUCTS 287 For steady fully developed laminar flow inside a circular tube, the pressure gradient dPjd7 is constant. Then the solution of Eq. (7-18) for constant dP/de, subject to the boundary conditions (7-19), gives the fully developed velocity rete (ef -()] o~ Here the velocity (?) is always a positive quantity because for flow in the positive 2 direction, but the pressure gradient dP/dz is a negative quantity. ‘The mean flow velocity ug over the tube cross section is determined from the definition as Wo) faa R'aP wen gh [fama - EE oan since (is given by Eq, (7-20) “The physical significance ofthe mean velocity, implies that the flow rate through the tube is determined fom Flow rate = (area cross section) tig = Rly ‘Now from Eqs. (7-20) and (7-21) we obtain [EtG] = This relation shows that the velocity profile u()/uq in the hydrodynamically developed region is parabolic, The velocity uy at the tube axis is obtained from Eq, (7-20) by setting r = 0; RP Gude ‘A comparison of the results given by Eqs. (7-21) and (7-23) shows that the velocity at the tube axis is equal to twice the mean flow velocity; (723) io 724) ‘The friction factor ffor laminar flow inside a circular tube in the hydrodynamically developed region is determined by obtaining the velocity gradient from Eq. (7-22) du) iy __ By 5 dr ep ORO 25) and introducing this result into Eq, (7-17), We find f (7-260)where D is the tube inside radius and = nD _ tind (7-266) eo Re is the Reynolds number. In the literature, the friction factor also has been defined on the basis of ‘hydraulic radius, If f, denotes the friction factor based on the hydraulic radius, i is related tothe friction factor f defined by Eq. (7-26a) by f= 4f,. That is, Eq. (7-264), ‘on the basis of J,, would be J, = 16/Re, where Re = py, D/. This result is some- times referred to as the Hagen-Poiseuille relation for friction factor in tubes, because Hagens [1] experimental data were later verified theoretically by Poiseuille [L ‘Heat transfer coefficient The heat transfer coefficient for flow inside a circular ‘ube in the thermally developed region is related to the dimensionless temperature gradient at the wall by Eq. (6-235) as eu) (7-28) To determine h the temperature distribution in the low is needed, and it ean be established from the solution of the energy equation, Tn the hydrodynamically developed region, the energy equation for laminar flow of an incompressible fluid inside a circular tube with negligible viscous-energy dissipation is discussed in Example 6-15 and given by Eq. (6-71a) as 1 eri eT) #T a oe ae (3) +e om In general, this isa partial diferental equation fr the determination of tempera- tre distribution inthe flow, and its solution is rather involved, However, for forced convection inside a circular tube in the thermally developed region, with constant wall temperature or constant wall eat flux, it can be shown [3,4] that the axial temperature gradient term in Eq (7-29) becomes a constant; that i, 2. asst ‘Then the partial diferential equation (7-29) reduces to an ordinary differential equation for the fully developed temperature profile T(), because the term 227/02? vanishes for constant 27 /2z. We now examine such a heat transfer problem for the constant wall heat ux and constant wall temperature boundary conditions for forced convection inside a circular tubeFORCED CONVECTION FOR FLOW INSIDE DUCTS 289 Constant wall heat flux In Ref. 3 it has been shown that under constant wall heat flux conditions, the temperature gradient in the direction of flow anywhere in the flow is constant and equal to the axial gradient of the mean temperature of the fluid. That is, a ene ©) = constant 730) ‘This result implies that under constant wall heat flux, the mean fuid temperature (2), in the thermally developed region, increases linearly with the distance = along the tube. ‘When Eq, (7-30) is introduced into Eq, (7-29), the term 0°7 /2z* vanishes for constant 27/42, and the following ordinary differential equation is obtained for Tey: 1a (ar aT) ral) 30% oa ‘This equation is written in terms of the dimensionless temperature 6(7), defined by Eq. (7-28), as 14 (4) 1d i 12(-4) fdr) EO (re ~ 1409) (7-324) ‘where the fully developed velocity profile u(r)is given by Eq. (7-22) as ee ‘Equations (7-32a) and (7-326) are combined and written more compactly as d (a0 r\r E(ij=nfi-(Q)] wocecr an se oa dean a a[T,(2) ~ TA2)] dz tant 7-33) ‘The boundary conditions for Eq, (7-33) are taken as 25 ura on) tno ark as ‘The first boundary condition states that 0 has symmetry about the tube axis, and the second results from the definition of @ given by Eq, (7-28) that 6 must be zero at the wall,290 Hear TmANSTER Equation (7-33c) is similar to the steady-state heat conduction equation in cylindrical coordinates, and it can readily be integrated subject to the boundary conditions in Eqs. (7-34) to give ar~-ae [Zak (Y-H()] aan ‘The unknown constant 4 appearing in this equation can be determined by utilizing the definition ofthe bulk mean fuid temperature. According to the definition of the bulk mean fluid temperature given by Eq. (6-225), we write _ opine dr Sn yg BRE ‘where the fully developed velocity profile u(r) is given by Eq, (7-32h), that is, u(r) = aft - (i) | (7-37) Equations (7-35) and (7-37) are introduced into Eq. (7-36), and the integrations are performed. We obtain (7-36) 0, = War? (7-380) Als, fom the definition of) given by Ea (7-28), we write Tse) ~ Tae) - 2-388 fn T= Tee) ed) ing (7-38) and (7-38), we find AR = #9 7-8) Introducing this result for AR? into Eq. 7-35) we obtain 9603, 1 fr\*_ Afr)? a9 = +e) -4(0) 0) Equation (7-40) is the dimensionless temperature profile for forced convection in a circular tube in the hydrodynamically and thermally developed region under the constant wall heat flux boundary condition. We recall that this temperature profile ‘was used in Example 6-9 to determine the heat transfer coefficient. Given the temperature profile in the fluid, the heat transfer coefficient h is immediately obtained from its definition given by Eq. (7-27) 48 k “he (7-41) nD _ 48 Now BaF = 4364 (7-41b) where D is the tube's inside diameter and Nu is the Nusselt number.FORCED CONVECTION FOR FLOW INSIDE DUCTS 291 ‘The result given by Eqs. (7-41) represents the heat transfer coefficient for laminar forced convection inside a circular tube in the hydrod ynamically and thermally developed region under constant wall heat flux boundary condition. Constant wall temperature The heat transfer problem described above for the hhydrodynamically and thermally developed region also can be solved under con- stant wall temperature boundary condition; but the analysis is more involved and is not presented here. The result is Nuss D = 36 7-42) which represents the Nusselt number (or the heat transfer coefficient) Jor laminar forced convection inside a circular tube in the hydrodynamically and thermally developed region under constant wall temperature boundary condition. Evaluation of physical properties Inthe results given by Eqs. (7-1) and (7-42), the thermal conductivity of the uid k depends on temperature. When the fui temperature varcs along the tube, k may be evaluated at the fuid bulk mean temperature ,, defined as Mn+ (7-43) ‘where 7; = bulk fluid temperature at the inlet and T, = bulk fluid temperature at the outlet, ‘Thelogarithmic and arithmetic mean temperature differences, The logarithmic mean (LMTD) of the two quantities AT, and A7; is defined as AT, = (7-40) ‘whereas the arithmetic mean (AM) of AT, and A7; is defined as. Tau = KAT, + AT) (7-446) In Chap. 11 we show that for the heating or cooling of fluid flowing inside a duct, the appropriate mean temperature difference between the fluid and the tube wall is the LMTD, where AT, = T, ~ Ti. = inlet temperature difference AT, = Ty ~ Tou = outlet temperature difference ‘Note that AT is always less than ATyy. I'the ratio of AT, and AT; is not greater than 50 percent, then ATj, can be approximated by the arithmetic mean difference within about 1.4 percent.292 MEAT mansrER Example 7-2 Engine oil is pumped with a mean velocity of tq = 0.6 m/s through a bundle of n = 80 tubes each of inside diameter D = 25.m and length L = 10 m. The physical propertis ofthe oil are v = 0.75 x 10™* m'/s and p = 868 kg/m’, Calculate the pressure drop across each tube and the {otal power required for pumping the oil through 80 tubes to overcome the fluid friction to flow. ‘SoLUTION The Reynolds number for the flow is 1D _ (0.640.025) vy 075x10 ‘The flow is laminar, hence the friction factor f is computed by Eq, (7-26): Ce) Re ~ 200 ‘The pressure drop across a tube, according to Eq. (6-192), becomes Louk aP=S5 Re = 032 0.023, 2 = 2x 10 N/m? The flow rate through the 80 tubes is M= ce >on _ o20( 1.) g/m?)(0.6? m/s?) = 80 (025 m"y06 ms) = 2356 « 10-* ms ‘The pumping power requirement is determined by (6-196): we ara (0S) (56 «09% -oakn = (ma™ "\GaRs) 063 hp Example 7-3 Consider the heating of atmospheric air flowing with a velocity of u, = 0.5 mjs inside a thin-walled tube 2.5 cm in diameter in the hydro- dynamically and thermally developed region. Heating can be done cither by condensing steam on the outer surface of the tube, thus maintaining a uniform surface temperature, or by clectric resistance heating, thus maintaining aFORCED CONVECTION FOR FLOW Iss DUCTS 299) uniform surface neat flux. Calculate the heat transfer coefficient for both of these heating conditions by assuming air properties can be evaluated at 350 K. Souution The air properties at 350 K are v= 20.76 x 10° m?/s 0.03 W/m - 8) The Reynolds umber forthe ow is nD _ (053002), vy 2076 x 10° Hence the ow is laminar. The Nusselt number for laminar fw inside a Cirulr bein the hydredyoumiely end thermally develope pion ice by Eqs. (7-41) and (7-42), respectively, for the constant wall heat flux and constant wall temperature boundary conditions. Therefore, the heat transfer coefficients for these two cases are determined as follows: Heating by condensing steam: 602 yg 08 Wis 346 35008 Ma 8 = 4.39 Wim? -°C) Elec estan ting 0.03 Wms). b= 4364 0025 m 4364 = 524 Wim? °C) Flow Inside Ducts of Various Cross Sections ‘The Nusselt number and the friction factor for laminar flow in ducts of various «ross sections have been determined in the region where velocity and temperature profiles are fully developed. I the duct cross section for flow is not circular, then the heat transfer and friction factor, for many cases of practical interest, can be based ‘on the hydraulic diameter D,, defined as 44 ; (7-45) P where 4, = cross-sectional area for low and P = the wetted perimeter. Then the [Nusselt and Reynolds numbers for such cases are Dy mu =2 (7-460) Dy (7-466)294 HEAT TRANSRER ‘The basis for choosing D, asin Eq. (7-44) is that fora circular tube D, becomes the tube diameter D, since 4, = (r/4)D* and P = xD. ‘The Nusselt number in the hydrodynamically and thermally developed region, {for laminar flow inside a circular tube was given earlier for two different boundary conditions, namely, the constant surface temperature and the constant wall heat ‘flux. However, for geometries other than the circular tube and parallel plates, the constant wall heat flux boundary condition can involve two distinct limiting First, the surface heat flux in the flow direction is uniform while the surface {temperature remains uniform around the periphery at a flow cross section. This situation arises in ducts having highly conductive materials resulting in a negligible wall thermal resistance. The reason is that highly conductive material tends to make the temperature around the perimeter at any cross section uniform. Except for a circular tube and a parallel-plate duct, the flow resistance around the duct is not uniform, Second, the surface heat flux both in the low direction and around the periphery js uniform. This situation arises in the ducts having walls of very low conducting material, To distinguish the Nusselt number for the three different cases—the ducts subjected to uniform surface temperature and the ducts subjected to two different cases of uniform heat flux boundary conditions —we adopt the following notation: Nur Nuys Nusselt number for uniform surface temperature Nusselt number for uniform surface heat flux in the flow direction while the surface temperature remains uniform around the periphery Nuy2 = Nusselt number for uniform surface heat fux both in the flow direction and around the periphery Shah and London [5] have compiled these three different types of Nusselt number and the quantity f-Re (ie. the product of the friction factor and the Reynolds number) for laminar flow through ducts of various cross sections in the hydrodynamically and thermally developed region. Their results are listed in Table 7-1. We note that only for flow inside a circular tube and a parallel-plate ‘eometry are Nuyy and Nuy: the same. These results are strictly applicable in the region where the velocity and temperature profiles are fully developed, The determination of hydrodynamic and thermal entry lengths is discussed next. Example 7-4 Air at atmospheric pressure and with a mean velocity Of tg = (0.5 m/s flows inside thin-walled, square cross-section ducts of sides b = 25cm. The ait is heated from the walls ofthe duct, which are maintained ata uniform temperature by condensing steam on the outside surface, Calculate the friction factor and the heat transfer coefficient in the hydrodynamically and thermally developed region. Air properties can be evaluated at 350 K. SOLUTION The air properties at 350 K are 20.76 x 10-$ m/s k = 0.03 Wiim-s)FORCED CONVECTION FoR FLOW IRSIDE DUCTS 295 Table 7-1 Nusselt number and friction factor for hhydrodynamically and thermally developed laminar flow in ducts of various cross sections* Geomeiry 0, > 100) Noy Nu Nuys FRE 3657 43st 4364 a0 241s 8333 297% 3608 3.001 S691 3391 4123 3017 @220 ne ee S397 6490 2904236 1s) eas 82359600 Game sol 585 600 * From Shah and London (5) The hydraulic diameter of the duct is 4b = 25em and the Reynolds number becomes tins __(0.5K0.025) Le v 20.76 x 10° co296 vba TRANSFER From Table 7-1, for a square duct we obtain f Re = 5691 5691 2 J = Fie = 945 x 10 and Nuy = 2976 sore & = 2976 003 Wilen- °C) = 29765, = 2916 Hydrodynamic and Thermal Entry Lengths It is of practical interest to know the hydrodynamic entrance length Ly and the thermal entrance length L, for flow inside ducts The hydrodynamic entrance length L, is defined, somewhat arbitrarily, as the length required from the duct inlet to achieve a maximum velocity of 99 percent of the corresponding fully developed magnitude. ‘The thermal entrance length L, is defined, somewhat arbitrarily, as the length required from the beginning of the heat transfer section to achieve a local Nusselt number Nu, equal to 1.05 times the corresponding fully developed value. Figare 7-3 Hydeodynamic and thermal entrance lengths: a) Re tanaer stats tthe duct inlet, () het transfer stare an thermal eastonFORCED CONVECTION FOR HLOW mibe DUCTS 297 If heat transfer to the fluid starts as soon as fluid enters the duct, both the velocity and thermal boundary layers begin to develop immediately, and Ly and L, are both measured from the tube inlet, as illustrated in Fig. 7-3a In some situations the heat transfer to the Muid begins after an isothermal ‘calming section, as illustrated in Fig. 7-36. For such a case, Lis measured from the duct inlet because the velocity boundary layer begins to develop as soon as the ‘uid enters the duct, but L, is measured from the location where the heat transfer starts, because the thermal boundary layer begins to develop in the heat transfer section. ‘The hydrodynamic and thermal entrance lengths for laminar flow inside ‘conduits have been given by various authors [3, 5-8]. We present in Table 7-2 the hydrodynamic entrance length L, for laminar flow inside conduits of various cross sections based on the definition discussed previously. Included in this table are the thermal entrance lengths for constant wall temperature and constant wall heat flux boundary conditions for thermally developing, hydrodynamically developed flow. In this table, D, is the hydraulic diameter, and the Reynolds ‘number is based on the hydraulic diameter. We note from Table 7-2 that for a given geometry, the hydrodynamic entry Jength L, depends on the Reynolds number only, whereas the thermal entry length ‘Table 7-2 Hydrodynamic entrance length Zand thermal entrance length L, for laminar flow inside ducts* LoiDy Constant wall Constant wal Geometry Re temperaiore heat fu Go 0036 0033 008, an ra oon 0.008 oon = 02s ors 00st aoa 0% oos 00 aos 19 09 oo 0.066 * Based on the results reportd in Refi. $10 8 The thermal ent) leoghs ae for the hydrodyeamialy developed, thermally developing flow conditions.depends on the Péclét number Pe which is equal to the product of the Reynolds and Prandt! numbers. Therefore, for liquids having a Prandt! number of the order of unity, L, and L, are of comparable magnitude. For fluids, such as ols, which have ‘large Prandtl number, L, > Ly; and for liquid metals which have small Prandtl number, L, < Ly ‘The thermal entry lengths given in Table 7-2 are for hydrodynamically de- veloped, thermally developing flow. As we discuss later, in many cases the velocity and temperature profiles develop simultaneously atthe entrance region. Sucha low is called the simultaneously developing flow. The thermal entry lengths for a simultaneously developing flow also depend separately on the Prandt| number. For exampk, for simultaneously developing flow inside a circular tube, under constant wall temperature, the thermal entry length L, is lL pig 007 for Pr = 07 which should be compared with L Bog 70033 for Pr+o0 Which corresponds to that given in Table 7-2 for the hydrodynamically developed, thermally developing flow. Thus L, increases with decreasing Prandtl number and a weak function of the Prandtl number for Pr > 007 Example 7-5 Determine the hydrodynamic and the thermal entrance lengths in terms of the tube inside diameter D for flow at a mean temperature Ty = 60°C and Re = 200 inside a circular tube for mercury, air, water, ethylene ‘glycol, and engine oil, under constant wall heat fux boundary condition. ‘SOLUTION The hydrodynamic entrance length L, for laminar flow inside a circular tube, is obtained from Table 7-2 as Ly 0.056 Re D = (0.056,200)D = 11D ‘Thus, Ly is approximately 11 diameters from the tube inlet for all the fuids considered here. ‘The thermal entrance length, given heat transfer under the constant wall heat flux boundary condition, is obtained from Table 7-2 as L, = 00083 Re Pr D = (0.043)(200) Pr D = 86 Pr Donce CONVECTION FOR FLOW INSIDE DUCTS 299 Here L, depends on the Prandtl number, and for the fuids considered in this ‘example it is determined as follows: 4 Fluid Pr B oo ot 0 6 3 2% Ethylene geal 5080, Engine ol 1050 9080 ‘We note that for low at Re = 200 the thermal entrance length varies from a fraction of the tube diameter for mercury to about 9000 diameters for engine oil, while the hydrodynamic entrance length is about 11 diameters for all the fluids considered here. 7-2 THERMALLY DEVELOPING, HYDRODYNAMICALLY DEVELOPED LAMINAR FLOW Consider the flow of a fluid inside a duct, a ilustrated in Fig. 7-3, in which there is an isothermal section to allow for the velocity development before the fluid centers the heat transfer zone. As illustrated in Example 7-5, for fluids having a large Prandtl number such as oils, the hydrodynamic entrance length is very small in comparison with the thermal entrance length. In the analysis of heat transfer for such situations, itis reasonable to assume a thermally developing but hydro- dynamically developed flow. For flow inside a circular tube, for example, if one ‘neglects the axial heat conduction and viscous-energy dissipation in the fluid, the ‘governing energy equation is obtained from Eq. (6-7la) of Example 6-15 [or Eq. (6-716) of Example 6-16] as 1 gated) _ 18 (ar aa ae (" 2) (410) and u(r) is the fully developed velocity distribution in flow and is given by Eq. (7-22) 5 womnfi-(3)] oo where R is the inside radius ofthe tube. ‘The solution of the partial differential equation (7-47) subject to appropriate boundary conditions at the tube inlet and wall surface gives the temperature distribution T(r, 2) in the flow. Given the temperature distribution, the local heat300 seat rRansreR transfer coefficient h(z) and the local Nusselt number Nu, can be determined Once (2) and Nu, are available, their average values hy and Nu from 2 = 0 to 2 = Lalong the conduit can be computed really. A classic solution of laminar forced convection inside a circular tube subject to uniform wall surface temperature was given by Graetz [9] in 1885 and later quite independently by Nusselt [10] in 1910. The reader should consult Refs 3 and 11 to 13 for a diseussion of the original Gractz problem. A vast amount of literature [14-32] now exists on the extensions ofthe Graetz problem for boundary conditions other than the uniform surface temperature, geometries other than a Circular tube, the effects of energy generation, and viscous dissipation in the fluid ‘An extensive compilation of literature on the subject of foreed convection inside ‘ducts is given in a monograph by Shah and London [5] Figure 7-4 shows the local and average Nusselt numbers for thermally de- veloping, hydrodynamically developed laminar flow inside a circular tube plotted against the dimensionless parameter (x/D)(Re Pr), where x isthe axial distance along the conduit measured from the beginning of the heated section. The inverse ‘of this dimensionless parameter is called the Graet= number Ga: xD Re Pr Jn this figure, the Nusselt numbers are given for both constant wall temperature and constant wall heat flux boundary conditions. We note that the asymptotic values of the Nusselt numbers for the constant wall heat flux and constant wall temperatures ar, respectively, 4364 and 3,66, These values are the same as those siven previously by Eqs. (7-416) and (7-42) for the region where the flow is both hydrodynamically and thermally developed. Figure 7-5 gives the local and average Nusselt numbers for thermally de- veloping, hydrodynamically developed laminar flow between parallel plates plotted against the dimensionless parameter (x/D,)(Re Pr), where D,, is the by- 10, which is consistent with the restriction Eq. (7.52ch ‘The results given in Table 7-3 show that the theoretical prediction of the Nusselt ‘number for simultaneously developing flow given in Fig. 7-1 is reasonably accurate and includes the effects ofthe Prandtl number. Example 7-7 Engine oil is cooled from Ti, flowing with a mean velocity of u, = 0.04 m/s through a circular tube of inside diameter 2.5 em. The tube wall is maintained at a uniform temperature T, = 40°C. Determine the tube length L. SOLUTION The physical properties of oil atthe bulk mean temperature + Tog _ 120 +80 > 2 are taken as ep = 200S/(ke-C) p= 840 kg/m? —Pr= 276 v= 02x 10-*m)s k= 0137 Wm-°C) Then the Reynolds number becomes ‘nD _ (00040025) y 02x 107 ‘hence the flow is laminar. ‘We treat this problem as an entrance region heat transfer problem of simultaneously developing flow. Figure 7-7 or Eq. (7-49) or (7-51) can be usedto determine the Nusselt number. However, to perform these calculations, we need the Gratz number Re+Pr _ (50276) _ 13800 b>” L/D LD Here, since the tube length L is unknown, Gz cannot be determined. ‘Another relation is obtained by writing an overall energy balance for L as Ge a) (j2°)omseokta Ta) = MDL) Ta Cy We take AT, as the logarithmic means of Tjy ~ Ty and Toy ~ Ta that is, ee Me IS[CTy = TayTex = TO) OTe = Bw = TO Equation (bs rearranged in terms of dimensionless parameters: Be HUNT, = Tad = Now (5) Tx o ‘The numerical values are (60)276)4)(120 ~ 80) = nu,(5) ar, where an 120 ~ 80 = in [0 — 400 — AOI (Or, solving for Nu, we obtain su 2013 ite = Tp o Equations (a) and (¢) can be used in conjunction with the appropriate cor- relation for the Nusselt number, and the two unknowns L/D and Nu, can be found, ‘We use the Sieder and Tate equations (7-51) for this purpose. Introducing Eqs. (a) and (e) into Eq, (7-51), we obtain 2391.3 _ go(13800)""(0.17)""* 7 Lip) \o2%, where p= 0.17 and j,, = 0.21 are the viscosities evaluated at the fluid bulk ‘mean and the wall temperatures, respectively. Solving for L/D, we obtain Lo oa or = 10390029 = 103 An iterative solution is needed if Fig, 7-7 is used.310 HEAT TRaNsrER 7-4 TURBULENT FLOW INSIDE DUCTS ‘Turbulent flow is important in engineering applications because it is involved in the vast majority of ud flow and heat transfer problems encountered in engineer- ing practice. Here we show how to determine the friction factor and heat transfer in turbulent flow inside conduits Friction Factor and Pressure Drop Consider fully developed turbulent flow at a mean velocity of vq through a circular tube of inside diameter D. The pressure drop AP over the length L of the tube can bbe determined according to the relation given by Eq, (6-192) apasiee ON 7-33) ‘where /'s the friction factor for turbulent flow. The friction factor for laminar flow inside a circular tube can be found by a purely theoretical approach and has been shown to be f = 64/Re. In the case of turbulent flow, however, some empiricism is introduced in its derivation because a semiempirical velocity profile is used in the analysis. It is instructive to describe here the development of the relation defining the friction factor in turbulent flow. A force balance over the length L ofa circular tube, shown in Fig, 7-9,is written rDig= ap (7-58) Where tq is the total shear stress at the wall, Combining Eqs. (7-53) and (7-54), we obtain Sen eet Pin Col Jaa (7-55) igure 79 A force balance over ength L of tubeFORCED CONVECTION FOR FLOW INSIDE DUCTS 311 where 12 = t= dimensionless mean velocity vtele ‘We assume that the velocity distribution for turbulent flow is given by Eq. (6-39), that is, 2SIny* +55 (7-56) where dimensionless distance measured from tube wall Te average vale of oe he cost etn of he Now trough the abe deermnch uz asin(® fi) +125 os where R isthe tube radius. The term /a/p appearing in this expression is obtained from Eq. (7-55) as fo, ff ie wi ie Equation (7-$8) is introduced into Eq, (7-57) eva (eo 1 = 25In(Re Vf) ~ 25m zs) Ls ae je * = 5.156 log (Re \/7) — 2582 (1-59) Equation (7-59) is substituted into Eq, (7-55). The following expression results for the friction factor f° 1 vi where Re = taD/v = Reynolds number, Equation (7-60) is derived by utilizing @ Semiempirical velocity profil. A comparison with experiments suggests that Ea (7-60) should be modified as follows: 035 log (Re \/7) ~ 091 (7-60) 1 = 20 log (Re Vf) - 08 (61a) vi312 Mexr TRANSFER J

10°. For smooth pipes, simpler but approximate analytic expressions have been sven for the friction factor in the form 0316 Re~°* for Re <2 x 10* 7-82) Fm 0184 Re? — for2x 10 < Re<3x 10 | (7-63) ‘These results apply to hydrodynamically developed turbulent flow. The hydro- ‘dynamic development for turbulent flow occurs for x/D much shorter than that for laminar flow. For example, hydrodynamically developed flow conditions ‘occur for x/D greater than about 1010 20.pon worg) said serous psu mo 1} doupaisoud 30) (2/34) (a/TIS = AV HOHE! M288 14 00) WON TTL AmB nanan os ey ye hk et 1000 too sto Joon woes coo as314 ear rRanseeR Example 7-8 Atmospheric air at T, = 300 K and a bulk stream velocity of 4, = 10 mys flows through a tube with D = 25 inside diameter. Calculate the pressure drop per 100-m length of the tube for (a) a smooth tube and (4) ‘commercial steel tube. SoxuTion The physical properties of atmospheric air at 300 K are P= LITIAKgim® y= 1684 x 10°€ ms Then the Reynolds number is tgD _ (10302025) Ren = i684 x 10 ‘The flow is turbulent. The relative roughness of the comunercial steel tube is 4 _ 0.0045 Br ag = 00018 ‘The friction factors fat Re = 14,846 for the smooth and commercial steel ‘tubes are determined from Fig. 7-11 as (0028 for smooth tube 4 {pos fosommer ae be (@) The pressure drop for the smooth tube is =f EO gpg 100 (LATIAXIO®) _ kN, APS py = OB aG5 y= OS (©) The pressure drop forthe commercial steel tube is = nosis 100 (1774910) kN ‘ap = caste, ITN) ap BN Heat Transfer Coefficient ‘Since the analysis of heat transfer for turbulent flow is much more involved than. that for laminar flow, a large number of empirical correlations have been developed to determine the heat transfer coefficient. We present some of these correlations. ‘The Colburn equation [58] The Nusselt number for fully developed turbulent flow inside smooth tubes can be determined by recalling the Reynolds-Colburn analogy given by Eq, (6-1Sa) as St, Prt (7-64)FORCED CONVECTION FOR FLOW INSIDE DUCTS 318 and the definitions of c, and f, given by Eqs. (6-4) and (7-55), respectively, as (7-650) (7-656) From Eqs. (7-65a) and (7-65b) we conclude that //8 for flow inside a circular tube is analogous to ¢,/2 for flow along a flat plate. With this consideration, the analogue of Eq. (7-64) can be written for turbulent flow inside a circular tube as sipee =f 766) We substitute the friction factor from Eq, (7-63) into Eq, (7-66); ‘St Pr? = 0,023 Re“? (7-610) Nu where Sr (7-61) Equations (7.67) are rearranged to yield the Colburn equation for turbulent flow inside a smooth tube: (023 Re?* Pr? (7-68) Nu where Nu = AD/k, Re = tqD/v, and Pr = v/a. Equation (7-68) is applicable for: 07 < Pr < 160 Re > 10,000 L > @ smooth pipes and small to moderate temperature differences. Fluid properties are evaluated at the bulk mean temperature T, Dittus-Boelter [59] equation A slightly different form of Eq. (7-68) is given by Dittus and Boeiter: Nu = 0023 Re®* Pr* (7-69) where n = 014 for heating (Ty > T) and n = 0.3 for cooling (Ty < 7) ofthe fluid ‘The range of applicability isthe same as for the Colburn equation.316 mar TRansren ‘Sieder and Tate [53] equation For situations involving a large property varia. tion, the Sieder and Tate equation is recommended: Nu = 0.027 Re®* vo (tsp 0-70), This equation is applicable for 07. < Pr < 16700 Re > 10,000 L 5 smooth pi 570 pipes All properties are evaluated at the bulk mean temperature T;, except j, which is ‘evaluated at the wall temperature. Petukhov [61] equation The previous relations are relatively simple, but they sive maximum errors of +25 percent in the range of 0.67 < Pr < 100 and apply to turbulent flow in smooth ducts. A more accurate correlation, which is also applicable for rough ducts, has been developed by Petukhov and coworkers at the Moscow Institute for High Temperature: Re Pr ‘us wo EOE) my wee 7 « 2x00 - a) en (0.11 heating with uniform T,, (T,, > T;) aos = [iss sew eter 72 = 1 [0 uniform wall heat flux or gases Equations (7-71) are applicable for fully developed turbulent flow in the range 10 < Re <5 x 10® 05 1 when the liquid is ‘cooled. All physical properties, except y4 are evaluated at the bulk temperature. ‘The friction factor f in Eqs. (7-71) can be evaluated by Eq, (7-61) for smooth tubes or obtained from the Moody chart (Fig, 7-11) for both smooth and rough tubes.FORCED CONVECTION FOR FLOW INSIDE DUCTS 317 We have given four different equations for the determination of the Nusselt number for turbulent flow inside circular tubes. The Petukhov equation (7-71) is the most up-to-date correlation that is applicable for both smooth and rough tubes, and it appears to correlate with the experimental data very well over a wide range of parameters, Therefore, it should be preferred to the other correlations. Nosselt (62] equation The previous relations apply to the region L/D > 60 "Nusselt studied the experimental ata for L/D from 10 to 100 and concluded that h, in this region, is approximately proportional to (D/L)'"*. Henes, he replaced Eq, (7-10) by on ‘where L isthe length measured from the beginning of the heat transfer section, and the fuid properties are evaluated at the bulk mean fluid temperature Notter and Sleicher [63] equation The Nusselt number is determined theoretic- ally from the solution of the energy equation by using an appropriate velocity profile for turbulent flow. The resulting Nusselt number for the hydrodynamically and thermally developed region was expressed in the form Nu = 5 +0016 Ret Pr? 7-73) where 024 06m a= 088 P* and b= 033+ 050 which is applicable for o1 < Pr < tot 10 < Re < 10° L 573 Equation (7-73) correlates well with experimental data and provides a more accurate representation ofthe effect of the Prandt! number. It may be preferred to Eq, (7-72) Noncireular Ducts, So far, discussion of the friction factor and heat transfer coefficient for turbulent flow has been restricted to flow inside circular tubes. Numerous engineering applications involve turbulent forced convection inside ducts of noncircular ‘ross section. The friction factor for a circular tube given by the Moody chart318 wear meansreR (Fig, 7-11) applies to turbulent flow inside noneieular ducts ifthe tube diameter D is replaced by the hydraulic diameter of the noncircular duct, defined by Eq (749), as 44. P ‘here 4, is the cross-sectional area for flow and P is the wetted perimeter. For noneircular ducts the turbulent flow also occurs for Re > 2300, where the Reynolds number is based on the hydraulic diameter. ‘With noncircular ducts the heat transfer coefficient varies around the perimeter and approaches zero near the sharp corners. Therefore, for certain situations difculties may arise in applying the cireular-tube results to a noncircular duct by using the hydraulic diameter concep. Irvine [64] discusses some problems of heat transfer in noncircular ducts Dd 0-74) Effects of Surface Roughness ‘The heat transfer coefficient for turbulent flow in rough-walled tubes is higher than that for smooth-walled tubes because roughness disturbs the viscous sub- layer. The increased heat transfer due to roughness is achieved at the expense of increased friction to fluid flow. The correlation of heat transfer for turbulent flow in rough-walled tubes is very sparse in the literature. The Petukhov equation (7-71) can be recommended for predicting the heat transfer coefficient in hydro- dynamically and thermally developed turbulent flow in rough pipes, because the friction factor fcan be obtained from the Moody chart (Fig. 7-11) once the relative roughness of the pipe is known. Effects of Property Variation ‘When heat transfer takes place to or froma fluid flowing inside a duct, the tempera- ture varies over the flow cross section of the duct. For most liquids, although the Specific heat and thermal conductivity are rather insensitive to temperature, the viscosity decreases significantly with temperature. For gases, the viscosity and thermal conductivity increase approximately by 0.8 power of the temperature. Therefore, the property variation affects both the heat transfer coefficient and the friction factor. To compensate for the effects of nonisothermal con: Sieder and Tate equation (7-70) and the Petukhov equi viscosity correction term in the form (1/4 ‘The variation of the transport properties with temperature changes both the velocity and temperature profiles within the fluid, and an accurate determination of the effects of such changes on the friction factor and the heat transfer coefficient is a very complicated matter. Here we discuss some of the recommended simple correlations, used in engineering applications, in order to adjust the analytic ‘or experimental correlations for constant-property conditions for the effects of property variation.FORCED CONVECTION FoR FLOW SIDE DUCTS 319 For liquids, the variation of viscosity is responsible for the property effects. ‘Therefore, viscosity corrections of the following power-law form are found to be sulficiently good approximations: Nie (7-156) where 1, = viscosity evaluated at bulk mean temperature ie = viscosity evaluated at wall temperature ‘Nuj.,.Nu = Nusselt number under isothermal and nonisothermal conditions, respectively Jans f= friction factor under isothermal and nonisothermal conditions, respectively In the case of gases, the viscosity, thermal conductivity, and density depend on the absolute temperature. Therefore, temperature corrections of the following form are found to be adequate for most practical applications: w= (3 (760) £-(@) oe where J, and T, are the absolute bulk mean and wall temperatures, respectively. ‘A number of experimental investigations and variable-property analyses have ‘been reported in the literature to determine the values of the exponents n and k appearing in Eqs. (7-75) and (7-76). We present in Table 7-4 recommended values of these exponents. Thus, by using the corrections given by Eqs. (7-75) and (7-76), the Nusselt number and the friction factor for ideal isothermal conditions can be adjusted for the effects of property variations, ifno viscosity correction is included in the equation Table 7-4 The exponents n and m associated with Eqs. (7-75) and (7-76) stow Fuuid—condivon ar Reference Laminar Liguid Cooling or Beating 0.18 fe) Gas Cooling orheaing 0 1 8 Turbulent Liguid Cooling o2s (9 Kiguis Heating our (i Liquid Cooling or heating -028 Gar Cooling o oD) Ga Heating oot)$320 ear TRANSFER Example 7.9 Water flows with a mean velocity of tq = 2.mys inside a cireular Pipe of inside diameter D = Sem. The pipe is of commercial steel, and its wall is maintained at a uniform temperature T, = 100°C by condensing steam on its outer surface. At a location where the fluid is hydrodynamically and thermally developed, the bulk mean temperature of water is T, = 60°C. ‘Calculate the heat transfer coefficient h by using the Petukhov equation (7-71). Souvtion Various properties for water at T, = 60°C are taken as p= 985 kei fy = ATL > 10°* kei 8) k= 0651 Wiim-"C) Pr= 302 and the viscosity atthe tube wall temperature T, = 100°Cis a, = 282 x 10°* kgm) Then fnD _ (985)2H008) _ 9 44 , 198 Reg AT KO For the heating problem, we set n = 0.11, and the viscosity ratio becomes \OMt rarnyont (2) "= Ga) "= The relative roughness ofthe tube wall fr commercial steels 004s 5 = 0085 — 0009 ‘The friction factor fis determined from Fig. 7-11 as Jf = 00205 We now apply the Petukhov equation (7-71) Na = 20x 1°82 (0205) where X = 107 + 12.7802? ~ 0S) Then and Example 7-10 Solve the problem considered in Example 7-9 for a smooth pipe by using the following correlations: (a) The Notter and Sleicher equation (7-73).FORCED CONVECTION FOR FLOW INSIDE DUCTE 32 () The Petukhov equation (7-71). (©) The Sieder and Tate equation (7-70) (@) The Dittus and Boelter equation (7-69). SOLUTION The physical properties at 7; = 60°C are taken as k= 0651 W/(m-°C) Pr=302 Re = 204 x 108 = 471 x 10 kam s) py, = 282 « 10°* kell 5) The friction factor for smooth pipe at Re = 204 x 10° is obtained from Fig. 7-11 as f= 00152 (a) The Notter and Sleicher Eq. (7-73) gives 024 024 Pe 8 505 = b= 033 + 05e-°*" = 0.412 Nu = 5 + 0016(2.08 x 10°)9##4(3.02)9412 88 0651 ea 788 Foy = 10.267 Wien? °C) 0846 oss - 7 (6) The Petukhov equation (7-71) gives Ny = (204 X 10°N3.02) (00152)/3.71)0* Xx al trs) where x= 107+ w2xa00% — 9242)" Then Nu = 741.65 0. ac h = 141.65 9551 — 9656 when? °C) (6) The Sieder and Tate equation (7-70) gives 890-8¢3,02)1(471)* Nu = oorra04 « yr ncaony"(32) Then Nu= 704 0651 ae = 9166 whe? “C) = 708[322 HEAT TRANSEER (@) The Dittus and Boetter equation (7-69) gives Nu = 0.023(208 x 108)°(3.02)"* = 633 0651 ‘005 b= 633 242 Whim? -°C) 7-5 HEAT TRANSFER TO LIQUID METALS, ‘The liquid metals are characterized by their very low Prandtl number, varying from about 0.02 to 0.003. Therefore. the heat transfer correlations in previous sections do not apply to liquid metals, because their range of validity does not extend to such low values of the Prandi! number. Lithium, sodium, potassium, bismuth, and sodium-potassium are among the common low-melting metals which are suitable for heat transfer purposes as liquid metals. There has been interest in liquid-metal heat transfer in engineering applications, because large amounts of heat can be transferred at high temperatures with a relatively low temperature difference between the fluid and the tube wall surface. The high heat transfer rates result from the high thermal conductivity of liquid metals compared with that of ordinary liquids and gases. Therefore, they are particularly attractive as heat transfer media in nuclear reactors and many other high-temperature, high-heat-lux applications. The major difficulty in their use les in handling them. They are corrosive, and some may cause violent reactions whea they come into contact with water or ai ‘As discussed in Chap. 6 and schematically illustrated in Fig. 64, when Pr < 1, as in liquid metals, the thermal boundary layer is much thicker than the velocity boundary layer. This implies that the temperature profile, and hence the heai transfer for liquid metals, is not influenced by the velocity sublayer or viscosity. So in such cases one expects rather weak dependence of heat transfer on the Prandtl ‘number. Thus most empirical correlations of liquid-metal heat transfer have been made by plotting the Nusselt number against the Péclét number, Pe = Re- Pr. This situation, discussed earlier with reference to flow along a flat plate, also applies to flow inside a circular tube, as illustrated in Fig. 7-12. There the Nusselt ‘numbers from various sources for the heating of liquid metals in long tubes sub- jected to constant wall heat flux are compiled by Lubarsky and Kaufman [65] and plotted against the Péclét numbers, The data appear to correlate reasonably well, but there is also the scatter of the data. The reason has been attributed to difficulties inherent in liquid-metal experiments, namely, having to deal with high tempera. tures and very small temperature differences, The nonweiting of some liquid ‘metals on solid surfaces also has been considered as a possible explanation why some measured values of the Nusselt number are lower than the theoretical predictions. The reader should consult Refs. 66 and 67 for extensive data on liquid ‘metal heat transfer and the heat transfer characteristics of liquid metals.FORCED CONVECTION FOR FLOW INSIDE DUCTS 323, Meuse Nest number Na ; Bi Figure 7-12 Measured Nusselt number fr the heating of liguit metals in long, round tubes with constant wall heat ux. (Eom Ref 65) We summarize now some empirical and theoretical correlations for heat ‘transfer to liquid metals in fully developed turbulent flow inside a circular tube under uniform wall heat flux and uniform wall temperature boundary conditions. Uniform Wall Heat Flux Lubarsky and Kaufman [65] proposed the following empirical relation for calcu- lating the Nusselt number in fully developed turbulent flow of liquid metals in smooth pipes: Nu = 0805 re™* | on where éclét number = Pe = Re Pr for 10? < Pe < 10%, L/D > 60, and properties evaluated at the bulk mean fluid temperature. ‘Skupinski, Tortel, and Vautrey [68], based on heat transfer experiments with sodium-potassium mixtures, recommended the following expression for liquid ‘metals in fully developed turbulent flow in smooth pipes: Nu = 482 + 00185 Pe"? 8) for 346 x 10° < Re < 905 x 10%, 10* < Pe < 10*, and L/D > 60. The physical properties are evaluated at the bulk mean fluid temperature. ‘Equation (7-77) predicts the Nusselt number lower than Eq (7-78); therefore it is on the conservative side.34 HEAT TRANSFER Uniform Wall Temperature Seban and Shimazaki [69] utilized the analogy between momentum and heat transfer and proposed the following expression for liquid metals in smooth pipes under uniform wall temperature: Nu = 50 + 0025 Pe™* 7-79) for Pe > 100, L/D > 60, and physical properties evaluated at the bulk mean fluid temperature. ‘The expressions also have been developed for the Nusselt number in fully developed turbulent flow of liquid metals in smooth pipes subject to uniform wall temperature boundary conditions by empirical fits to the results of the theo- retical solutions. We present now the results of such solutions: Sleicher and Tribus [70]: Nu = 48 +0015 Pe! Pr?3° for Pr < 0.05 (7-80) ‘Azer and Chao [71] 0-4 005 Pe?" PPA for Pr <01,Pe< 15000 | (781) Nu Notter and Sleicher [63]: Ni 48 + 0.0156 Pe? Pr°®* for 0.004 < Pr < 0.1, Re < 500,000 (7-82) ‘The physical properties in calculating Nu, Pe, and Pr in these expressions are evaluated at the bulk mean fluid temperature; the expressions are applicable for L/D > 60. Figure 7-13 shows a comparison of the Nusselt number under uniform tube ‘wall temperature determined from Eqs. (7-79) to (7-82) with the experimental data of Sleicher, Awad, and Notter[72] for heat transfer to NaK in turbulent flow inside circular tube with uniform wall temperature. It appears that Eq, (7-82) agrees well with the experimental data ‘Thermal Entry Region ‘The previous relations for liquid metals in turbulent flow are applicable in the fully developed region. Sleicher, Awad, and Notter [72] examined the heat transfer calculations of Notter and Sleicher [63] in the thermal entry region for both uni form wall heat flux and uniform wall temperature. They noted that the localFORCED CONVECTION FOR FLOW IRSDE DUCTS 325, 1 SetunShiaat Ee (7-99) oaro: Ea (61) Rat? Figue7-13 Comparison of Nusselt number rom various eoeeations fad the exprimestal resls for heat transfert NaK in turbulent flow inside a circular tube with ko ‘00 To'ooe uniform wall temperature. (From a Slecher eral. (721) ‘Nusselt number for the thermal entrance region can be correlated within 20 percent with (7-83) where Nu = 63 + 00167 Pe®* Pr°°% for uniform wall heat flux (7-84) Nu = 48 + 00156 Pe Pr°-°" for uniform wall temperature (7-85) and applies in the range 0.004 < Pr < 01. Effects of Axial Heat Conduction in Liquid Metals Liquid metals have very high thermal conductivity. Therefore, in the thermal entrance region where the temperature gradient in the axial direction is high, the heat conduction in the fluid in the axial direction may become important. In general, the effect of axial conduction in the Fluid is negligible for Pe > 50; this condition implies that the axial heat conduction in liquid metals may become important for laminar flow. Heat transfer in laminar forced convection in circular tubes with axial heat conduction in the fluid has been investigated theoretically [73-80]. Figure 7-14 shows the effects of axial heat conduction in the fluid on the local Nusselt number Nu, at the thermal entrance region for laminar flow inside a circular tube subject to uniform wall heat flux in the downstream region (i, x > 0) while the upstream region (ie. x < 0) is insulated. The axial heat conduction becomes important for Pe < 50, and its effect is to reduce the local Nusselt number in the thermal entrance region; but the length of the thermal entrance region is very short (i, about 1 diameter),326 wear TRAnsren A 1 L 0 10 10 L Gar! 26/0)KRe-PA Figere 7-14 Eecs of axial heat conduction in Mut on the lea! Nuset number for lamina ow ina ‘tela abe with uniform wal beat ux. (From Vick and Osi (90)) Example 7-11 Liquid NaK (56 percent Na) flows with a mean velocity of tq = 3 mys inside a smooth tube of inside diameter D = 2.5 em and is heated bythe tube wall maintained ata uniform temperature T, = 120°C. Determine the heat transfer coefcint at a location where the bulk mean fuid tempera ture is T, = 95°C and the low i fully developed by using Eas. (7-79), (7-80), and (7-82), Compare the results ‘SoLUTION The physical properties of NaK (56 percent Na) are taken as p= 887 kg/m? w= 058 x 10°? ke/(m-s) k=256W/m-°C) Pr = 0026 Thea = PlmD _ (887V3X0.025) _ Ree 058 x 19> ~ 115000 and Pe = Re Pr = (115,000(0026) = 2990 ‘Hence Eqs. (7-79) (7-80), and (7-82) are applicable.FORCED CONVECTION FOR FLOW INSIDE DUCTS 327 Using Eq, (7-79), we find Nu = 50 +0025 Pe®* = 50 + 00252990)" = 20.1 k 256 h=Nus=201 i gay 7 20582 Wim? 6) Using Eq. (7-80) yields Nu = 48 + 0015 Pe" prose 48 + 0.015(2990)"°1(0026)" °° = 12.1 k 256 Aes hem Nag = 121 Gas = 12877 Wilm? -“C) Using Eq, (7-82), we find 48 + 010156 Pe®®$ prot 4.8 + 0.0156(2990)°-*4(0.026)°" = 15.3 wk 256 ha Nuj = 153 Go Nu 15,667 Wim? -°C) Equation (7-82) gives the heat transfer coefficient somewhere between those ‘obtained from Eqs. (7-79) and (7-80). A comparison of these three equations with the experimental results for NaK, given in Fig. 7-13, hows thatthe result {rom Eq, (7-82) is closer to the experimental data. So, of the three results calculated, the one from Eq, (7-82) is preferred. Example 7-12 By using the heat transfer coefficient for the fully developed flow determined as case 3 in Example 7-11, calculate the heat transfer co- cffcient for the thermal entry region at locations 5 and 10 diameters from the inlet. : 1 u-(g where h for the fully developed region is determined in case 3, of Example TAl,as = 15,667 W/(m? -°C)328 ext TRansreR fh, = (1 + 315,667) = 21,934 Wim? -°C) hg = (1 + #9X15.667) = 18,800 Wim? - °C) 7-6 ANALOGIES BETWEEN HEAT AND MOMENTUM. TRANSFER IN TURBULENT FLOW In Chap. 6, based on the analysis of laminar low along a flat plate. we developed relation between the local heat transfer coefficient h, and the local drag coefficient ns iven by Eq, (6-15a) as where St= (7-866) Pr For turbulent flow inside a circular tube a similar expression is given by Eq (7-66): St Prt (7-810) gy — Nuc Re ; where = (7-81 Equations (7-86) and (7-87) are the relations between heat and momentum transfer for turbulent flow along a lat plate and a circular tube, respectively. Given the ddrag coefficient or the friction factor, the heat transfer coefficient can be found from these equations. Is instructive to examine how equivalent expressions car bbe developed for turbulent flow along a flat plate or inside a circular tube, by ‘making use of the analogies between heat and momentum transfer. Here we consider only the flow inside a circular tube. “The first and simplest such relation for turbulent flow was developed by Reynolds and is now known as the Reynolds analogy for momentum and heat transfer. More refined analogies were developed later by Prandit [81 von Karman [82], Martineli [83], Deissler (60, 84], and many others. The development of all these analogies is based on the solution, with various degrees of approximation, of the relations for the shear stress and the heat flow in turbulent flow given byFORCED CONVECTION FOR FLOW INSIDE DUCTS 329 Eqs. (6-32). If we assume fully developed turbulent flow, hence u = u(y) and T = T(), Eas. (6-32) become du area 1.88) fa 04m 7-88) 4 ~(a4ey tt 789) Py 4 where y isthe distance measured from the tube wall. To provide some insight into the nature of the approximations involved in the development of such analogies, we now derive the Reynolds analogy. Reynolds Analogy Reynolds assumed that the entire flow field consisted of a single zone of highly turbulent region. That is, he neglected the presence of the viscous sublayer and the buffer layer. In such a turbulent core, the molecular diffusivities of heat a and of momentum v are negligible in comparison with turbulent diffusivities. That is, Vetq and de (7-900) In addition, he assumed that the turbulent diffusivities are equal: meee (7-906) With the assumptions in Eqs. (7-90), the expressions given by Eqs. (7-88) and (7-89) simplify, respectively, to os ; at (7-92) a Ay combining ies two equations we fad tra - Lu 09) The inteaton of Fe, (79) fom te wall onions T = T= 0 0 the ‘mean bulk stream conditions T = Ty, remains constant results in =, with the assumption that q/t (7-94)380 ear Mansi ‘Now, the heat transfer coefficient hand the friction factor f for flow inside a tube are defined as a= WT. -T) (7950) ase [see Eq. (7-55)) (7-956) ‘The substitution of Eqs. (7-95) into (7-98) yields | x= .f 096) Reamer) ‘This result is known as the Reynolds analogy for momentum and heat transfer in fully developed turbulent flow in a pipe. Its valid for Pr = 1 Prandtl Analogy Prandt assumed that the flow field consisted of neo layers.a viscous sublayer where the molecular diffusivities are dominant, that is, mtv and yaa (797) and a turbulent zone where the turbulent diffusivities are dominant, that is vam 24y and =m 7-98) ‘These assumptions are utilized to simplify Eqs. (7-88) and (7-89) for each layer, the equations are integrated, and the definitions of the friction factor and the heat transfer coefficient are introduced. The following result is obtained: (7-99) 8145/78 ‘This relationship is known as the Prandtl analogy for momentum and heat transfer for fully developed turbulent flow in a pipe. We note that for Pr = I the Prandtl analogy reduces to the Reynolds analogy. ‘Von Karman Analogy ‘Von Karman extended Prandt!s analogy by separating the flow field into three distinct layers: a viscous sublayer, a buffer layer, and a turbulent core. He made assumptions about the relative magnitudes of the molecular and turbulent diffusivities of heat and momentum in the viscous sublayer and the turbulent core similar to those made by Prandtl, but in addition he included the effects of theFORCED CONVECTION FoR FLOW nese DUCTS 331 buffer layer by assuming that the molecular and eddy diffusivities in this layer were of the same order of magnitude. The following result is obtained (7-100) which is known as the von Karman analogy for momentum and heat transfer for fully developed turbulent flow in a pipe. We note that for Pr = I this result also reduces to the Reynolds analogy. This relation appears to be good for a Prandtl ‘number up to about 30. ‘The previous relations are given in terms of the Stanton number. They also can be expressed in terms of the Nusselt number by noting that Nu Re- Pr where Nu = AD/k and Re = u,D/v. The effect of the property variation can be included by utilizing Eq. (7-75), that is, st (7-101) 7-102) With these considerations, we write the von Karman analogy in the alternative formas _ Re: Pr: 18 ( 1)" * na = EPA) riot ‘5 Pr + 1\)(f\"? sfornem® 2)" osm and the exponent n may be determined from Table 7-4. ‘Note that the von Karman analogy, expressed in the alternative form Eq (7-103), resembles the Petukhov equation (7-71) except for the definition of the Z term in the denominator. However, Eq. (7-71) should be preferred to predict the heat transfer coefficient, because the denominator has been adjusted to closely correlate with the experimental data over a wide range of the parameters. For ‘example, for Pr = 1 in the von Karmén analogy Z = 1 and the corresponding. term in the Petukhov equation becomes X = 107, which is about a 7 percent difference. where Zz Example 7-13 Solve the problem considered in Example 7-9 by using the alternative form of the von Kirmén analogy, Eqs. (7-103). Compare the result- ing heat transfer coefficient h with that obtained by the Petukhov equation om),382 Hear TmANSTER ‘SoLUTION The physical properties for water from Example 7-9 are: k = 0651 W/m? -°C) Re = 204 x 10° Pr= 302 py = ATL x 10-4 Keil) fag = 2.82 x 10+ eps and the friction factor is f= 0.0205 For the heating problem we set n = 0.11. We now apply Eq. (7-103): ==) _ 204 x 1083.02) eeamsyceznyes __ 1763.42 Zz 3s )\z83, Z where Zatt slr -D+ n “ WG)" at+ s[on -D+hn ( = set) ea)" 761 6 8 ‘Then Nu = 950.27 0651 ne 95027 Tas = 12372 Whim? °C) This result is very close to 12,307 given by the Petukhov equation (7-71) in Example 7.9. However, the deviation may increase at larger Prandtl numbers. and 7-1 HEAT TRANSFER AUGMENTATION In recent years, energy and material saving considerations have prompted an expansion ofthe efforts aimed at producing more efficient heat exchange equip- ment through the augmentation of heat transfer. The potentials of heat trant- fer augmentation in engineering applications are numerous, For example, the heat ‘exchanger for a projected ocean thermal energy conversion (OTEC) plant [85] requires a heat transfer surface area on the order of 10,000 m?/MW(c). Clearly, ‘an increase in the efficiency of the heat exchanger through augmentation may result in considerable savings in the material need Desalinization is another application in which large heat transfer surfaces are required and possibilities exist for the use of augmented systems [86, 87]. Numerous ‘other examples can be cited for the application of heat transfer augmentation, ‘A vast amount of literature exists on this subject, and a comprehensive Survey ‘on heat transfer augmentation is available [88-90]. Here we briefly discussFORCED CONVECTION FOR FLOW INSIDE DUCTS. 333 principal augmentation techniques for single-phase forced flow in ducts, and we Cite the pertinent references that present data on heat transfer and friction factor characteristics of various augmentation systems. ‘Augmentation Techniques We recall from our discussion of the analogies between momentum and heat transfer that increasing the friction factor increases the heat transfer coefiicient. ‘The Moody chart (Fig, 7-11), for example, shows that in turbulent flow increasing the relative roughness of the surface increases the friction factor. This chart is ‘based on the random sand-grain type of surface roughness. Other types of surface roughness have been produced, and their friction factorsand heat transfer character~ istics have been tested for possible use in heat transfer augmentation. We discuss some of these enhancement techniques. Roughened surfaces Surface roughnesses can be produced by the machining of the surface as well as by casting, forming, and welding processes. Clearly, an infinite ‘number of geometric configurations are possible, each having its own heat trans- fer and pressure-drop characteristics. Experimental techniques generally are used to determine the heat transfer coefficient and friction factor for flow, since no unified analysis is yet available for predicting them by purely theoretical means. Bergles and Jensen [91] and Webb, Eckert, and Goldstein (92, 93] presented heat transfer and friction factor data for various surface roughnesses. Extended surfaces The use of fins on the outer surface of tubes to enhance heat transfer is well known, Internally finned tubes have been used also to enhance heat transfer to fluids flowing inside tubes. Heat transfer and friction factor correlations have been presented for internally finned tubes under laminar flow [94] and turbu- Tent flow [95-97] conditions. A theoretical approach has been applied to predict the effects of internal fins on the turbulent flow of air [98}. Enhancement devices Enhancement devices such as twisted tapes have been employed in the form of inserts into the tubes, to promote increased heat transfer for the laminar and turbulent flow of viscous fuids [99, 100]. Coiled tubes Coiled tubes can serve as a heat transfer enhancement device because the secondary flow produced by the curvature causes an increase in the heat transfer coefficient. Effects of curvature in enhancing heat transfer in curved pipes have been reported [101-103] Optimization Methods ‘The increase in heat transfer with augmentation is accompanied by an increase in the friction factor. For example, in some situations the heat transfer coefficients are increased at most about 4 times while the friction factors are increased as much,334 ear manson as 50 times or more [91]. An increased friction factor implies an increased power for pumping the fluid. So the results of augmentation, when itis applied to produce ‘more efficient heat exchange equipment, should be weighed against the increased power requirement for pumping the fui. ‘Studies have been conducted to develop methodologies for assessing heat transfer, pumping power, and surface area for a system resulting from enhance- ‘ment versus those or the unenhanced system [104-108], For a given enhancement technique, if the heat transfer and the friction factor data are available as.a function of the Reynolds number, it may be possible to optimize the system to reduce the heat transfer surface, to obtain increased heat transfer capacity, or to reduce the power required for pumping the Bud. No unified correlation of heat transfer and friction factor characteristics of enhancement techniques is yet available. Each specific enhancement method has its own correlation. For this reason, we do not present here various correlations for specific cases; instead, pertinent references have been cited. However, we give now an example, to illustrate a typical data set on heat transfer augmentation and its use in the optimization processes as a means to reduce the heat transfer surface area for a heat exchanger. In an actual optimization study, numerous other questions arise in connection with the area reduction resulting from heat transfer enhancement. For example, the manufacturing procedure, manufacturing cost, ‘material, and many other factors associated with the augmentation technique used should be considered before a final decision is made, [Example 7-14 Heat transfer augmentation is applied on the inner surface of circular tube subjected t0 uniform wall temperature by condensing vapor ‘on the outer surface. Figure 7-15 shows heat transfer and friction factor data for different types of enhancement on the inner surface of the tube for experi- I we sonine 2 Fa SeTRION TNA Se 1 2 Pa sargigr 2 F 4 S708 Figure 718 Heat ansler and tetion factor date for wales owing in internally coughened tubes (From Ref 91)FORCED CONVECTION FoR FLOW INSIDE DUCTS 335 ‘ments performed with water low. Examine the effectiveness of each enhance- ‘ment method in reducing the heat transfer surface area while the total heat, transfer rate, the pumping power requirement, and the tube length remain the same as for the unaugmented, smooth tubes. SOLUTION This example is based on the optimization study [91] performed n connection with the inside augmentation of a proposed shelland-tube ‘heat exchanger for an ocean thermal energy conversion power plant, described in Ref- 17. The total heat transfer rate Q is considered to remain the same for both the augmented and the reference (unaugmented) systems. Asa first approximation, we write Q = Agho AT = Ah, AT where the subscripts 0 and a refer, respectively, to the reference and the augmented cases. Here we assume thatthe temperature diffrence AT doesnot change forthe augmented and the reference systems. And we use hinstead ofthe ‘overall heat transfer coefficient u because our analysis s an approximate one. The area ratios obtained from the above heat transfer relation and rearranged Ay __ Nu/Pe°* 0.023 Reg" Ay hy NayiPr?* ~ NugPr®* ‘here the correlation Nuy = 01023 Re?* Pr i used forthe reference cae. ‘The pumping power fequirement W for both systems isthe same and can be expressed as @ W = cfoAaus = Agus since the constant cis defined as W = (flow rateypressure drop) “(vin|5) a0 where 4 is the surface area, wis the low velocity, ns the number of tubes in the bundle, and f isthe friction factor. ‘The area ratio is now obtained from the above pumping power relation- shipas fous _ fae} _ 0.184 Re5* Ay” fa ~ foRen” ~ f.Re? where the correlation fo = 0.184 Re5°? is used forthe smooth tube From Eqs. (a) and () we obtain 0.023 Reg" _ 0.184 Re NuPe* ~~ 7.Red O)2336 ear rRanseen Be oan | [5 Loe rane ef ost f.Rez \'? re, <0ns(glt o Now the calculation procedure is as follows 1. Choose Rey. 2. Determine from Fig. 7-15 the corresponding values of Nu/Pr°* and J; then calculate from Eq, (c) the corresponding value of Rey 3. Given Reg, calculate the area ratio Ay/Ay from Eq. (a) Figure 7-16 shows the area ratio 4,/4y plotted against Rep for each of the various enhancement techniques. The results suggest that the enhancemert with the low transverse ribs is most efficient, and such tubes ae also acceptable with manufacturing considerations. For more accurate results, the calculations should be performed with variable temperature differences and by using the overall heat transfer ec- efficient instead of the h values. The reduction in the heat transfer surface area obtained with such calculations will be less than that shown in Fig. 7-16 However, the area reduction—even with the inclusion of outs for condensation and the tube resistance~may amount to about 30 percen: This is a considerable savings when large surface areas are involved, 7-8 SUMMARY OF CORRELATIONS ‘We summarize in Table 7-5 the correlations of the heat transfer coefficient and the friction factor for forced convection inside ducts presented in this chapter,FORCED CONVECTION FoR FLOW INSIDE DUCTS 337 ‘Table 7-5 Summary of correlations for forced convection inside ducts* Equation ortable 026) any on ‘Table 7-1 ‘Table 72 0-9) sn crs) 7-2 0-8) rs) 09) 70) om om Flow Conreaton ‘eae 8 anima Re nes att Laminar Nua Laminar Noa Lamina sand Laninar .0sGe Lamia Now saaan(te) J=CaRIog Re 16 Turolent j= onteRe Turtle sorter? Torben a = 00m Ret? “oro Nu = 0023 Re Pe urbe a Ds for eat 10 for cooking ayers Tabula w= osm net» (") nw = REP Torbtent oe (hee were en xe sor+ renee — 1 (4) = 011 beating wih osform 7, 2 cotng wih usiform 7, 1 nlm val bent a oF es Torben nw = 0036 net nt (2)" w= 006 nerve? Remarks Filly developed fow na crular tube Folly developed fow ina eevlar tube, constant wal heat Hx Folly developed fow ina ctelar be, constant wall temperture Fally developed flow in duets of Hydrodynamic bd thermal ity lng for dacs af Enirance region, cla tube, ‘constant wal emperatre Eniranee region. circular ube ‘consant wal temperate, O38 Pr-< 16700an0 (Gn (2) 22 Smooth cubes, Re > 10 Smooth tubes, Re < 2% 10" Sot tubes, Dae We Re <3 x 10" 0.7 < Pr < 160; Re > 10.00: 1. > 60: smooth pipes 07 < Pr < 1s Re > 10.000; iD > 0: smooth pipes 07 < Pr 16700: Re > 10000; LD > 60; smooth pipes Smacth oF rough pies lot < Re < $x 10 05 © Pr-< 200 with $06", 05 Pr < 2000 with 10", Properties, excep. aFe vale bulk moan fempersture L w< ke38 HEAT TeanseeR ‘Table 7-5 (Continued) where « = 0.88 — 24 oe oie oy poy(i gat ley rent) “ ” turbulent ota een Oo 2 wane). san(8)) ae om rat i) Nee an soonest om (og rt =) where Nas dtermine from a, (7-44 fr uniform wall eat un Eq, (7-85 for uniform wall Remarks OL Pr< io 10" < Re < 10° Loos boas ‘See Table 7-4 for recommended ‘See Table 7-4 for eecommendst ales of m Ligutd metae uniform wall hea fax: D> 60108 Pe 10" Liquid metals: uniform wall heat fx; L/D > 60; 3 10 < Re < 90 x 10 lo? = Pe = 10 Lui metals form wal Temperature: L/D > 0; Pe> 100 Liguid metals: uniform wall Temperature; L/D > 6; Preoos Liquid metas: uniform wall Temperature: L/D > 5; Preol:Pe > 500 Lig metals: uniform wal Temperature: L/D > 60: Re = 500.00;0.006< Pr <0.1 Lig metas entrance region; Oo Pr <1 Unless otherwise sated, uid properties are evaluated atthe bulk mean Ad finperatue 0 “+ Tug2. The fm temperature T, 8 defined a8 T; = (7. + T,)2 Inthe visoaty rahe ‘#8 the viscosity at bulk mean fd temperature andj. at te wall temperature. Also Re Din Gz = (Re: PeVIL/D) and Pe = Re Pr.FORCED CONVECTION FOR FLOW INSIDE DUCTS 339 PROBLEMS “1 A lbricatng ol of viscosity and thermal conductivity fils the clearance L. between wo rotating Cinders, which ean be regarded a two parallel plates in mation fr the purpose ofthe analysis. Let tnd, be the velocities ofthe nner and outer cylinders, especie). Develop a relation for the ‘ost distibation in the ol ye for (a henner and ote ylnders rotting in the same dtston btn, > uo and) the inne and outer eylnders rotating in opposite directions. Also devloparcation forthe shes sts in the fd resulin rom the rotation. 72 Alubsicatng oil of viscosity and thermal conductivity kis contained inthe clearance L between the journal and the bearing which un be regarded as two parallel pats Let, be the velocity ofthe ‘pper plate while the lowe plate s considered stationary. Heat is generated inthe ol layer during {oration at real of vsous-nergydbsipaton. Develop expressions forthe temperature distribution in hed forthe fllowing cases “@) The loner pat maintained at «temperature Ty and the upper plate ata temperate T;. with T, > Ty, (@) The lower plate is kept atx trperatre Ty while the upper pate insulated. Also develop an expression for the temperature ofthe insulated surface 13 Develop an expression for the maximem temperature ris in a lubricating ol contained batween & [Durnl ands beating ifthe veloc ofthevotalingetrface eu, mis andthe journal an the Bearing are ‘oth mintained at the same temperatre Ta. Calculate the maximum temperature rie fra rotation ‘velocity ofa = 10 mjsand T, = 20°C The viscosity ofthe ois given asp = 0.21 gm 3) a °C, 1045 a HPC, and y= O8 at FC. The viscosity canbe taken a the mean fd temperatre. The {thermal conductivity of otis k= 014 Wim -"C) 44 Consider Covet flow with heat transfer between two parallel plates separated by a distance The spacing between te plates ie filed witha Mai of viscost wand thermal conductiviyK The lower plete stationary. andthe upper plat moving with a elo w, and mantis at a temperature Ti, with T, > Ty where Te the temperature of the lower pte Establish the criteria under which thre wil be no beat rar inthe upp plate, Also develop an expr fr eat transfer rate at the lower plate. 745 Two large parallel plates separated bya distance L = 0.2m are maintained at T = 20°C. The upper plate is moving with a speed, = 100 m5, while the lower plate i stationary. In one ease the fluid between the pats ts waernthe othe case tai For each Aus determine the flowing) the ‘maximum temperature ie nthe ud (the shea sss inthe id and) the heat ix atthe plate surface. “Answer: (a) O98 and 21°C; (6) 0987 and $0.33 Nn? (€) 41.85 and 2516 Win 114 Consider Couette ow between two pall plates with the lower pata y = Ostaionary and the "pperpiateaty = Lmoving witha veloc = ,, Develop solution forthe velocity distribution) forthe case jax = C0. ‘Hydrodynamicly and thermally developed fw sie ica tubes 17 Determine the fiction factor andthe pressure drop for fully developed laminar ow of ethylene iyo! at APC [y= 096 x 10-7 kplon= 2) p = 1101 kg’) through a S-om-diameter,SO-m-long {ube ata rate of kas “Anomers 0242; 786 Ni? 748 Engine il t 40°C [4 = 0.21 kgm 2h. p ~ #75 kg/m) Bows inside a2 S- ‘ordinate measured ally in the direction of ow and rhe radial coordinate By writing force balance ‘equation forthe pressure and the vicous shear frees acting om an annular element frais, radal thickness A. and ail thickest As inthe fd, show thatthe > momentum uation i 1a] ae ral! ar ne where isthe scosty, dP isthe pressure gradient, and (ithe axial How velocity 743 By solving the: momentum equation 4 [ae] tat ral ae | ae for constant and Ps, deeop an expression forthe velocity ditribution uf) for fly developed Ieminar Howie acca tabe frie R. The ich vlocy dab (@) Develop a relation forthe mean Dow velo, (0) Develop an erreasion fo the ton ator / defied as aids f° GpuaiD hr isthe deo of id ad D isthe inside diameter ofthe tube. (© Determine hea wwe ithe ceterine loc andthe mean flow elo 7-44 By sling the moma equation for constant and Ps deep an expreion for he ‘elo dition in fly developed lamina ow nes ccular aan of nse ius, and uid radi ‘Hyérodyuumically and thermally develope lm inside noncicaar dts ‘18 Determine the friction fctr for flow of ethylene pycol at 0°C(» = 192 % 10* ma throughs parll plate channel havinga spacing of ~ 5 cm betweon the plate witha velocity ou, = 025 ms. ‘Assume the flow is laminar ad aly developed ‘Answer 00737 76 Consider the fully develope laminar fw of viscous fd through a parale- pate channel of spacing Landa circular tube of diameter D. Determine the ratio L/D for which the fection factors for flow between a paale plate chanel apd circular tube te equal 7-17 Compare thefietion factors ascent wit the fully developed laminar ow of engine oat 40°C (= 084 « 10-* mip through a parle pate channel of spacing LIS em and acral tube of diameter D = 2m with velocity of 05 mis 17-18 Consider hydrodynamical flly developed laminar fow between wo paral plates. Letx be the coordinate measured aially in the direction of low andy be the coordinates normal tothe wal. Fy wing a force Balance equation fo the pressure and the viscous seat frees acting om 4 fren ‘olumeclementof ses Ax nd Ay and length unity nthe fi, show tha the x momentum eqestions: Puy) _ Lah a as where isthe wsosty, dP x isthe pressure gradient in the Now, and) ithe axial ow velocity.FORCED CONVECTION FOR FLOW INSIDE DUCTS. ML 1719 Consier uly developed lamina flow between wo parallel pats separated bya distance 2 Let {he Osan be taken along the ostter plane in the diet of flow and te Oy ats be normal othe ‘walls The x momentum equation sen by Puy) _ 1470) OF ede (@) By solving tis momentum equation fr constant and dPts/4s, develop 2 relation fr the fully developed velocity profile 4)- (@) Develop eelation forthe mean velit uy. (6) the eton factor is defined by here pis the density and Dy ~ the hydraulic diameter, develop a elation forthe ection factor f (@) Determine the ratio li, where the Now velocity at the midplane ofthe channel and, i the mean flow velocity 7.20 Consider he fw of water ata rate of 1015 kysthrough a square duct 2m by 2cm whose walls, ave maintained ata uniform empeature 10°C. Assume that the ow is hydrodynamical and therm ally developed. Determine the dct length equite to hea the water rom 3010 17°C Anower: L327 721 Consider the fow of wateratarateof 001 kx/ through an equilateral wianglar duct ofsides2 cm and whore walls are Kept a aniform temperature 100°C Asume thatthe Now i byrodynamically ‘nd thermally developed. Determine the dat length requied to heat the water from 2010 17°C. Anower! Sat 7.22 Engin oil at 50°C wth fow rate of 10"? ky eaters a L-mong equilateral triangular duct of sides 05 om whose walls ae maintained at «uniform temperature 120°C. Assuming hydrodynamiclly ‘nd thermally fly developed flow, determine the oul! temperature ofthc 723 Oil at 50°C enters anon conduit whose cross section can be approximated as two parallel Plats with a spacing of cm, The Bow rate of 5 kgm!» The walls are subjected to uniform beating at rte of 1000 W/m, Assuming hydrodynanscally ad thermally ally developed flow, de- termine the average heat transfer coeficen the temperature is ofthe ol for flow through the co ‘hit and the wall emperature, [Properties of oil maybe taken tthe anticipated mean temperature of PC asy = OO32Kpl 8) cy = 2150W -ake°C), =O 14 Wie -"C)p = SSOKglm" and Pr a7 7-24 Consider hydrodynamically developed laminar ow of veosty inthe x direction between two ‘tale plats Let be the coordiate axis normal ote plates, Consider difeeatial volume element [Ae Ay 1 By wring an energy balane equation for heat conducted the direction and heat convect in the x direction fr thi volume element, show tha the eneray equations 1 yy ATI) _ BTC) ye? whore the thermal fast ofthe lid, T(x, 3) isthe temperate, and 4) isthe lamina velocity roe 7-S Water at 20°C with a flow cate of 0015 kt enters a 2S-m-ID tube whichis maintained ata ‘uniorm temperature of °C (a) Determine the dermal entry length (©) Assuring hydrodynamialy and thermally fully developed flow, determine the heat transfer ‘veicient and he tube length eure to heat the waterto 70°C. (@) Determine the heat tranaer ate tothe wat (@) Determine the fection factor and the pressure drop. “Answer (2) 44m (8) 927 W/? °C) and 10.81; (3.14 KW) LOAN 7-26 Deiermine the friction factor f forthe uly developed laminar fw of engine ol at 6O"C witha flow rate of 01 kyl through a)» elas tube of Tom diameter, (0) 4 square duet 1m by Tem,342 wear Teanseen (6)aectanguar duct 0m by 2em.and (dan equilateral triangle dt of sides Lem and having shaip 727 Determine the preure drops over the length L, Prob, 726 10m for cach of the cases considered Hydrodynamic and the cary lenge 7-28 Determine te hydrodmamicemry length or Now at IPC and at rat of 01S kel of wate, tylene glycol and engine ol through cicslar tube ofisie diameter 25cm “Answer: 227,0207, 0015 0 129 Determine the shermal entry lengths for laminar flow at GO-C ata rate of O01 ks of wate, sthylee glycol, and engine ol through accu tbe of inside diameter? em and subyected to un form wal temperature. 7.30 Determine the thermal entry lengths for hycodynamicaly developed laminar ow at 40°C with aw rate of 50 kgm) of water, lyern ethylene gy, nd engine oil trough a parallel plate channel having spacing of | cm and subject to uniform wal emperatare “F31_Repeat Prob. 7-30 for uform wall Bet hi 17-32 Repeat Prob. 730 and 7.3 fora ow ete of 20 kgm?) 7.38 Compare te thermal entry lengths for hydrodynamically developed laminar flow of the sam: ‘uid atarate of kg/sthrough the following dct subjected to tniform wal temperature (a) ciculr ect of diameter D(a square duct of side D, and (c)a rectangular duct sides 4D and 2D. 7-34 Determine the hydrodynamic entry length, thermal entry length, and heat transfer cote for fully developed fw fr engine olat 60°C Rowing at O01 kgs through a square duct | em by |e ‘in cos section and subjected toa uniform wall emperature,(Physial properties of engine ol at 60°C ‘may betaen as p= 864 kal’, c, = 2047 W kg “Chk = 0180Whom. °C) p= COPS Eg Pr 1050) 7.8 Repeat Prob. 7-H fora flow through a rectangular duct 0 cin by 1m in cross section and subjected to uniform wal tmperaure, 7-36 Glycerin a 20°C enters a square duct | em by | em in cro section with a Row rte of 001 ks The walls ofthe duct ae subjected to uniform heat Bux everywhere (hat, Nupy condition ot Tobie 7 (a) Determine the hydrodynamic entry length, (6) Determine the thermal enry length. (6) Determine the eat tanserenefcent forthe veglon where the velocity and temperate rofies are fly developed Anam: (@) 0.0006 (6) 56 m; (6) 103 WAm?- °C), 7-37 Repeat Prob 736 fora rectangular duet having cos section 0 em by 2em. 7-38 Engin oilat 27°C enters a0 3 8). Figure 8-2362 HEAT TRANSFER Figure #2 Velocity and themat Doundaey yer for lguidmetal eat waasfer Pr 1 illustrates the velocity and the thermal boundary layers when both start to de- velop from the leading edge of the flat plate. Let T,, and ube the temperature and the velocity of the fuid, respectively, outside the boundary layers; Tis the surface temperature of the plate. We assume incompressible, constant-property fluid in ‘two-dimensional, steady flow with negligible viscous dissipation. The energy ‘equation governing the temperature distribution T(x, y) in the thermal boundary layer is obtained from Eq, (6-83) by neglecting the viscous-dissipation term: or, at wr oe (628) For convenience inthe analysis, we define a dimensionless temperature M(x, ») as T(x, 9) = Te a= 29) He, y= TE 629) where O(x, y) varies from zero atthe wall surface to unity at the edge of the thermal boundary iayer. Then the energy equation is written in terms of Ox, y) as 0 ae wR rep aage forx>o 30) and the boundary conditions are taken as O=0 aty=0 (8310) O=1 ty = a(x) &31b) ‘where Eqs. (8-31a) and (8-31b) are, respectively, the statements of temperatures equal to T, at the wall surface and T,, at the edge of the thermal boundary-layer thickness 5(x). ‘The exact analysis of this temperature problem is quite involved, because the velocity components u and v should be determined from the velocity problem given by Eqs (8-1) to (8-3), before the energy equation (8-30) can be solved. However, an approximate solution of this problem with the integral method isa relatively easy matter. The basic steps are as follows: 1. The energy equation (8-30) is integrated with respect to y over the thermal boundary layer. and the velocity component v(x) is eliminated by means ofFORCED CONVECTION FOR FLOW OVER woDIES 363 the continuity equation (8-1). The resulting equation, called the energy integral equation, is given by (see Note 2 at the end of this chapter for the derivation) indsysd, 8-32) where 5, = 5a), = u(x, »)-and 0 = O(x,y), So far, the analysis and Eq, (8-32) are exact, bu this equation cannot be solved because it involves three unknowns, namely, 62)u(x, y),and 6x, 9). Therefore, additional relationships ae needed, 2 Approximations are introduced at this stage to develop simple analytic ex- pressions for u(x, y) and O(x, y) consistent with the physical reality. Since the velocity boundary layer is very thin, the flow velocity over the large portion of the thermal boundary layer i uniform and equal tou, asillastrated in Fig. 82. “Therefore, asa first approximation, the velocity profil is taken as ux, 9) ‘The temperature profile (x, y) can be represented with a polynomial approxi- ‘mation within the thermal boundary layer. Suppose a cubie approximation is chosen for 6(x, ) in the form Ox, 9) = co + ex(ady + cao? + esl)? — in Sy = 5x) (B34) tty, = constant (6-33) and the four conditions needed to determine the four coefficients are taken as O-0 ay=o (8-350) O=1 aty=3 (8-35) a yro ty 350) 26 Gerda (8.354) We note that the first two conditions are the boundary conditions for the problem given by Eqs. (8-31), the third condition is based on the definition fof the thermal boundary layer, and the last condition is obtained by evaluating the energy equation (8-30) at y = 0 and noting that w =. = 0 at the wall surface. The application of conditions (8-35) to Eq. (8-34) gives the temperature profile in the form 1 A») 836) 3. The velocity and temperature profiles given by Eqs. (8-33) and (8-36) are introduced to the energy integral equation (8-32). We obtain all 37)364 HeaT TRANSFER where the right-hand side of this equation has resulted from the relation [28/2y),.0 = 3/26). When the integration with respect 0 y is performed, Eq. (8-37) reduces tothe following ordinary differential equation for the thermal boundary-layer thickness 6, ble “Bde 2, 38) ‘The integration of Eq, (8-38) with the condition 5, = 0 for x = 0 gives the thermal boundary-layer thickness as Base (8-390) (8:39) 4, The temperature gradient at the wall for the cubic temperature profile, Eq. (6-36), becomes ao Aheo 2 and the heat transfer coefficient defined by Eq, (8-27) is written in terms of (x, y) as 3 (8-40) | re 2 on From Bas. (8-0) and (8-4), we have ho = 3 an) By introducing 5, from Eq, (8-396) to Eq, (8-42), the local heat transfer co- cffcient h(x) is determined as ke fig 3k fgxd 3k poe hx fis XV 3K Re Pe (843) 2JBVax” 2yexV v2 28x ‘The local Nusselt number Nu, forthe laminar flow of liquid metals over a lat plate maintained at a uniform temperature becomes Max _ 3g Nu, = 49% = 3 te Pr =[ 0.530 Pet? 8-44) me UR (6-44)once convict FoR row ove nots 345 where Re, = "* = local Reynokls number Prandtl number = Re, Pr= <* = local Pélét number The solution given by Eq. (8-44) is obtained by an approximate analysis. This result should be compared with Poblhausen's [9] exact solution of this heat transfer problem for the limiting case of Pr — 0, given by | Nu, = 0564 Pe! (exact) for Pr +0 (45) This equation was derived under the assumption Pr -+ 0; in practice, this assump- tion implies liquid metals (that is, Pr < 0.05). The approximate solution given by Eq, (8-44), is reasonably close to the above exact result. ‘AL the beginning of this analysis we stated that for liquid metals the velocity boundary layer is much smaller than the thermal boundary layer. To check the validity of this statement, the velocity boundary-layer thickness 4(x) given by Eq, (8-13) is divided by the thermal boundary-layer thickness 4(x) given by Eq, (8-39). We obtain oy) fOr pa 2. FR 2 fie 2652 Pe For liquid metals with Pr = 0.01, we find (8-46) which shows that for liquid metals 6(x) < 5,4). Ordinary Fluids #= Laminar Flow We now examine the determination of the heat transfer coefcient for laminar flow of ordinary fuids having Pr > 1 over a flat plate maintained at a uniform temperature. Its assumed that a fluid ata temperature 7, flows with a velocity 1. ‘vera flat plate. The x axis is chosen along the platc inthe direction of flow with the origin x = 0 at the leading edge, and the y axis is perpendicular to the plate in the outward direction. The plate is maintained at a temperature Tin the region 0 xq. That is, heat transfer between the plate and the fuid does not start until the location x = xo Figure 83 illustrates the velocity and thermal boundary layers for the physical situation just described. We note that the velocity boundary layer is thicker than366 wear TeansreR Secon A NM HET— Velacy and hema Bap oundary layers oF a Aud having Pek the thermal boundary layer because Pr > 1; and 6(x) starts to develop at the leading edge of the plate, while 8,(x) starts to develop at x = Xp where the hea’ transfer section heging. Again we assume incompressible, constant-property Avid in two-dimensional, steady laminar flow with negligible viscous dissipation. The boundary-layer energy equation is taken as 1 Be fore > xy ean ox” by a and the boundary conditions are taken as 0 ay=0 (480) 1 aty= 56) (6-48) where 0 is defined by Eq. (8-29). ‘ince the exact analysis of this temperature problem is rather involved, again wwe consider its solution by the integral method: 1. The energy equation (8-47) is integrated with respect to y over the thermal boundary layer, and the velocity component 1(x, ») is eliminated by means of the continuity equation (8-1). The energy integral equation is determined as 4 a a [fa - oa] = 2% whichis the same as Eq. (8:32). This equation cannot be solved because i involves threeunknowas,namely,3{3).4,»)andO(x, 9). Therefore, additional relationships are needed 2. Approximations are introduced to develop analytic expressions fori, y) and Ax, )) For the velocity profleu(x, ). we choose acubiepolynomial approxima tion as given by Eq (88) and take tin the form 20-30) em ind I. Then A < 1, and in Eq. (8-55) the term yiy * can be netieied in comparison to ba Equation 38) simplifies to sa oan 3 4 oo art gag 100 ae rises ust pots gio = Me 5)368 HEAT TRANSFER 2d _ Vda? de 3 de ‘The velocity boundary-layer thickness 6 was previousty determined as [se Eq, (8-12)) (8-580) and by differentiating we obtain (8-58) The substitution of Eqs. (8-58) into Eq, (8-57) yields an? 394 F ee (8-58) This isan ordinary differential equation of the first order in A?, and its general solution is written as (see Note 3 at the end of this chapter for derivation) Ba WG) = OM + 69) ‘The integration constant C is determined by the application of the boundary condition 4, = 0 for x = xp, which is equivalent to A(x) = 0 forx =x (8-61) We find A(x) = Pro where Pr= : = Prandtl number It is assumed that the heat transfer to the uid starts atthe leading edge of the plate, we set xp -+ Oand Eq, (8-62) simplifies to 560) x) This relation shows thatthe ratio of the thermal to velocity boundary-layer thickness for laminar flow along a flat plate is inversely proportional to the cube oot of the Prandtl number. AG) = GD! Pr-8 = 0976 Pr"? (662)FORCED CONVECTION FOR FLOW OvER BODIES 369 ‘The substitution of &(x) from Eq. (8-58a) into Eg. (8-63) gives the thermal boundary-layer thickness as, 343) = 433 gor pp 6-64) where 4, For the cubic polynomial approximation considered here for x,y), the local heat transfer coeficient h(x) has been related to the thermal boundary-layer thickness 5,(x) previously, by Eq. (8-42), as 3k 260) By introducing 8,) from Eq. (8-64) into Eq, (8-65) the local Nusselt number Nu, is found: fa) = (8-65) _ Meds Nu ‘This approximate solution is remarkably close to the exact solution ofthis, problem given by Pohlhausen [9] as Nu, = 0.332 Pr? Re! (exact) for Re, < 5x 10° | (8-67) Note that the heat transfer relation given by Eq. (8-66) was derived by an approximate analysis on the assumption 5, < 6 or Pr > 1. However, its com- parison with the exact results shows that its valid inthe range of 6 < Pr < 10, which covers most gases and liquids. 0331 Pr Rel — for Re, <5 x 10 (8-66) For very large values of the Prandtl number, Pohihausen’s exact calculations show that the local Nusselt number Nu, is given by Nu, 339 Pr? Rel’? (exact) for Pr + 20 and Re, < 5 x 10° | (8-68) To calculate the heat transfer coefficient from the above relations, itis recom ‘mended that the fluid properties be evaluated at the arithmetic mean of the wall temperature T,, and the external-flow temperature T,,, that is, at T; = 4(Ty + Tao) which is called the film temperature. In engineering applications, an average heat transfer coefficient hy over the length of the plate from x = 0 10 x = L is defined as ra nat fais 66)370 ext TRANseER ‘Noting that h, ~ x°1!, we find the average heat transfer coefficient for laminar flow along.a flat plate over the length x = O10 x = L,as hg = 2M) 670) ‘Then the average Nusselt numbers for laminar flow along a fat plate are given by Nu, = 0664 Pr! Re}? (exact)06.< Pr<10 | (&71a) Nu, = 0678 Pr!” Re}? (exact) Pr+ co en) where ale Nog = "9 ‘and the properties are evaluated atthe film temperature. Equation (8-718), derived for the limiting case of Pr -» 20, is applicable for fluids having large Prandtl ‘number, such as ols. ‘Turbulent Flow {AN transtion takes place from laminar o turbulent flow inthe range of Reynolds rumbers from 2 « 10° to 5 x 10° for Row over & iat pte. Heat transfer cor Felations can be developed for turbulent ow over fat plate by ting the relationship between theheattranserand the drag coeficients given by Ea. (6-150) St, PP = fe, @7 For example, ifc, is obtained from Eq. (8-24), we find St, PH = 00296 Rez? for $x 10? < Re, < 10" | (8-73) oF ¢, is obtained from Eq, (8-25) to yield St, Pr = O.185(log Re,)"25 for 107 < Re, < 10 | (8-736) and all properties are evaluated at the film temperature. More recently, Whitaker [10] used the experimental data of Zukauskas and -Ambrazyavichyus [11] and modified the Colburn [12) expression Eq. (7-68) 0 {develop the following cortelation for turbulent boundary layer along a fat plat: Nu, 029 Reo Pr°+? 74)[FORCED CONVECTION FOR FLOW OVER DoDIES 371 valid for Re, > 2 x 10* to 5 x 10°; and all properties are evaluated at the film temperature, In practical applications, the average heat transfer coefficient hy over the distance 0 < x < L of the plate is of interest. When the flow is turbulent, itis always preceded by a laminar boundary layer in which the equation governing the ‘heat transfer is different from that for turbulent flow. Therefore, the averaging must be performed over both regions as now described. Let the flow be laminar over the region < x < cand turbulent over the region ¢ Re,, where Re, transition. Clearly, Nu, as given by Eq. (8-76), depends on the value of the critical Reynolds number for transition from laminar to turbulent flow. The free-stream turbulence level affects the transition. When high-intensity turbulence is generated in the free stream, the transition to turbulent flow takes place at a lower critical Reynolds number. However, if care is taken to eliminate free-stream turbulence, the transition to turbulent flow is delayed. For a critical Reynolds number Re, = 2 x 10°, Fq, (8-76) yields Nu, = 0.036 Pr°+%(Re?:* — 17,400) + 297 Pr”? em Nu, (8-756) 1oL/v and Re, = critical Reynolds number for[372 Wear TRANstoR ‘The last term on the right-hand side can be approximated as, 297 Pr'it = 297 Prot? and the viscosity correction can be introduced by multiplying the right-hand side of the resulting expression by (J1,/j44)°2. Then the following equation is obtained: “yn 7 All physical properties are evaluated at the ftee-stream temperature except ty, ‘hich is evaluated atthe wall temperature. For gases, the viscosity correction neglected, and for such a case the physical properties are evaluated at the fl temperatere. Equation (8-78) gives the average Nusselt number over the laminar and turbulent boundary layers over a at plate for Re, > 2 x 10°.Ithas been proposed by Whitaker [10] and used to correlate the experimental data of several investgae tors [11, 13, 14] for ar, water, and ol covering the following ranges 2 x 10? < Re, < 55 x 10° 070< Pr < 380 Noy 036 Pr°**(Ref snon( i 026 < Be Equation (8-78) correlates the experimental data reasonably well when the free- stream turbulence is small. If high-level turbulence is present in the free stream, Eq, 6-78) without the constant 9200 correlates the data reasonably well. <35 Example 8-4 Atmospheric air at T,, ug, = 20 m/s flows over a flat plate L uniform temperature T,, = 325 K, (a) Calculate the average heat transfer coefficient hover the region where the ‘boundary layer is laminar. () Find the average heat transfer coefficient over the entire length L = 1.5m of the plate. (© Calculate the total heat transfer rate @ from the plate to the air over the length L = 1.5 mand width w = 1 m. ‘Assume transition occurs at Re, = 2 x 108 275 K and a free-stream velocity 1.5m long that is maintained at a SoLUTION The physical properties of atmospheric air are taken as follows at (Ty + Te)2 = 300 K: k= 0026 Wim-*C) Pr = 0.708 y= 168 x 10° miss 198 x 10° kgi(m-s) MeFORCED CONVECTION FOR FLOW OVER BODIES 373 For Re, = 2 x 10% the location x, where the transition occurs is determined as 168 m (@) The average heat transfer coefficient for the laminar boundary layer, if we neglect the viscosity correction, is determined by Eq. (8-71) a) 0026) 5 o99):2(2 x 102 = 06st) 0708 12 x 105) = 410 Wim? 3) for 108) ~ 9200] (©) The heat transfer rate is Q = whh(T, = To) = (1)(1.5X49.19325 — 275) = 3683 W 83 FLOW ACROSS A SINGLE CIRCULAR CYLINDER Flow across a single circular cylinder frequently is encountered in practice, but the determination of the drag and heat transfer coefficients is a very complicated ‘matter because of the complexity of the flow patterns around the cylinder. Figure 8-4 illustrates with sketches the flow characteristics around a circular cylinder; clearly they depend on the Reynolds number, defined as uD Re (8-79)374 wear reansren Reed sere sac Ree s00 Re> S009 Unseparted Bow Pao vertes Pesodis vores Hiaty tarde Figure 8-4 Flow around a cgcla eylndr at various Reynolds numbers ‘where D is the cylinder diameter and u,. i the free-stream velocity. For a Reynolds ‘number less than about 4, the flow remains unseparated and the velocity field can be analyzed by the solution of equations of motion {15}. For Reynolds numbers above about 4 the vortices startin the wake region and the analysis of velocity and temperature distribution around the cylinder for Re > 4 becomes very complicated (16). Drag Coefficient Consider flow at a velocity u,, across a circular eylinder of diameter D, and let F be the drag force acting on the length L. of the cylinder. A drag coefficient cy is defined as. F 680) icDicea2| Here LD represents the area normal to the flow. The drag coeficient cp defined by Eq, (8-80) isthe average value of the local drag coefficient over the circumference of the cylinder. Thus, given cp, the drag force F acting over the length L. of the cylinder can be calculated according to Eq, (8-80). Figure 8-5 shows the drag coefiicient cp for flow across a single eylinder. The physical significance of the variation of cq with the Reynolds number is bette: envisioned if we examine the results in Fig. 8-5 in relation to the sketches in Fig 8-4, For Re < 4, the drag is caused by viscous forces only since the boundary laye- remains attached to the eylinder. In the region 4 < Re < 5000, vortices are formed in the wake; therefore, the drag is partly due to the viscous forces and partly due to the wake formation, that is, the low pressure caused by the flow separation. In the region 5 x 10? < Re < 35 x 10°, the drag is caused predominantly by the highly turbulent eddiesin the wake. The sudden reduction in dragat Re = 3.5 x 10° {i caused because the boundary layer changes to turbulent, thus causing the point of flow separation to move toward the rear of the cylinder, which in turn reduces the size of the wake, hence the drag. Heat Transfer Coefficient Figure 8-6 shows McAdams’ [17] correlation of the average heat transfer co efficient hy forthe cooling or heating of air lowing across a single circular cylinder. The properties are to be evaluated at (T,, + T,)/2. This correlation for gases doesFORCED CONVECTION FOR FLOW OVER HOOKS 378 € reciiSaL ena elie tacit le Tonia ToT os seed atone Figure 85 Drag coefficient for Row across a singe cirelar cinder (From Sclicing (3}) ‘not show explicitly the dependence ofthe results on the Prandt! number, because gases have a Prandt! number of about unity. Therefore, more elaborate correla- tions have been developed by several investigators [18-20] in order to include the effects of the Prandtl number and hence extend the applicability ofthe results to fluids other than gases. Whitaker [18] correlated the average heat transfer coefficient hy for the flow of gases or liquids across a single cylinder by bad ra Nu, coaners +oreresyes(te)" | sn cs which agrees with the experimental data [21-26] within +25 percent in the range of variables 40 < Re < 10° 067 < Pr < 300 025 < "= <52 Me where the physical properties are evaluated at the free-stream temperature except for iy, Which is evaluated at the wall temperature. For gases, the viscosity correc- tion is neglected, and for such a case properties are evaluated at the film tempera~ ture. We note that Eq. (8-81) involves two different functional dependences of the Nusselt number on the Reynolds number. The functional dependence Re®* characterizes the contribution from the undetached laminar boundary region, and‘ussrne 2uunoy a J Bujo09 0 SopRoy Jo} AGUNG DEEN SBELINY og mg lee sr Me yg Oley les ec Wess or One TO ae tt iFORCED CONVECTION FoR FLOW OVER noDHES 377 (© Water ethene yt A Nivoxen 10-9 Water F © wae eo a FW itr satin, i 0 0 1 1° oF 1 Reo Be Figere 87 Nuselt number for Now across singe eirular einer. (From Whtaker [18] Re®? characterizes the contribution from the wake region around the cylinder. Figure 8-7 shows the correlation of Eq. (8-81) with the experimental data of various investigators [21-26] for different uids. ‘A more elaborate but more general correlation is given by Churchill and Bernstein [19] for the average heat transfer coefficient h for flow across a single cylinder as (8-82) which i applicable for 10° < Re < 10” and Pe = Re Pr > 02 Equation (8-82) underprediets most data by about 20 percent in the range 20000 < Re < 400,000. Therefore, for this particular range of Reynolds number, the following modified form of Eq (8-82) is recommended: 0.62 Re"? Pr' Re a wg 08 + REL (gta) ] = for 20,000 < Re < 400,000. In Eqs. (8-82) and (8-83), all properties are evaluated at the film temperature. Equations (8-82) and (8-83) were developed by correlating the experimental data of many investigators, and the fluids included air, water, and liquid sodium with both constant wall temperature and constant wall heat flux.378 EAT TRANSER For the range of Péclét number less than 0.2, Nakai and Okazaki20] proposed the correlation Nu, = (08237 — In Pet)" for Pe < 02 (84) Properties are to be evaluated at the film temperature. ‘Variation of (8) around the Cylinder In the above discussion we focused our attention on the determination of the average value of the heat transfer coefficient for the cylinder. Actually the local value of the heat transfer coefficient h(@) varies with the angle @ around the cylinder. It has a faicly high value at the stagnation point 0 = 0 and decreases around the cylinder as the boundary layer thickens. The decrease of the heat transfer co- efficient is continuous until the boundary layer separates from the wall surface or the laminar boundary layer changes into turbulent; then an increase occurs with the distance around the cylinder. The variation ofthe local heat transfer coefficient (0) with angle 0 around a circular cylinder has been investigated by Eckert and Sochngen [27] for low Reynolds numbers and by Giedt [28] for high Reynolds ‘numbers. To illustrate the complicated heat transfer mechanism around the cylinder, we present in Fig, 8-8 Giedt’s data. At high Reynolds numbers, the heat transfer coefficient exhibits two minima around the cylinder. For example, in the ‘curve for Re = 140,000, the first minimum occurs at the transition from laminar ‘to turbulent boundary layer at an angle 6 = 80°; the second minimum occurs at 9 = 130°, where the flow separation takes place. Example 8-5 Atmospheric air at T, = 250K and a free-stream velocity 4g = 30 m/s flows across a circular cylinder of diameter D = 2.5 cm. The surface of the cylinder is maintained at a uniform temperature Ty, ~ 350 K. (@) Calculate the average heat transfer coefficient hy (b) Determine the heat transfer rate Q per I-m length of the cylinder. () Find the average drag coefficient cp. (@) Calculate the drag force F acting per 1-m length of the cylinder. SoLUTION Equation (8-81) is used to calculate the heat transfer coefficient. ‘The physical properties of air atthe film temperature T; = 300 K are k = 00262 Wim v= 1684 x 106 mijs Pr = 0.708 = 1983 x 10°? kgi(m-s) p= LITT kg/m? ‘The Reynolds number becomes D__0x0025) Toad x 10° = 44537YoRcip CONVECTION FoR FLOW ovER RoI 379 way z Figere 48 Varintion ofthe local cq eat tansfercoficient i) 30a 8 TOOT TTA TRO around civular cylinder for (0, Ange messed from the option poi. epoes ow of ai (From Gied (281) (@) Equation (8-81) is applied to calculate hy: Nu, = (04 Re?’ + 0.06 Re®) Pr?* = (0.4(44,537)° + 0.06(44,537)?°9(0.708)°4 = 139.2 k 0.0262 ; hig = 5 Num = Coop (139.2) = 1459 Whim? °C) If Eq, (8-83) were used, the heat transfer coefficient would be hy = 143.7 W((m? -°C); the two results are very close.2380 Hear TmANSFER (®) The heat transfer rate Q becomes Q = hg(XDLXT. ~ To) = (145.9X0n x 1025 x 1)(350 ~ 250) = 1146 W/m (©) With Re = 44,537, the average drag coefficient cp is determined from Fig. 8-5; eel (@) The drag force F is determined by using Eq. (8-80). A ou LD 1aytyaors) A720) = 146N Example 86 Atmospheric air at T,, = 275 K flows across a L-mm-diameter electric wire that is maintained at & uniform temperature T, = 325 K. Ifthe e dissipates 70 Wim, calculate the ree-stream velocity 1. of the ar. SoLUTION Since the Reynolds number is not known, itis more convenient 10 usc the heat transfer data given in Fig. 8-7. The physical properties of air at the film temperature T; = 300 K are = 0.0262 Wim -"C) y= 1684 x 10- m/s Pr = 0.708 = 1.983 « 10°* kem-s) ‘The heat transfer cocficient hy is Q 70 =DLAT ~ xxi xT w ~ 456 When’ °C) ‘Then IigD _ 4856 x 10-8 Nu, = be - m= one = 17 Prot = (0.708)FoRceD CONVECTION FOR FLOW OVER BoDIES 381 From Fig. 8:7 we obtain Re = 10° 16.84 x 10-* 10 (10°) = 16.84 m/s 84 FLOW ACROSS A SINGLE NONCIRCULAR CYLINDER “The results of experiments for the average heat transfer coefficient hy for the flow of gases across a single, noncircular long cylinder of various geometries have been correlated by Jakob [29] with the following simple relationship: haDe (ey 5 Table 82 Constants ¢ and n of Eq. (8-85) Flow diretion . eee Re~ noe nome QLYo, soov-1n0 assy 0222 we corto, 250-1500 612 02% md nome QTfo, 2500-1500 062s 281 so0o-109mn 06m aus soq0-19500 068 ae sn-109g00 0675 ow 2500-8000 06590160 ee ee = —= O_fo. 19300-190000 0.782 ass se () fo aowo-ssome sot cons Sores: Fakeb 2)[382 Hear TRansre where the constant c, the exponent n, and the characteristic dimension D, for various geometries are presented in Table 8-2. The physical properties of the uid are evaluated at the arithmetic mean of the free-stream and the wall tempera- tres 85 FLOW ACROSS A SINGLE SPHERE ‘The flow characteristics across single sphere are somewhat similar to those showa in Fig. 8-4 for a single cylinder. Therefore, the dependence of the drag and heat transfer coefficients on the Reynolds number for a sphere is expected to be of the same form as that for a single eylinder, Drag Coefficient IC F is the total drag force due to flow across a single sphere, the average drag ‘coefficient ¢y is defined by the relation 6.86) where 4 is the frontal area (that is, A = xD3/4) and us the fre-stream velocity, We note that F/A is the drag force per unit frontal area of the sphere. Figure 8.9 shows the average drag coefficient cy for flow across a single sphere. A com parison of the drag coefficient curves in Figs. 8-5 and 8.9 fora single cylinder and 200] 100 Din cetiient ep Not? #6 ght eee He IGEE COQ HE Qh EIQ De Figure 89 Drag coeticient for flow ove a single sphere. (Fim Sclcting 7.)FORCED CONVECTION FOR FLOW OVER Ropes 33 sphere, respectively, reveals that the two curves have similar general character~ istics. Heat Transfer Coefficient For the flow of gases across single sphere, McAdams [17] recommends the simple correlation for 17 < Re < 70,000 (8-87) ‘here his the mean heat transfer coefficient over the entire surface ofthe sphere. ‘The properties are evaluated at (T., + T,)/2 ‘A more general correlation for the flow of gases and liquids across a single sphere is presented by Whitaker (18] in the form Nu, = 2+ (04 Re? + 0.06 Re?!) Pr? (ey (6-88) which is valid over the ranges 35 < Re <8 x 10% 07 < Pr < 380 p< 32 Be and the physical properties are evaluated at the freestream temperature, except Hy, Which i evaluated at the wall temperature. For gases the viscosity correction is neglected, but the physical properties are evaluated atthe film temperature. Equation (8-88) for a sphere and Eq. (8-81) for a cylinder have the same functional dependence of the Nusselt number on the Reynolds number, except for the constant 2 in Eq. (8-88). As Re +0 (ie, no flow), Eq. (8-88) assumes a limiting value Nu = 2, which represents the steady-state heat conduction from a sphere at a uniform temperature into the surrounding infinite medium. Figure 8-10 shows a correlation of Eq. (8-88) with the experimental data of Refs, 30 to 32 for air, water, and oil. Equation (8-88) represents the data reasonably well Example 87 Atmospheric air at T, = 250K and a free-stream velocity ug. = 30 m/s flows across a sphere of diameter D = 25cm. The surface of the sphere is maintained at a uniform temperature T, = 350 K by electric heating. (a) Calculate the average heat transfer coefficient (b) Find the heat transfer rate Q from the sphere.ttt HL 0 wate 130 Fo wert 4 rr) q i 4 & vol eee 9 0 107 7 e Re UeD Figure 8-10 Nusselt number for Row across. sgl sphere, (Fram Whitaker (18) (©) Find the average drag coefficient cp. (@) Determine the drag force F acting on the sphere. SoLtioN Equation (6-88) can be used to calculate the average heat transfer coefiienth,. The physical properties of air are the same asin Example 8-5. Therefore, 2D __ (30,0025) Re = te? 7 = 44537 vy 1684 x 10° 0.7, and N= 10, Here Re is defined as Re = PGan (8-956) # ‘The values of constant cy and the exponent n are listed in Table $3. All physical properties in Eq. (8-9Sa) are evaluated at the mean film temperature. Kays, London, and Lo [39] examined experimentally the effects of the row number on the heat transfer coefficient for a variety of tube arrangements. For tube bundles having less than N = 10 transverse rows in the direction of flow, there was some reduction in the heat transfer coefficient. Based on the results of, their experiments, the heat transfer coefficient hy for N < 10 could be determined by utilizing the following relation: hy =Cihyowy fort N= 10 (6:96) ‘Table 8-4 lists the values ofthe correction factor ¢, for both in-line and staggered tube arrangements, with N varying from 1 to 9. The results depend only slightly ‘on the Reynolds number.Table 8-3 Constants cy and n of Eq. (8-95) Ss) D 130 20 30 Arrangement St == = an oo" @ 5 &@ - % —— 02s 066 as ~ 044 osm oant ase 10 oan ase : mas osm 056s 0360 Smepred 1359 asia 0556 ass asi9 0556 ose 1500851 Os68 562 0452 068 0568 20 ao sie ses 04k 0556, osm 300g 0592 ase caso ase ost 138 osm ome 0100 0708 078 os 1500367 0586 9620 C11 om one 20 o4i8 05m osm 02 062 re 08 30029 4@1 0357 0sks aT asst ‘Source: Grinioon [94 More recently, Zukauskas [47] reviewed the work of various investigators ane proposed the following correlation for the heat transfer coefficient for flow across tube bundles: 697) where Pr, is the Prandtl number evaluated at the wall temperature, and (0 for gases for liquids which is valid for 0.7 < Pr < SOQand N > 20. For liquids, the physical properties are evaluated at the bulk mean temperature, since the viscosity correction term is included through the Prandtl number ratio. For gases the properties are evaluated at the film temperature and the viscosity correction term (Pr/Pr, is omitted. ‘Table 8-4 Correction factor c, for Eq. (8-96) ieee 2 se eee ee) Iniime 06080087090 092 094 096 a9 am Staggered 068 075 083089 092 095 097 098 099 Source: Kays, London, and Lo (39)FORCED CONVECTION FOR FLOW OVER HODIES 389 ‘Table 8-5 Constant c; and exponent m of Eq. (8-97) Semey Fe a = = nies wih ow (Grol Lngeand moderate nga ich can beep sng iors to om 08 dei eie On ou Sat 0° ow bow 20 pret ihe an ht fr ig be ora a3s{%)" se wor oas() 080 g< $ wrx 040 ow Sea 2x Wwe — aon oss ‘Source: Zukauskas (17), ‘The coefficient ¢; and the exponent m were determined by correlating the experimental data for ait, water, and oil reported by numerous investigators. ‘Table 8-5 lists the recommended values ofc and m of Eq, (8-97). Equation (8-97) correlates the experimental data very well for tube bundles having N = 20 or more rows in the direction of flow. For bundles having less than N= 20,ows, the Nusselt number can be found from where the correction factor cy is given in Fig, 8-12 for both in-line and staggered tube arrangements. In the previous discussion we considered two correlations, given by Eqs, (8-95) and (8-97), for the heat transfer coefficient for flow across tube bundles. The latter equation is a more recent correlation which has been developed as a result of comparison with the experimental data over a wide range of flow rates, oof ied Figure 8:12 Correction factor 1S TE 20 Eg (698) (From Zukauskas " tony390 HEAT rRAnsrER | 1 207 Fa : t Zz ; | Ap ~ Ibe = vg Nt optir = ie, — 4 Mas 4 Soa a i Wo 2 € 6a 2 4 HO? 4 HTD «CHI? 4 OHI D4 Figure 8-13 Friction factor andthe correction factor Z fr use in Eq, (8-9) fr in-line tbe a iment (From Ref 47) Prandtl numbers, and tube arrangements, The agreement with the experimental data was shown to be very good; therefore, it should be prefered. Pressure Drop Correlations Zukauskas [47] correlated the pressure drop due to fluid friction for flow across tube bundles by 6-99) where f = frietion factor G = Plime, = maximum mass flow velocity, kg/(m? -s) N = number of tube rows in direction of flow Z = correction factor for effects of tube bundle configuration (Z = for a square or equilateral triangle tube arrangements) Figures 8-13 and 8-14 show the friction factor f for in-line arrangement with square tube and staggered arrangement with equilateral triangular tube, respectively. In these figures x; = S7/D. x, = S,/D, and xp = Sp/D denote, respectively, the «dimensionless transverse pitch, longitudinal pitch, and diagonal pitch. The cor- rection factor Z is plotted as an insert in these figures. Note that Z = 1 for the square arrangement in Fig. 8-13 and for the equilateral triangular arrangement in Fig. 8-14, For tube arrangements with S; # S, of Sy # Sp. appropriate correction factors should be obtained from these figures and included in the pressure drop expression, Eq. (8-99).FORCED CONVECTION FOR FLOW OVER BODIES 381 or , Labor econ «Lf * Ro] Ss | | Tt T L Seno 2 4 RID 4 OE 2 4 OHO 2 4 OME? Fire 814 Feicion factor / and the cotrection Factor Z fr use in Eq. (8-99) for staggered twhe arrangement. (Pom Ref. 47) Liquid Metals ‘The previous heat transfer correlations are not applicable to fluids having very low Prandtl number, such as liquid metals. Hoe, Dropkin, and Dwyer (48] and Richards, Dwyer, and Dropkin [49] reported experimental data on heat transfer rates for mercury flowing across staggered tube banks. In these experiments the mercury flow was across a tube bank 60 to 70 rows deep and consisting of $n tubes arranged in equilateral-triangle array having a pitch-to-diameter ratio of 1.375, The average heat transfer coefficient for these experiments were correlated by (8-100) where Re = DGuy/s¢ for 20000 < Re < 80,000. The physical properties are evaluated atthe arithmetic average of the bulk fluid and wall surface temperatures. Kalish and Dwyer [50] give heat transfer data for NaK flowing through tube bundles Example 88 Water at T, = 24°C is to be heated to T; = 74°C by passing it through a tube bundle in staggered tube arrangement. Tubes have an outside diameter D = 2.5 em and are maintained at a uniform surface temperature292 ear Tansee T,, = 100°C. The longitudinal and the transverse pitches for the staggered arrangement are given by Seats and D p-?° The velocity v., of water just before entering the tube bundle is, = 03 m/s (a) Calculate the average heat transfer coefficient hy, () Calculate the number of tube rows N in the direction of flow needed te achieve the above temperature rise of water. SoLUTION _ The physical properties of water atthe bulk mean fluid tempera: ture 24 4 74)/2 = 49°C ate cp = AIA (ke °C) p = 989 kg/m* k = 0644 Wi(m -°C) = 562 10°* kgm -s) Pr= 3064 ACT, = 100°C, Pr, = 1.74, The maximum velocity ta, fora tube bundle with staggered tube arrangement is determined by either Eq. (8-92) or Eq. (8-91). depending on which gives the greater value for Uma... For the particular ‘geometry, Eq, (8-91) gives greater tga’; hence, from Eq. (8-91) Le Sy{D imu = Me SID 1 354) = 06ms ‘The maximum mass-flow velocity Gay becomes G, 9890.6) = 593.4 kgm? -s) as = Pia ‘Then the Reynolds number is Re = DUinar _(00254593.4) nie Scr DqlOgsaea cei (@) Equation (897) now can be used to determine the heat transfer co- efficient ig with the assumption that NV > 20, For the staggered tube arrangement, with 5,/D = 20, $,/D = 1.5, or Sy/S, = 20/18 <2, and Re = 264 x 10, the constant cy and the exponent m of Eq. (8:97) are obtained from Table 8-5 as( 5) 371FORCED CONVECTION FOR FLOW OVER RODS 393 ‘Then Eq, (8-97) becomes HP amneenon(t)" k a 10089 exseapnf*)" £085 asuesanesasen('t) hy = 8230 Wim? -°C) [IF Eq. (8-95a) were used, the resulting heat transfer coefficient woul! be £6570 Wi(m? °C), which is within 20 percent.) (®) The accompanying figure illustrates the definition of various quantities that we now use to write an energy balance to determine the number of transverse rows N needed: ‘Heat transferred from {tube surface to water ) ~ (by water Aig hTa = MT ~ T) att p = total mass flow rate, kg/s (cise) where M ference between fluid and wall surface temperatures inlet and outlet temperatures of water, respectively394 wear mansn ‘The logarithmic mean temperature difference is used to determine AT, (R= %)~ (T=) _ h-h in. — TAT, — BN in (= TT Now, Ag, As, and AT,, as defined above, are introduced into the energy balance equation (a): AT, a h-T (DLN ET RTT] 7 LMA PATE ~ Ti) Solving for N yields 1S; web jy Te Ti =D by Ty Te (039894174) 100 — 24 330 100-74 102.7 = 103 -te ‘Thus, 103 rows are needed. Example 89 Air at atmospheric pressure and temperature T, = 325 K flows through a tube bundle in in-line tube arrangement, as shown in Fig. &-11a ‘Tubes have an outside diameter D = 1.9 cm and are maintained at a uniform temperature T, = 375K. The longitudinal and transverse pitches for the bundle are given by Sr Sk 2 D”D ‘The bundle consists of tubes L = 0.75 m long tubes, N = 15 tube rows in the direction of flow, and m = 20 tubes per row. The air velocity just before the ‘tube bank is u,, = 8 m/s. (@) Find the pressure drop AP across the tube bundle. (®) Determine the heat transfer coefficient hy. (©) Find the exit temperature 7; of the ar. (@ Determine the total heat transfer rate Q. ‘SoLUTION The exit temperature of the air is not known; therefore, the bulk fluid temperature cannot be determined. To start the calculations, the physical properties of atmospheric air are evaluated at the mean temperature, taken as Tot Ty _ 375 +325 _ 390K 2 2FORCED CONVECTION FOR FLOW OVER ODES 395 ep = 1009 Hfkg-°C) p= 0998 kg/m? k= 003 W/im-°C) x= 2075 x 10°* kgm +s) Pr = 0697 ‘The maximum flow velocity tga, is determined by Eq, (8-91): Sy/D 2 = 16mjs ‘The maximum mass flow velocity Gy becomes Gn = Plgns = 0998 * 16 = 1597 kg/m? 8) ‘Then DG _ (00191597) 2075 x 10 = 14623, (a) Equation (8-99) is used now to determine the pressure drop: Nee ap = Maia I “The fition factor fis obained from Fig. #13, for Re Xen 5yD = 28 14623 and f=022 The correction factor is Z = 1, since Xy = X,. Then (b) The heat transfer coefficient hy is determined by using Eq, (8-97). For the in-line arrangement with Re = 14,623, the coeffciemts c; and m are ‘obtained from Table 8-5: 2 = 027 m= 063 and for gas we set n = 0. Then Eq. (8-97) becomes. aD 0.69 py0.36 am = 027 Re’ ig(.019) .82(0.697926 52 STOR 7 02H14.623)°410.697) hy = 157.5 Wi(m? °C)396 wear ANH () To calculate the exit temperature T;, we consider an energy balance (ie transferred fa) at ie tube surface to air) ~ (away by air MefT, ~ 1) ) where M = Agi p = total mass flow rate, kas freetiow area just before tube bundle ins; total heat transfer surface (surface area per tube)(number of rows)(aumber of tubes rer row) — (RDLXNXm) difference between wall surface and air temperatures inlet and outet temperatures of air, respectively ‘The logarithmic mean temperature difference is used to determine AT. R-T Ty “nT Bl Now 4. Aj.and AT, are introduced into the energy balance equation (2): Boh (OLN nah = TOT = TD) (LoS Xin peeXTs ~ Tr) Solving for the logarithmic term yields L-h_.y2 Tats Sree, 1 ists X15) 9 (x0.998 1009) 7 =385K (Given the ext temperature T,, the heat transfer rate is caleulated from Eq, (a): Q= MeJT; — T,) = M(1009\(343.5 — 325) = 18,666.56 where Ma Ante (LS) 2 (0.75)20)(2 « 0201948X0.998) = 4551 ke/sFORCED CONVECTION FOR FLOW OVER woDIES 397 Then Q = (18,666,5,4.551) = 8495 kW Calculations can be repeated by evaluating the air properties at T+ Ty _ 325 +3435 pases 3343 K but the improvement would be negligible. 8:7 HEAT TRANSFER IN HIGH-SPEED FLOW OVER A FLAT PLATE {At flow velocities approaching or exceeding the velocity of sound encountered in applications such as high-speed aircraft, missiles, and reentry vehicles, the elects of compressibility, viscous dissipation, and property variation with tempera- ture become important. The general analysis of such problems is very involved. However, the heat transfer rate in high-speed flow along a flat plate at a uniform temperature T, can be predicted, for most practical purposes, by using the [ow speed heat transfer coefficient h, for flow along a fat plate with the temperature difference T,, ~ Ty». That is, at high-speed flow along a flat plate, the local heat flux q, at the wall can be computed from 9, = ATs ~ Tan) 101) where Ty all temperature adiabatic wall temperature local, low-speed heat transfer coefficient ‘Note that in Eq, (8-101) the adiabatic wall temperature Tr, replaces the uid free- stream temperature T., commonly used in low-speed heat transfer. To use Eq. (8-101) for high-speed heat transfer calculations, the local heat transfer coefficient i, is obtained from the low-speed heat transfer correlations for flow along a flat plate presented earlier. For example, for laminar boundary-layer fiow, his obtained from the Pohlhausen equation (8-67), that is hex Nu, = "e Foc turbulent boundary-layer flow, itcan be obtained from Eqs. (8-734), (8-730), and (8-74); that is, he Pees 332 Pr!’ Re! for Re, < 5x 10° (8-102) 10296 Rez? Pr? for $x 10° < Re, < 107 (8-103) 185(log Re)"**"*Pr-2 for 107 < Re, < 10" (8-104) 1029 Re2* Pr°# for Re, > 2x 108 t0 5 x 10%, (8-105)398 HEAT TRANSFER ‘The adiabatic wall temperature T, is determined from (8-106) where T,, and 1» are the free-stream temperature and velocity, respectively. The parameter ris called the recovery factor, and itis related to the Prandtl number as follows: For the laminar boundary layer: r= Prt 06 < Pr < 15(3,51] (1071 PE19Pr> for Pr oe [52] (8-108) For the turbulent boundary layer (53). r= Pri (6-109) ‘The physical significance of the recovery factor is better envisioned by consider. ing an ideal pas at a temperature T., with a velocity 1. that is slowed down adi- abatically to zero velocity. The conversion of kinetic energy in the gas to internal energy will result in a gas temperature Tp given by a 110; Tea Te +5; where Ty is called the stagnation temperature. A comparison of Eqs. (8-106) and (8-110) reveals that for r = 1, T,. is equivalent to the stagnation temperature Ty Note that r = 1 for Pr = 1; thus for a gas (Pr = 1) flowing at high speed along @ flat plate, the adiabatic wall temperature is the same as the stagnation temperature For fluids with Pr > 1, the recovery factor ris greater than unity, and the adiabatic wall temperature exceeds the stagnation temperature. For a fluid with Pr < I the recovery factor ris less than unity, and the adiabatic wall temperature is les than the stagnation temperature. In high-speed flow, temperature gradients in the boundary layer are generally large; hence the properties of the fluid vary significantly with temperature, Al- ‘though the exact analysis of such problems is very involved, the effeets of variation of properties may be approximately included in the heat transfer relations given above by Egs. (8-103) to (8-105) if the properties ofthe fluid are evaluated at the following reference temperature [54] Ty, + 0.5(Ty ~ Ty) + 0.22(Tye ~ Te) any Readers should consult Refs. 55 to 58 for further details of high-speed heat transfer. Example 8-10 Air at a pressure P= Jy atm, temperature T,, = 250 K, and velocity u,, = 600 m/s flows over a flat plate L = 0.80 m long and w = 030m wide. Calculate the amount of cooling needed to maintain the plate surface at 4 uniform temperature T,FORCED CONVECTION FOR FLOW OVER RODIES 39) SouUTION This isa highspeed flow problem, and fit involves both laminar and turbulent regions, the heat transfer anaisis foreach region should be considered separately because the recovery factor, and hence the adiabatic wall temperatures, is different for each flow rexime. Laminar Flow Region To start the calculations, we evaluate Pr and cj a 300 K Pr= 0708 cy = 1006,)¢ke-°C) ‘Then the recovery factor is determined by Ea. (8-10): rm Pr'? = (0.708)? = 0.841 and the adiabatic wall temperature Ty is by Eq (8-106), a 00? Ta + 15 = 250 + 0841 Tg r. 4005 K The reference temperature Tis determined by Eq. (8-111) T= Ta + OST, ~ Te) + 02% Tou ~ Te) 50 + 0.5(300 ~ 250) + 0.22(400.5 ~ 250) 308.1 K The physical properties of air at T, = 308 K and at P = Yyatmare taken as Lars 30 p= 1998 x 10° kgi(ms) Pr 6 = 1007 Jilke °C) Note that the Prandtl number and c, at T; are sufficiently close to the values used above to compute the recovery factor; therefore, there is not need to repeat the eafeulations ‘Assuming that the transition takes place at a critical Reynolds number 5 x 10°, we can find the distance x, = 0.0383 kgim? 0.0269 W/(m-°C) a0 Re, He = 1998 x 10- pu B= ses x 600 © * 043m The average heat transfer coefficient for laminar boundary-layer fow over O's v2 043 mis determined by Ba. (71). fe = 068 Re? Pet? 0269 = pa (864K5 « 10°)" 00.710)" = 262 Wien? °C)400 ExT TRAnsrER ‘The heat transfer rate Q! over the laminar flow region is determined by Eq, (8-101): Q Wx: hhg(Te — Ton) = 0.3(0.4326.2300 — 400.5) —340W Turbulent Flow Region To start the calculations, we evaluate Pr and cy again at 300 K Pr= 0708 c, = 1006 Jj{kg-°C) ‘Then the recovery factor ris determined by Eq, (8-109): Pr? = (0,708)""? = 0.891 And Try is found by Eq, (8-106) a 600! Toe = To +r $= 280 + 0891 5g = MBA K ‘The reference temperature T, is determined by Eq. (8-111) Te + OTs — Te) + 02UTe ~ Te) 150 + 0.5(300 — 256) + 0.22(409.4 — 250) 10K ‘The physical properties of air at 310 K can be taken the same as those given above for the laminar flow region, ‘The local heat transfer coefficient for the turbulent flow region can be determined from Eq. (8-103); Tr, hh, = 00296 pu, Pro? Rez?? (0.0383 x 600)-° i998 x 105) * ~ ausereoxsdeonnnar( = 528x-9? gran [hed = 884 WHme"0) ‘The heat transfer rate Q' over the turbulent flow region is Q, = hy(L ~ XXTo ~ Tow) 8.40.3 08 ~ 0.43300 ~ 409.4) = 70.2 WCL +11 =! ameada 1 > "24> sor ey e240 MOL waqeqim por se womnqin pur sey ease (tee) 0H gud 9990 "ON seo waco“ "NC sean, cw ewe we» eae oro (es) (sew) ae ew) reojaqia row OL + 5 5 Toy sje ep # 210 wot rome aes) OL 55 #20} md IY BHO MOLE awe oop Hore) ovo ray ues wonmne) UNE ota woven "-Ipoq 1940 wop.s9su09 pood0) 405 suoTEye1I09 Jo KrEMMUUNS 9B MITEL, ao2pufoserosDuou Bus # sre MOL 20> 920) 2p eu roe wo. r0< day = ay Situs wore moLg a or > 28> ,01 20) sep zs>h>ero woe > Ad > 90 01> 24> oF soy apa 28 98 wot st> Gy sro owe > 2a>oen WO1x 55> %¥> O12 eee eee (ore ew oe ee (Shame ston ausmoeon ace > a> ames sear ares (PomnueD) 9-9 7481,ioe > 2a > onioc Sieuateeise sam ain panties 0338 may NIA sezpung ag sssb8 mo 30} doup ans oe 5 y 5 1 fumey setpung ag 205 (Le) ba Ua nN 2 1 os > 4 > £125; sopung agm sSIEE MOL o1> ws runny seypung am 19 (6) ba Ag ony 01:09 OF 01 NL0 < 24 toon > 2a > cone oj por moy jo woop ut wou a280eN su 2 of BuRy sexpang aq SIR MOL oe > a> 40 Ol 2u> se oj aay fuse ss. MOLL ‘oo. > oy > LL sy anyde fas» sans se jo mn soso 20) 80 + £9 = "ON ya pow cg kyo uo avez perf “ alma 2 aton! pr wonaa.e oes zat ous uno @ lage (Jacarr-g eae sw won pu ua2ya09:A/™9g = 243 ok PNR Cope tame puree etm 7a ec0= "ON or a os) uss) (60) (69) oon) wa z404 wear TRANSFER The total amount of cooling is determined by adding the heat transfers for the laminar and turbulent regions Total cooling = 340 + 709 = 1049 W 88 SUMMARY OF CORRELATIONS In Table 8-6 we summarize the correlations of heat transfer and drag coefficients for forced convection over bodies presented in this chapter. PROBLEMS ‘Drag force fot laminar fm over fat plat 81 Atmospheric airat27°C ows along fat plate witha vlaciy of, ~ 8 ms, The critical Reyokls ‘numb for transition fom laminar to turbulent fom i Re, = 3 ~ 10" (@) Atwhat distance rom the leading edge of the plate doe transition occu? (©) Determine the velocity boundary ayer thicknes and the local drag coefcient tthe location where the transition oscue, (6) Determioe the average drag coefcien over the distance whete the ow is lamina. Answer: (2) 98 mb) 7 mm, DOHA) DU018S £82 Atmosphere ar at 7"C hows overa fat pate witha velocity of. = 10 mis Pot the variation of the thickness ofthe velocity boundary layer (2) andthe lea drag coefficient aa funtion ofthe Aistace along the pate up to the location where the transition takes pace ro lina Yo turbulent flow at Re = 5 10 £83 Determine the thickness ofthe velocity boundary layer (and te lea shear sees, ab = 21m ‘tom te leading edge ofa fat pate forthe boundary-layer flow of air and hydrogen at stmospheric pressure and °C with velocty of uy = 2s £84 Determine the thickness ofthe velocity boundary ayer A) andthe local hea sre atx = OS m Som the eng cg of ft plate forte boundary layer Row ofengineoiland ethylene pyolat 80°C witha velocity of = 2 me, {85 Determine the thickness ofthe velocity boundary lye ts), the local dracon ,, and the Sm fom the leading eige of fat pate forthe boundary layer flow ofairatT, = 7PCwitha velocity of, = 2 m/s (205, (01 and (c)2atm. 186 Consider a fat pat of length LL = 1m inthe x ditetion and with w = 3 m inthe y direction Airat T, = 27°C and atmospheric prsure ows slong this plate inthe x diction with a veloity af ‘ys. Calculate the total drag fore exerted on the plate What woul the drag ore bef the air ‘ow were inthe y direction? “Answer 00263 N, 00182 N £87 Determine the drag force exerted on 2-m-ong Bat plate pr I-m wider the fw of the following Aid at atmospheric pressure and 350 K witha velocity of a. ~ Sms (a at (b)hydeopen, and (c) {88 Show thatthe second-degree polynomial epresentation ofthe velociy profile for Nw along. at plat subjet to the conditions o| =o bee ay at Bl no6 £89 show thatthe fourth-dereepolynomia representation of he velocity profile for Aow along. a fat pte pen by Lor) 140 Using he momen ie are a [flee -omar] Bl aod saci pole epee yas expreson efor 9) in (42) we 35) for laminar boundar:ayer flow along a at plate, derive expressions forthe boundary layer thickness 1X) the local dragcoeticient andthe average drapcoeisentc, ovr the lagth 2 L- {11 Repeat Prob. 8-10, using nea velocity prot ive in therm w fagiven by om given by ” 3 4412 Repeat Prob. 10, using a second degree polynomial representation of the velocity profile given inthe frm “or-5-() ‘ag fore for turbulent fom over x at plate 8:13 Atmospheric air at 27°C flows with a teestream velocity of w= 10 m/s along. 2 at pate {= 4mlong.Computsthe dragcoeffcent and 4m fom the lading ge Assuming an alurbulent ound ayer determine the drag ore exerted per Im width ofthe plate ‘Anower” 356% 10% 31 x 10">s091 N {14 A fui a 80°C flows with a reesiteam velocity of, = 8 m/s along a fat pte L = 5 m fog. Compute the lca drag coeficent andthe shar ses at the ralingedge hati, L = $m ofthe plate for (a) ai, (b) CO; () water, and (eter lol 15 Air at 27°C flows with a fee-steem velocity of. ~ 40 mis along a Mat plate L = 2 Tong. Cleat the bouaay-layer thickness a the ond ofthe plat fora at a) (8) and) 2 ate “Answer: (a) 318 (8) 307, (€) 282 em £8.16 Ina low-speed wind tunnels at I atm and 200 K flows over a at plate with fre steeam velocity ‘of $0ms Determine he ength the plateostudy boundary layersat Reynolds numbers upto? » 10" ‘What ithe boundary-layer thickness at the ailing ede of the plate? B17 Air ata presure P= 06 aim and temperature T, = ~15°C ows with a eestream velocity tr. = 120 mover the wing ofan airplane. The wings L~ 2 longi the ection flow and canbe ‘pstded aaa plate Determine thelocaldragcoeiint ad he sear stress athe ralingedge ofthe ‘ning, What she drag force pr meter widih othe wing? eat rans for lamina fl oer at plate {8.18 Determine he velocity andthe thermal boundary-layer thickness aL. = 05 m om the leading ge ofa plate a 50°C for fow at T, = MPC, velocity 4 = 0. miso ai at atmospheri pes ‘ue water ang (engine oi Compare the ratio 8/8 of ternal to velocity boundary lay thickness406 ea TRaNsrin ‘19 Determine the velocity and the thermal boundary-layer thicknesses L. = 0 m from the lsding ge ofa at pliteat 74°C for ow at T.,~ 80°C and amoapherc pressure witha velocity OT fair. hydrogen, and heim, respectively. Compare the rato 8/8, Answer: 3 = 92,261, 254 rsd, = 95.248, 264 mam {820 Engine oil a 4°C flows with velocity of = I ms over a 2-mong Rat plate whose surface 'smainianed ata uniform temperature of HC, Deermine he average eat transfer cfc ove the 2am length ofthe plate Answer: TAS Win? °C) 1221 Air at atmospheric pressure and 40°C flows with a velocity of ua = Sims over &2-anlong Nat Plate whose surface is kept at a uniform temperature of 120°C: Determine the average heat taster fin oer the 2m length othe plate Alo find the rate of ea transer betwen the plat and the ‘per Lom with ofthe plate {822 Airat atmospheric pressure and T. = SC fows witha veloc of, ~ 10 ms oer among Aa plate maintained at T, = 200°C. Caleulate the average drag and het rafer cocci over the ‘em length ofthe plate. Determine the rate of ea transfer between the plate and at per meter wilh of the plate {822 Mercury at T_ = 30°C flows with a velocity of _ = 01 m/s over a fat plate maintained at T, » 120°C Assuming that the transition fem laminar otarbulent flow takes plage at Re, = 3 10%, termine the average eat transfer coeficien over the ength ofthe pate mbere the Now lamin “Anower: 230 Whe? °C) £24 The following information ie pven: Pret forcily Pr=07 forges Pret for iguid metals ‘Sketch the velocity and thermal boundary-layer thickness with distance fom the lading edge of at pate fo laminae ow ofeach ofthe above Bui {828 The local beat transfer coeficient h, fr laminar ow along a fat plate varies as <-", with x ‘being the distance measured from the leading ee ofthe plate. Develop am expression forthe average ‘ae ofthe heat wransfer coeficient hover the distance rom the lading ede ofthe plate {826 Determine the rato ofthe average Nusselt number Ni over the length Lf lt plate tothe focal value ofthe Nuss number Nay atthe lation Lite local heat tans coefiiem hy varie as where isthe distor fom the lading ee ofthe plate 1827 Atmosphere airatT, = 24°C Rows witha velocity of = 4 ms along a Rat plate L thats maintained at» uniform temperature of 107°C (a) Determine the thermal boundary-layer thickness 3 and the local at transfer coefficient at the tralng edge that, L'= 2m) of he pate (b) Find he average beat transfer coer ver the eit ent the plate (©) Caeulate the heat trans rate rom the plate othe ai per meter with of the plate, {828 Consider the ow of sir, hydrogen and helium at atmospheric presure and 77°C with eveloity fv, = 4mysalongs fit plate L 2m long Determine the thickener o the velocity andthe thermal ‘ouidary ayers and the valu fhe local het rans coeficent Im rom the eading edge of he plate {£29 Determine the thickness ofthe ermal boundary layer 3, and the lea! het transer coeficent ‘ha distance L = m from the lending ede of «Rat plate for the ow ofa at 77°C with velocity cof = dims at presures 05.10, and atm 30 Aira simospheri pressure ata mean temperate of C flows over fat plate £ = 2mong wth ‘elocty ofa ~ 4 m/s Plt the local and average het transfer coclcents sanction ofthe dtance From the leading ede of he pate 8.31 Ethylene lycolata mea temperature of WC Rows overs S-n-ong fat plate witha veo of ug = 2 my Cakeulate the average heat transfer coeticen oer the entie length of the plat “Anon 640 Wi? "C). alongFORCED CONVICTION FOR FLOW OVER BoDIES 47 {£32 Helium at atmosphere preure and 20°C flows with velocity of 10 m/sover aft plate L = 2m Tong that maintained ata uniform emperatare of 140°C. Calculate the rate of heat los fom the plate Derieter wath the plate What the dag orc exerted onthe plate? 2:39 Atmospheric airat 2°C Hows witha velocity of2 my over the 3 m by 3m surface ofa wall which banc solar nerdy fu at arate of $00 Wim! and diate heat by conection into the arsteam ‘Aewming that theater surface of the wall has egigible het los, determine the average temperature ‘tthe wall under equilibrium conditions “Anower. 180°C 1.34 Atmospheric ar at 2°C flows witha speed of 4 mys over a 05m by 0.5 m fat plate which i ‘favlorny heated with an electric heater at rat of 2000 Wr. Calulte the average temperature of the plate 1838 Consider the laminar boundaryayer flow ofa guid metal with velocity ug and temperate along fa plate Kept ata uniform trperalur 7. Derive the expression fr the thermal boundary Iyer heen fs) ad the lel Nal mer Na = hk by wing linear role fr the tempera fare distribution given ia the form Tena Ty T.-% 56) 1836 Repeat Prob .35 using second degre polynomial representation or the emperatue profi 1.37 Consider laminar boundary-layer flow of fd having a Prandtl aumber Pr = 1 with a velocity tie and themperature longi plate kept at onfor temperature 7. Derive the expressions for the thermal boundary layer tikes (x) and the local Nusselt number Ny, = hs) by using ‘elocity profie for the velocity distribution and a second-degree polynomial representation forthe Temperitore dstibution. Compare hires with hose derive i ths chapter by sing cubic velocity tnd temperature profiles 1.38 Consider the laminar boundary-layer flow of gui tl wit yeloity vg and temperate T., ‘ver fat plate maintained at uniform temperate T.. Taking the temperatore profile inthe form MDT or ‘where T the wal temperature apd 9, the thermal boundary layer thickness, develop an expression {orth acl heat trae coelcent , ae = Hest rae for turbalet lw over 2 at plate £839 Air at atmosphere preseure and 24°C flows witha velocity of = 10s along » fat plate ‘Ls dim which is maintained at seiform temperature of 130°C. Assume Re, ~ 2% 10° (a) Cakeulte the local bet tans oeeint atx = 2 .and 4m from the leading edge of he plate (b) Find the average eat trans oeticient over L = 4m. {6} Determine the hea transfer rate rom the pate ote air per meter width of the plate ‘Answer (a) 228,211,199 Wim? -°C) (9) 233 Wis? "C);(€) 9880 Wi {40 The lca eat wanser coefficient fr turbulent flow along fat platecan be determine rom 29 = 00096 Rez! for Sx 10° < Re, < 10° sy ‘Asuming that this relation i valid rom the Fading edge of at pate, develop an expression forthe {erage vale ofthe het transfer coefiiet over the length Lo at plate 41 A hid at T, = 40°C ows with a velocity = 8 ms longa at plate L = 3m long which ‘aitaied at a uiform temperature o 100°C Calculate the loa! beat transfer coetiient atthe end of the plate and the average heat rar coeficent over the entre ngth ofthe plat fo (a air a atmos phere pressure and (0) ethylene heal Assume Re, = 2 10%408 wear TmANSHER 8.42 Hum atm. = 30 mand 30 K fows over aa pate = S mlongand w = t mide which s maintained a uniform temperature of 0 K. Cale the average hes rane coefiient and the total heat ansfer rate (Assume Re, =2 x 108) 849 Air at T, = 20°C flows along a fat pate = 41m which is maintained at a ifm temperature ‘of 130. Calculate the average het tanse coefficient over the entre ngth of the plate and the heat tranfr rate per meter with ofthe plate for (a) = 5 (0) y= 10, and () vy = 20 (ASME Re,= 2x 10°) {44 A highly conducting thin wall L = 2m long separtes the hot and cold atstreams Bowing on ‘both sides paleo the plate surface. The ho stream atthe atmosphere presre ad es temper {ure T= 280°C and velocity ig ~ 50 ms. Thecld sretram als at atmosphere pressure a has temperature 7, ~ 50°C and velocity uj ~ iS m/s Calculate the average heat transfer coef or ‘both sistreams and the total heat transfer rate between the streams pe meter with ofthe separating lat. (Assume the wall atthe arithmetic mean of T,2ndT-fr the calelation of pial properties and fake Re, = 2 105) Flow acroat single ytindr 845 A fuid t 27°C Rows witha velocity of 10 ms across a S-om-OD tube whose surace sep at 3 efor temperature of 120°C. Determi the average heat tans coeficients andthe hea raster "tes per meer length ofthe tube fo (a) aia atmospheric presi) wate, () ethylene pyc, Answer: (a) 38.3 (6) 31,80, (6) 14828 Wi? °C) {8.46 Consider the flow of air at 2"C witha velocity of w= = 20 m/s actos single eyinder of outside diameter D = 25 cm. The surface ef the syindet i tsintained at aunifoctemperatare of T, = 12°C. Determine he average heat uansler coeficet and the het tans rate rom the tube to ‘heal pr meter lent ofthe tbe for (a1 (8)2.and() atm 8.47 A fd at 80°C flows with a fce-streat velocity of 10 ms accom tube wth D = Sem OD. Determine average drag euellcient andthe dag free per meter length ef the ube for () at 1 st, (©) CO, a 1 ate, (o) water, (dethylene gles, nd) engine ol ‘8.48 Consider the How of air at 77°C and with amass flow rate of uc = 25 kgm?) across a single exlinder of outside diameter D = 2. em, Detemine the dag force exerted per meter length ofthe tube forsirfowingat (a) ,() 2nd @)4 8m. 849 Engine iat 20°C fons witha velocity of m/sacros 2S-cm-diametr tube whichis maintained ats anform temperature T,~ 100°C. Determine the average het transercoeicent and the he ‘wane rate between the tube surface and the oper meter length the ube “Answer: 3206 Wit -"C), 20.14 Wit {50 The thermal insulation iemoved fromthe I-m-lngth section of stam pipe of outside diameter 1 = 25cm and earyng high pressure hightemperature steam at 180°C. Airat °C owing arrss the eiposed section wth s velocity oft, = 6 ms Determine the average heat ranslercoeiceat snd the rate of eat oss fom the Iamlength expose section of the ube int the ait {St Atmosphere airat 27°C flows witha velocity of ~ 10 ms across singe duct whos surfaces kept uniform temperature T, = 100°C. Determine the average ea trae coeficient and the heat lwansfrrate per meter length o tbe duct having the following dimension and configuration (@) A cteular tbe with D = 125m OD (@) A square duct of 1.2 em by 12cm eross section oriented euch that one fits lateral surfaces is erpendcult to the direction of flow 52 Water at T, = 20°C witha frestream velocity, = 1 mys owe serosa singe circular tube of ouside diameter D = 25m. The tube surface is maintained 3 a uniform temperature T. = IFC Cleat the average heat transfer coefficient hand the het transfer fate per meter length the ube 1859 Atmospheric air at T= ~ 300 K and witha freeze velocity of. = 30 mys Rows across 4 single cinder of outside diameter D = 2em, The elinder’ surface i at aniform temperate of T, = 400 K. Calculate the mean heat tanse coefficient hand the eat rans ae er mete length ofthe etinde.Flow across single sphere {854 Consider the fw of Mid at BOC and witha mass ow rat of pug = SO Kan +5) across a finge sphere of diameter D ~ Sem. Compute the everage drag concent andthe dag fore exerted fn the sphere for (a) ae 1 atm, (6) CO, a1 atm, () water, ethylene glyco, and () engine il. 2.55 A tid at 80°C Rows witha feesteam velocity of 10 ms across a Sem-dameter single sphere ‘Compute the average dag coefcent and the otal dag force exerted on the sphere for (a) ait at | an, (@)CO, a at, (0) water, (ethene glycol and () engine ol. {56 A fiat °C Rows with a velocity of? ms across a 2S.cm-iameter sphere. The surface ofthe Spheres maintained ata uniform emperatre of 100°C. Compute the average heat anser cefcent Ad the rate of ent rae rom the sphere othe Ri fr() ar at {atm (6)COs at Fate) water, (@ ethene pyc, and (6 engine oi 8.57 Atmosphoicsr at 20°C Rows witha fee-steam velocity, = Sm/sovera?-m-diameter spherical tank whit mmitlned at 80°C: Compute th average hem tansler cote en the hea transfer tate rom the phere tothe i “Answer: 63.2 WiC), 298 W 8.88 Water at 20°C flows with a fee-steam velocity of I ms over a 2S-emdiameter sphere whose ‘surface is mintineata uniform temperature of 140°C. Dterins the average heat transfer coeficient ‘od the rate of beat los fom the sphere to the a Flow cro tbe bundles 1859 Ai at stmospherc presure and 27°C flows over a tube bank consisting of D = Lem-diameter tubes 10 rows dep, The fw velocity before the ar eater the tube bundle is {mys Determine the verge heat ener coefct forthe following two cases: (a) Tubes are in equlateral-tanguar arrangement with Sx/D = Sy/D = 1.25 (0) Totes are in square arrangement with S,/D = Sy/D ~ 128 “Aner (a) 106, (8) 101 Wie -°C) 1860 Air at atmospheric pestre and 27°C flows over a tube hank consisting of D = t-em-diameter tbe 10rows dep. The ow velit before the sirenters the tbe bank su, = {Sam Determine the sverage fection ator and the pressure drop forthe folowing configurations: {e) Tuber rein equisteratirangular arrangement with SD = Sp/D = 1.25, () Tuber arein square arrangement with Sy/D = Sy/D = 128. Answer) f = 06, AP = 1987 Nim:(6) — 053, AP = 1755 Nim? {OL Air at 227°C and 1S alm pres ows ove a tobe bank consisting f 1 25-cm-OD tubes 10 rows deep inthe direction of ow and forming tack 40 tubes high Tues rei iin arrangement with S,)D = Sy/D ~ 2. Thea velocity belore entering the tube bank is, = 4 ms. Determine the ction {ett and the pressure drop for fows accoss this tube bank, {62 Repeat Prob 8651 forthe staggered arrangement with SyD ~ Sy/D = 2 {465 Aira 77°C ows overa tube hank conssingo2S-em-OD. 2-m-longtubesin inn arrangement swith $0 = {Sand S,/D = 2. The tbe bank i 1 rows dep, forming a stack 20 tube high, The ae ‘locity before entering the te bank iz, = 10 mye Tubes are maintained at a uniform temperature 11 100°C. Determine the average ricton factor andthe peste drop for ow across the abe bank ‘Anone 03; 410 Ni? 1464 Air at 100°C and S atm pressure flows over a tube bank consisting of 2.cm-OD tubes 40 rows ‘exp inthe Girection of flow: Gna forthe flow 20 kph’ =). Determine the average heat transer coeicient or a staggered aeangement with SUD = 1S and Sy/D = 20. {65 Hot ue ges at 575°C fw acrosatbe bank consisting of L2S 25, [eo bor ~ ax are a(f ato] © Which is the energy integral equation given by Eq. (8-32). 3. The derivation of equation (8-60). The differential equation (8-59) is of the following form: wo where 4 and B are constants. A particular solution ¥, of this equation is given y=2 @FORCED CONVECTION FoR MLOW OVER nODHES 415, and the homogeneous solution Yj, that satisfies the homogeneous part of this ‘equation is given as Yam Then the complete solution becomes e yeex* ® where C is the integration constant. This solution is of the same form as that given by Eq. (8-60).CHAPTER NINE FREE CONVECTION Tn Chaps. 7 and 8, we considered heat transfer in forced convection, in which the fluid motion was imposed externally by a fan, a blower, or a pump. Also in some situations convective motion is set up within the fluid without a forced velocity Consider, for example, a hot plate placed vertically in a body of fluid at rest, which {sata uniform temperature lower than that ofthe plate. Heat transfer will ake place first by pure conduction, and a temperature gradient will be established inthe fui. ‘The temperature variation within the fuid will generate a density gradient which, in a gravitational field, will give rise, in turn, to a convective motion as a result of ‘buoyancy forces, The fluid motion set up as a result of the buoyancy force is called {free convection, oF natural convection. The flow velocity in free convection is much smaller than that encountered in forced convection; therefore, heat transfer by ree convection is much smaller than that by forced convection. Figure 9-1a illustrates the development of the velocity field in front of a hot vertical plate owing to the buoyancy force. The heated fuid in front of the hot plate rises, entraining fluid from the quiescent outer region. Figure 9-1b shows a cold vertical plate in a hot fluid. In this case, the direction of, ‘motion is reversed; namely, the fluid infront of thecold plate moves vertically down, again entraining fluid from the quiescent outer region, In both cases a velocity boundary layer is developed, with the peak velocity occurring somewhere within the boundary layer. The velocity is zero at both the plate surface and the edge of the boundary layer. Furthermore, in the regions near the leading edge ‘of the plate, the boundary-layer development is laminar, but, at a certain distance from the leading edge of the plate transition to a turbulent boundary layer begins Eventually, a fully developed turbulent boundary layer is established. ‘We now consider a fluid contained in a cavity or between two parallel plates arranged horizontally as illustrated in Fig. 92a. Suppose the lower plate is main- tained at a temperature higher than that of the upper plate (that is, T, > T,) a6REE CONVECTION 417 Figure 9:1 The laminar and tarbulent velocity ‘boundary layer for Ire convection oma vertical 6) Hot wat ©) Cosma plate ‘A temperature gradient will be established in the vertical direction. The layer will be top-heavy, since the density ofthe cold fluid at the top is higher than that of the hot fluid at the bottom. Ifthe temperature difference is increased beyond a certain critical value, the viscous forces within the uid can no longer sustain the buoyancy forces, and a convective motion is set up, giving rise to circulation patterns. Thus again we have natural convection in the fui, Suppose. in Fig. 9-26, the lower plate is cold and the upper plate is hot (that is, T, < T,).In such a case, the density ofthe top layer is less than that of the bottom ‘Hot wll Cold wall, Temperature <== Deny (a) Lower pte ot (0) Upper pate or Figure 9-2 Fuid contained berween two horizontal plates418 ear TmanseeR layer; then the fluid is always stable, and no natural convection currents can be set up. Energy transfer by free convection arises in many engineering applications, such as a hot steam radiator for heating a room, refrigeration coils, transmission lines, electric transformers, and heating elements. Free-convection currents also can beset upat high-speed rotation in which temperature gradients alter the density in the presence of centrifugal body forces. This results because the magnitude of the body force due to centrifugal effects is proportional to the fluid density ‘Buoyancy-induced flows in water are in evidence in nature also. For example, short-term circulations of water due to the solar heating and the seasonal thermal inversion of lakes are buoyancy-induced free-convection motions. Intthis chapter we examine the energy transfer by free convection for situations, in which the convection currents are set up by the buoyancy forces. The analysis of free convection is a complicated matter. To provide some insight to the physical significance of the factors affecting heat transfer by free convection, we briefly discuss the dimensionless parameters of free convection, and we present a simple, approximate analysis of laminar free convection on a vertical surface. Various empirical and semiempirical correlations of free convection are then given. 9-1 DIMENSIONLESS PARAMETERS OF FREE CONVECTION ‘To develop the principal dimensionless parameters of fee convection, we consider free convection on a vertical plate, as illustrated in Fig. 9-1. For simplicity in the analysis, we assume the boundary-layer flow is steady and laminar; the viscous- energy dissipation term in the energy equation can be neglected because of the small low velocities associated with free convection. Then the governing continuity, ‘momentum, and energy equations are obtained from the boundary-layer equations (6-81), (6-82), and (6-83), respectively, by introducing the appropriate buoyancy term into the momentum equation: ou Continuity = oy au x Momentum: + ot) pg Fy 02 ay Energy: a 3) Here the term ~ pg on the right-hand side of the momentum equation represents the body force exerted on the fluid element in the negative x direction. For small temperature differences, the density p in the buoyancy term is considered 0 vary with temperature whereas the density appearing elsewhere in these equations is considered constant,nae convecrio 419 To determine the pressure gradient term dP/ax, the x momentum equation (9-2) is evaluated at the edge of the velocity boundary layer, where u 0 and P— Pa We obtain Fala oe where pp denotes the fuid density outside the boundary layer. Then the term pg ~ dPiéx appearing in the momentum equation (9-2) becomes ca = 09-5 = Ou ~ oo os It Bis the volumerric coefficient of thermal expansion of the fluid, the change of density with temperature is related to f by _ (ae p\oT}, By expressing the derivative term in this relation with finite difference, Eq. (9-6) is approximated by 8 0-6) dp= boat or on ~BoTe 7) ‘Then Eq. (9-5) becomes: = 09 F= ~6 te Ne os) Equation (9-8) is substituted into the momentum equation (9-2). We summarize the resulting equations for free convection on a vertical plate: eu, a nto (9) Qu, au eu te = oT — 7.) +55 0-10) aT _ er ae on If the fluid is considered to be an ideal gas, we have P o= ar 0-12)420 wear TRANSHER ‘Then the coefficient of expansion f in Eq, (9-7) becomes (alo) = 1 _ (TIT) 1 T-T T-T “T, 8 0-13) For liquids, the value of f can be obtained from the property tables in App. B-2 ‘To determine the dimensionless parameters that govern heat transfer in free convection, we need to nondimensionalize the above governing equations. The following dimensionless parameters are defined x y xei y= u= » tests U T= Te o-14) Here L is a characteristic length, Us isa reference velocity, Tis the wall surface temperature, and T,, is the fuid temperature at a far distance from the hot plate. ‘When these new variables are introduced into Eqs. (9-9) to (9-11), the resulting nondimensional equations become au wv x tay 15) you | au unt 016) mya udev on Here the Reynolds and Prandtl numbers are defined as Pre! ons) ‘The dimensionless group in the momentum equation can be rearranged as GKTy = Te)L_ gBL(Ty = To)? _ Gr 019) UU Re where the Grashof number Gr is defined as or R= ea ‘The Grashof number represents the ratio of the buoyancy force to the viscous force acting on the uid. We recall that in forced convection, the Reynolds number represents the ratio of the inertial to viscous forces acting on the fluid. Therefore, the Grashof number in free convection plays the same role as the Reynolds number in forced convection. For example, in forced convection the transition from laminar to turbulent flow is governed by the critical value of the Reynolds number.‘mee convection 421 Similarly, in free convection, the transition from laminar to turbulent flow is ‘governed by the critical value of the Grashof number, Equations (9-15) to (9-17) imply that when the effects of free and forced convection are of comparable magnitude, the Nusselt number depends on Re, Pr, and Gr: Nu = f(Re, Gr, Pr) 21) ‘The parameter Gr/Re?, defined by Eq, (9-19), isa measure ofthe relative importance of free convection in relation to forced convection, When Gr/Re? = 1, free and forced convection are of the same order of magnitude, hence both must be con- sidered. If (Gr/Re?) ¢ |. flow is primarily by forced convection. If Gr/Re?) > 1. free ‘convection becomes dominant and the Nusselt number depends on Gr and Pr only: Nu = (Gr, Pr) (9-22) In free convection flow velocities are produced by the buoyancy forces only, hhence there are no externally induced flow velocities. As a result, the Nusselt ‘number does not depend on the Reynolds number. For gases, Pr & 1; hence the Nusselt number for free convection is a function of the Grashof number only: Nu= (Gr) for gases 0-23) Sometimes another dimensionless parameter, called the Rayleigh number (Ra), which is defined as Ra = Gr Pr = BLTa = Te) py, _ BL" Te ~ Te) ee (9-24) is used instead of the Grashof number to correlate heat transfer in free convection, For such cases, Eq. (9-22) takes the form Nu = (Ra, Pr) (9-25) Example 9-1 Examine the dimension of each term in the momentum equation (92). SoLUTION We write Eq. (9-2) and show below the dimension of each term: Lae eu pox ‘yt N mm m eae See 7% =422 ear mRANSRER Here, ‘Thus, each term in the momentum equation has the dimension of newtons per kilogram. 9.2 AN APPROXIMATE ANALYSIS OF LAMINAR FREE CONVECTION ON A VERTICAL PLATE ‘Heat transfer by free convection on a vertical or an inclined plate has been the subject of numerous investigations [1-26]. To provide better insight into heat transfer by free convection, we consider here the simplest situation involving a vertical plate under isothermal conditions (i.e, subjected to uniform surface temperature) placed in a large body of fluid at rest, as illustrated in Fig. 9-1 or 9-3, Let T, and T., be, respectively, the temperature ofthe wal surface and the bulk temperature of the fluid. The fluid moves upward along the plate for T, > T,, Figure 9.3 Temperature and velocity profiles for feevia convEcTION 423 and flows downward for T, < Tas illustrated in Fig. 9-1. The following analysis is applicable for both cases. ‘Mathematical Formulation of Problem ‘The buoyancy-induced flow problem, as illustrated in Fig. 9-1, is considered a boundary-layer type of flow, and the governing equations of motion and energy are obtained from Eqs. (9-9) to (9-11); 6-29, 27), (0.28) The physical boundary conditions for the problem include zero velocity com- ponents at the wall surface, no axial velocity outside the boundary layer, tempera- {ures T, atthe wall and T,, outside the boundary layer. These boundary conditions are stated as follows: 0 e=0 Tat, aty=O(wally 0 T=T7, as y+ o (or outside boundary layer) (9-29) ‘The heat transfer problem posed here yields itselto exact solution, but the analysis. is more involved. Here we describe a simple, approximate method of analysis by using the integral method, considered in Chap. 8 forthe solution of heat transfer in laminar forced convection over a flat plate. For simplicity, we assume the Prandtl number for the fuid is close to unity, hhence the thicknesses of the velocity and the thermal boundary layers are almost equal or 6 = 6,. The basic steps in the analysis are described. ‘The Momentum and Energy Integral Equations ‘The first step in the analysis by the integral method is the development of the mo- ‘mentum and the energy iptegral equations. The momentum integral equation is developed by integrating the momentum equation (9-27) over the velocity bound- ary-layer thickness 6, utilizing the continuity equation to eliminate the velocity component , and making use of the boundary conditions for the velocity com- ponent 1. If we omit the details of such a development, the resulting momentum integral equation can be expressed in the form a(Lro)- 5 ito [rtd (20)424 wear TRANSHER ‘The eneray integral equation is developed by integrating the energy equation (9-28) over the boundary layer 5, = 6, uilzing the continuity equation to eliminate the velocity component v, and employing the boundary conditions imposed on the temperature, Omitting the details of this development, we find the energy integral equation to be 4 ar [fier- m4] - -«F {As expected, Eqs. (9-30) and (9-31) are coupled because u and T appear in both; hence they should be solved simultaneously. ‘The next step in the analysis isthe determination of suitable profiles for the istribution of velocity and temperature in the boundary layers. 31) Choice of Velocity and Temperature Profiles A suitable choice should be made for the velocity and temperature profiles, con- sistent with the physical reality. For free convection on a vertical plate, say for Tz > Tz, the physical nature of the problem gives rise to the velocity and tempera- ture profiles as illustrated in Fig. 9-3, Polynomial representations can be used to approximate these profiles. ‘The temperature profile is represented by a second-degree polynomial in the form T(x, 9) = a + Boy + en) 032) ‘The following three conditions can be applied to determine the three coefficients 4g, bo, and ¢y in terms of the boundary-layer thickness: T=T, aty=0 033) T=T. aty=d om) ar x aty=6 935) “The resulting temperature profile becomes Te, y)- Te _ (,_ 9) . 5 () 1 (0:36) Here we assume 3 = 3, ‘The determination of a suitable velocity profile, however, is more involved. An ‘examination of the velocity profile in Fig. 9-3 implies that the velocity component ‘ux, y) is zero both at the wall surface and at the edge of the velocity boundary layer, but has a peak inside the boundary layer. To represent such a situation, we ‘choose a cubic polynomial in the form ux, 9) = uBlay + buy + ey? + day?) (037)REE CONVECTION 425, where the coefficients a;, by, cy, dy and a reference velocity ng are considered to be functions of x and are yet to be determined. Four conditions are needed on the velocity to determine the four coefficients. Three of these conditions are taken as u=0 aty=0 (0-38) u=0 at 0-39) Booty 9-40) ‘where d is the edge of the velocity boundary layer. A fourth condition is obtained bby evaluating the momentum equation (9-27) at y = O and noting that u = v = 0 and T = T, at y = 0. We find S--fa-w oa With the application of the condit profile, Eq, (9-37), becomes an yge BOT — Ted] ¥ (, _ 9)? so ~ [aA ()-2) oan ns given by Eqs. (9-38) to (9-41), the velocity which can be writen more compat as wa sd=ne5(t 3) (0-43) als where uy = uo(x) = wf 5%Q(T, ~ T)/(4v) is an arbitrary function of x with the dimension of velocity. By differentiating Eq. (9-43) with respect to y it can be shown that the maxi ‘mum velocity u(x, y) occurs at a distance y = 5/3. Solution of Equations ‘The temperature profile equation (9-36) and the velocity profile equation (9-43) are introduced into the momentum integral equation (9-30) and the energy integral {equation (9-31), and the indicated operations are performed. The momentum and the energy integral equations, respectively, become 1 diay! 1 Yo Fos gx 39) = 5 Te ~ Tas — "2 (9-44) 1 a Te-To 39 T Fy iad) = 2a (0-45) ‘To solve these equations, we assume that u9(x) and (x) depend on x in the form ols) = eyx" and (x) = ex" (9-464, 6)426 ear manseen where cy, ¢2,m, and n are constants, Equations (9-46) are substituted into Eqs. (9-44) and (9.45) o yield m+ miter amen 0s 390T, ~ Toca — eS xm" 9-47) (mH MC: rat % ‘To make this system of equations invariant of x, we equate the exponents of x ‘on both sides of Eqs (9-47) and (9-48), and obtain (9-48) m+n (0-49) mtn (0-50) which yields m and n=4 51) When these values of m and n arc introduced into Eqs. (9-47) and (9-48), the x variable cancels. A simultaneous solution of the resulting equations gives cy and c; a8 20 2)-"FoBhT ~ To)" a osin(t) "A= oa ome PTY” om ‘These values of n and c, are introduced into Eq. (9-468); the boundary-layer thickness 5(x) becomes Bx) = 3930952 + poe: 7 “pews 09.54) « 4) sospmoon+ miter ——@s9 where the local Grashof number Gr is defined as PTs (9-56) Given &(), the temperature profile T(x, ) can be determined from Eq. (9-36), and the local Nusselt number can be obtained as described below.‘us CONVECTION 427 ‘Local Nusselt Number The local Nusselt number Nu, is related to the temperature gradient of the fluid at the wall surface by x -HOT/OVy a0 x be =Tek~ y= Tek Here the temperature gradient i determined from Eq. (9-36): or 2, = Te) a7 cele oe (0-58) Introducing Eq, (9-58) into Eq. (9-57), the local Nusselt number is related to the boundary-layer thickness. Nu, (9-59) From Eqs. (9-55) and (9-59), the local Nusselt number for laminar free convection ‘ona vertical wall subjected to uniform surface temperature is gx Nu, = "6 = 0.508 Pr!7(0952 + Pry Gri! 0-60) This result can be expressed in the alternative form as ya(_Pr_\ Wo, = 038 (592) a ‘where the local Grashof and Rayleigh numbers, respectively, are defined as ATs = Ta? Gr, (0-62) Ty = Tx? Ra, (0-63) = Gr,Pr ‘The results developed above for T, > also apply for Tj, > Ty. For the latter ‘case, the temperature difference should be replaced by T., ~ T,. ‘Note that functional forms of Eqs. (9-60) and (9-61) are, respectively, Nu, ‘{(Gr,, Pr) and Nu, = f(Ra,, Pr). These forms are consistent with those given by Eqs. (9-22) and (9-25), which were developed from the examination of the di- mensionless form of the governing equations. ‘Mean Nusselt Number In engineering applications, the mean Nusselt number Nu, over a distance x = 0 to x = L along the plate is also of interest. To determine Nu,., We consider the average heat transfer coefficient hq, defined as fe ou428 ear TRANSEER ‘According to Eq, (9-61), h, is proportional to x"; then h becomes Im = hd (9-640) “The average Nusselt number Nu, is Ly La 4 [xy Nag 9 mf jeden sL.. (9-640) Note that the mean Nusselt number given by Eq. (9-64c) is developed for the case when h, is proportional 10 x~", ‘Suppose the local heat transfer coefficient h, depends on x in the form hoot (9-654) ‘Then the mean heat transfer coefficients hy over x = Oto x = L becomes beef fine tha ok” e (0-656) T ‘and the average Nusselt number becomes 1 Now [Na] (0-650) Clearly, for n = 4 Eq, (9-65c) reduces to Eq, (9-64c). The relation given by Eq. (©-64c) can be used now to determine the corresponding average value of the local Nusselt number given by Eq, (9-61) Introducing Eq. (9-61) into Eq, (9-64c), we obtain wae 67755 ot wag = tet oe ont nag = Or = WAT = Tal: oan ‘The results for Nu, and Nu,, given by Eqs. (9-61) and (9-66) are applicable for laminar free convection on a vertical plate maintained at a uniform temperature The fluid properties are evaluated at the film temperature T, = (I+ Ta)/2. Under the present circumstances, the transition from laminar to turbulent flow has been observed [4, 15, 16] to take place in the range 10" < Ra, < 10° Therefore, the above laminar-flow solution is restricted to the range Ray < 10°.Table 9-1 Exact solutions ‘of the mean Nusselt umber for laminar free convection ‘on a vertical plate* ed Gr 003 oe ‘008 az oot 020, nz 280 03 sn 07 osie ' oss 0588 en 0653, 665 asm + Vali for Gr, Pr < 10% ‘Source: Slicing (141 Results of Exact Analysis The previous analysis, even though approximate, helps to envision the implications of heat transfer in free convection and the physical significance of various di- _mensionless parameters. The exact analysis of free convection on a vertical plate subject to uniform wall temperature has been performed over a wide range of PrandtI numbers by various investigators. Schlichting [14] compiled the mean Nusselt number obtained by several investigators [1, 3,6, 13]. We present in Table 9-1 the resulting expressions. for the mean Nusselt number; in this table, Nu,, and Gr, are defined as follows: Ia oH = ToL! ea = Gn and they are applicable for Gr, Pr < 10°, Example 9.2 A square plate 0.4 by 0.4 m, maintained at a uniform tempera- ture of T,, = 400 K, is suspended vertically in quiescent atmospheric air at Te = 300K. (@) Determine the boundary-layer thickness 6(x) at the trailing edge of the plate (at x = 0.4 m). (b) Calculate the average heat transfer coefficient h over the entire length of the plate by using the results ofthe exact analysis listed in Table 9-1,40 ear TRANS ‘SOLUTION The physical properties of air at atmospheric pressure and T; (300 + 400)/2 = 350 K are = 2075 x 10° m/s Pr = 0.697 1 k = 003 Wim -°C) B 2.86 x 107? K-! ‘The Grashof number at L = 04 m becomes GT = Ta)? Gros = __ 9.89286 x 107? 400 ~ 300,04)* - (20.76 10°? = 4.16 x 10 ‘Thus the Boundary layer is laminar, and the approximate analysis is applicable. (a) The boundary-layer thickness d(x) is determined from Eq, (9-55): Sipe = 393 Pr"(0952 + Pry! Gri! L 3 = aa 10994 0605" =15 «107m @ Sem (b) The average Nusselt number is determined from Table 9-1 ago aster Po = 507 Whim? °C) 9-3 CORRELATIONS OF FREE CONVECTION ON A VERTICAL PLATE. The analysis of free convection is complicated, and experimental data are often desirable in order to develop reliable heat transfer relations. Here we present some of the recommended empirical correlations of free convection on a vertical wall in laminar and turbulent flow, for both uniform wall temperature (ie, isothermal surface) and the uniform wall heat lux boundary conditionsues CONVECTION 431 Table 9-2 Constant ¢ and the exponent m of Eq. (9-69) Type ofow Range of G References) Laminar ——10¢10 10" oo Turbulent 10°10 10" oo 1 8 Uniform Wall Temperature McAdams [8] correlated the average Nusselt number with an expression in the form Nu, In = Gr, Prf = € Raf, 9-69) where L isthe height ofthe vertical plate and Gr, and Nu, are defined by Eq. (9-68). ‘The recommended values ofthe constant ¢ and the exponent n are listed in Table 9-2. The physical propertics are evaluated at T,= Hy + Te) An examination of the values ofthe exponent» given inthis table reveals that in turbulent flow hy i independent of the plate height L since Gr, ~ Land hy ~ LNG)" More recently, Churchill and Chu [25] proposed two equations for correlat- ing free convection on a vertical plate under isothermal surface conditions. One expression, which applies to only laminar flow and holds forall values of the Prandtl number, is given by nes [smears ear so 5.45 W(m? -°C) k 0.0: [2s = VF Note that the results obtained from Eqs. (9-69) and (9-70) are close. The result obtained from the purely theoretical solution in Example 9-2, h 5.07 W/m? - °C), underestimates the heat transfer coefficient, Example 9-4 A vertical plate L = 5m high and w = 1.5m wide has one of its surfaces insulated; the other surface, maintained at a uniform temperature T, = 400 K, isexposed to quiescent atmospheric air at T,, = 300 K.Calculate the total rate of heat loss from the plate, SoLuTion ‘The physical properties ofatmosphericairat T; = (300 + 4002 = 350 K are v= 2075 10° mis Pr = 0697 k= 003 Whim") Bm t= 286 x 10" K-* % The Grashof number for L = $mis = WT, = TedL* 8.137 x 10"are convECTION 433 Since the Grashof number (hence the Rayleigh number) is beyond the range (of Eq, (9-70), we use Eq. (9-71) to calculate Nu, or the average heat transfer coefficient h 038%Gr, PM? T+ eazy | ' oEfons = 9B fons « 0.387(8.137 x 10!" x 0697)! y 3 TE + (0492)0.607° 527 = 5.51 Wim? °C) ‘The total heat transfer rate becomes Q=hAT, ~ Te) = (SINS x 1.54400 — 300) = 4133 kW ‘Uniform Wall Heat Flux Free convection on a vertical plate subject to uniform heat flux at the wall surface has been investigated by Sparrow and Gregg [11], Vliet and Liu [19], and Vliet [20]. Based on the experimental data of Ref. 20 for air and water and of Ref. 19 for water, the following correlations are proposed for the local Nusselt number ‘under uniform wall heat flux: Q64Gr¢ Pr)" for 10 < Grt Pr < (aminar) | (9-72) Nu, = 0568(Gr? Pr)? for? x 10") < Grf Pr < 10 (turbulent) 0-7) where the modified Grashofmumber Gr is defined as Grt = Gr, Nu, = AE Teas (0-740) and the local Nusselt number as Nu, = (0-740) ‘And qj. is the constant wall heat flux.484 ear TRANS ‘To determine the average Nusselt number Nuq, we need to establish the dependence of hon x in the form given by Eq, (65a) From Es. (9-72) and (9-73) wwe have, respectively, re ~LGrop? ~ Leet? ~ 0-146) ae Leone Leta (0-14@) ‘Then Eq, (9-65¢) is used to determine the average Nusselt number. Therefore, the average Nusselt numbers for Eqs. (9-72) and (9-73) are, respectively, gg lNtde-t = L2S(Nudean for 10 < Grf Pr < 10" (9-148) Now L136(NUe-1 for 2 x 10" < Grf Pr < 10° (@-747) 3 (Nw 1-00 All physical properties are evaluated atthe film temperature. ‘When the wall heat flux q, is prescribed instead of the wall temperature T,,, the temperature diflerence T, ~ T,,, and hence the film temperature, is unknown at the onset ofthe problem. in such situations, an initial guess is made for the film temperature, and the calculations are performed. Ifthe guess and the calculated ‘values are different enough to affect the results, the calculations are repeated with the new value ofthe film temperature. In Refs. 19 and 20 the transition from laminar to turbulent regime begins in the range3 x 10'? < Gro Pr <4 x 10! andendsin the range 2 x 10" < Gr¢ Pr < 10". Therefore, the fully zrbulent regime is considered to occur at Gr* Pr = 10", but this value may be as low as 2 x 10" Churchill and Chu [25] also correlated some of the available experimental data under uniform wall heat flux conditions with Eq. (9-70), which was developed strictly for the uniform wall surface temperature. The two problems, because of the boundary conditions (ie. one is uniform heat flux, the other is uniform wall temperature), are different from a mathematical point of view. Therefore, rigorously speaking, the heat transfer results are not expected to be the same. However, they found that Eq. (9-70) correlated the data well; hence they concluded that Eq (9-70) also can be used to correlate free convection on a vertical wall subjected to ‘uniform wall heat flux. Similarly, they propose Eq. (9-71) for use with the uniform wall heat flux condition, IE. (9-70) isto be used for uniform wall heat flux condition, it is desirable to express the right-hand side in terms of the modified Grashof number Gr*. This ccan be done by noting that Ra,=GrPr and Gr Gry NyFier CONVECTION 435 Then Ray = Nig Introducing this result for Ray on the right-hand side of Eq. (9-70) and rearranging, wwe obtain 068) 0-75) T+ @492/P)” for laminar free convection on a vertical plate subjected to uniform wall heat flux. Example 9-5 A thin vertical panel [= 3m high and w = 1.5m wide is thermally insulated on one side and exposed to a solar radiation flux of 44, = 750 W/m: on the other side. The exposed surface has an absorptivity of a, = 08 for solar radiation, Assuming that the energy absorbed by the plate is dissipated by free convection into the surrounding quiescent air at atmospheric pressure and T, = 300 K,, calculate the surface temperature of the panel SOLUTION The problem is one of constant wall heat flux, hence the surface temperature is not known and T; cannot be determined to evaluate the physical properties of air. To determine a film temperature to evaluate properties, we choose an approximate value for h in free convection, namely, = 6 Wim? °C). The solar energy absorbed by the wall is, Gu = 244, = 08 % 750 = 600 W/m? ‘Then a first approximation to the wall temperature T,, is MT, Te) = 46 20 400K Tat +4e= M0 6 = 1400 + 300) = 350 K ‘The physical properties of atmospheric air at T; = 350 K are taken as Hence, T= MT, + To 076 x 107° m/s. k = 003 Wi(m-°C) Pr = 0697 286 x 107K" ‘Then the modified Grashof number at x = 3 m is determined by Eq, (9-74a) (9.82.86 x 10°°600(34) ky? (0.03X2076 x 10°5)? = 105x104 atx=3m Grt =486 ear TRANsreR ‘Therefore, Eq. (9-73) can be used to determined the local h, at x = 3 m: A oseoxces 989 (058x105 x 10% x 0697992 538 Wim? °C) atx = 3m ‘The average heat transfer coefficient over 0 < x < 3m is determined from Eq, (9-744): h 1136{hIa-9m@ = 725 Wil? °C) Given h, the wall surface temperature T, is 600 3828 K a T= T+ 1 = 300 fe Te FL 0 + 735 ‘To improve the result, anew film temperature 7; is calculated T, = 4300 + 3828) = 341K ‘Then these calculations are repeated by evaluating the physical properties of air at this new film temperature. ‘However, another iteration is not warranted, since the film temperature T;, = 380K used in the above calculations is sufficiently close to 7, = 341 K. and the resulting change in the physical properties will be negligible. 9.4 FREE CONVECTION ON A HORIZONTAL PLATE ‘The average Nusselt number for free convection on a horizontal plate depends on whether the surface is facing up or down and whether the plate surface is warmer or cooler than the surrounding fluid. Again we consider the cases for uniform wall temperature and uniform wall heat flux separately. Uniform Wall Temperature ‘The mean Nusselt number for free convection on a horizontal plate is correlated by McAdams [8] with an expression in the form st, x (Tw ToL? Gr, = AE (0-766) whereREE convEcTION 437 ‘Table 9-3 Constant ¢ and exponent n of Eq. (9-76a) for free convection on a hori- ‘zontal plate at uniform temperature Orientation of pate RangeofGr,Pr em Flowregine Hot surface facing up or cold surface eel Cue facing dowa 2xWwIx 10” OM Ff Hot surface facing down or cold sorta facie up 3x1 193x 10 027 Source: MEAdam: [8], ‘Thecoefficient cand the exponent nare listed in Table 93. The characteristic length of the plate can be taken as the length ofa side fora square, the arithmetic mean of the two dimensions for a rectangular surface, and O9D for a circular disk of diameter D, as suggested by McAdams [8]. We note that for turbulent flow Nui, isindependent of the characteristic length. For the case ofa hot surface facing down for a cold surface facing up, the turbulent flow regime is not reached even at Gr, Pr 3 x 10". Recent correlations [24, 27] suggest that improved accuracy may be obtained ifthe characteristic length L for the plate is defined as. 4 Lad i) where A js the surface area ofthe plate and P isthe perimeter, which encompasses the area.’ Equation (9-77) is similar to the definition of the equivalent diameter except for a factor of 4 ‘Uniform Wall Heat Flux Free convection under uniform wall heat flux has been studied extensively in Ref. 23for an electrically heated plat in vertical, horizontal, and inclined positions. We present here only the results for the horizontal plate. ie For the horizontal plate with the heated surface facing upward: ex *#* Nip = 019Gr, PA"? for Gry Pr < 2 x 10! 0-78) Nu, = 014Gr, Pr) for Sx 10° < Gr, Pr< 10" | 79) Here, the § power suggests that the heat transfer coefficient does not change locally or depend on the plate length. These two results have been developed by using 30- and S-cm-long heated plates, respectively. Equation (9-79) developed with the smaller plate shows higher heat transfer than Eq, (9-78) developed with the larger plate. It appears that the edge effect becomes negligibly small with the larger plate.498 sear TRANseER For the horizontal plate with the heated surface facing downward: [» = 058(Gr, Pr)" for 10° < Gr, Pr < 10" | @-80) Here the # power suggests that the flow regime remains laminar. The physical properties in Eqs. (9-78) to (9-80) are to be evaluated at a mean temperature, defined as h. ~ 0.25(T, — Tx) 81) and the thermal expansion coefficient f at (T,, + T,)/2. In these expressions, the Grashof number is defined as 0. Teal? oom and the mean Nusselt number over the length Las ul wm (9-82) Example 9.6 Consider a square plate 0.5m by 0.5m with one surface in- sulated and the other surface maintained at a uniform temperature of 7, 385 K whichis placed in quiescent air at atmospheric pressure and T, = 315 K. Calculate the average heat transfer coefficient for free convection for the follow= ing three orientations ofthe hot surface: (@) The plate is horizontal, and the hot surface faces up. (6) The plate is vertical. (©) The plate is horizontal, and the hot surface faces down, Soution The physical properties of atmospheric air at T; = 4085 + 315) = 350 K are taken as y= 2076 x 107? mis Pr = 0.697 k= 003 Wim °C) = he 2a 10K i ‘Then the Grashof number for L = 0.5 m becomes ra)? _ 982.86 x 10-3985 — 3150)? . (2076 x 10°F = 57 « 10Ise CONVECTION 439) (q) For the horizontal plate with the hot surface up, the average Nusselt number is determined from Eq, (9-T6a) and Table 9-3. For the turbulent flow condition, we obtain hale Nu, = "2 = 01461, Pr! e Sr, Pr)? hy = 5 (0.14YGr,, Pr) 0.03 " a = G5 (OAS x 10° x 0.697) = 618 Wim? °C) (6) For the vertical plate; the average Nusselt number is determined from Eq, (9-69) and Table 9-2. For the laminar low condition we have Nog = Meh = 0.596, Pr 0.03 is he hg = ig (0.59N5.7 x 10° x 0.697) = 50 Wim? -°C) (©) For the horizontal plate with the hot surface facing down, the average Nusselt number is determined from Eq, (9-76a) and Table 9-3 Nu, = Meh = 021Gr. Prt t= oars 108 0099" = 220 Wi(m? °C) ‘These results show that of the three orientations considered with a square plate, the highest and lowest values of the free-convection heat transfer Coefficient occur with the horizontal position, the hot surface facing up- ward and downward, respectively. 98 FREE CONVECTION ON AN INCLINED PLATE Rich [7] pointed out that the heat transfer coefficient for free convection on an inclined plate can be predicted by the vertical plate formulas ifthe gravitational term in the Grashof number is adjusted to accommodate the effect of the inclina- tion. Since then, free convection on inclined surfaces has been studied by several investigators [20, 23, 25, 26]. The orientation of the inclined surface, whether the40 ear TRANsrER < Ee Mn. Figure 94 The concept of poitive and negative intnation angles fom the vertical 10 define the onenation ofthe hc surface surface is facing upward or downward, is also a factor that affects the Nusselt ‘number. To make a distinction in the orientation of the surface, by following Fuji and Imura [23], we designate the sign of the angle @ that the surface makes with the vertical as follows: 1. The angle 8 is considered negative if the hot surface is facing up, as illustrated in Fig. 9-40 2. The angle @ is considered positive if the hot surface i facing down, a illustrated in Fig. 9-4. Figure 9-46 illustrates the limiting cases of @ + ~90°, the horizontal plate with hot surface facing upward, and @—» +90", the horizontal plate with hot surface facing downward, Uniform Wall Heat Flux Here we present the heat transfer correlations based on the extensive experimental investigations of Fujii and Imura [23] for free convection from an inclined plate subjected to approximately uniform wall ux to water. For an inclined plate with the heated surface facing downward: Nu, = 0.56(Gr, Pr cos 6)!" for +0 < 88, 10° < Gr, Pr < 10"! 83) For the plate slightly inclined with the horizontal (that is, 88° <9 < 90°) and the hheated surface facing downward, Eq. (9-80) is applicable. The power in Eq. (9-83) implies that the flow is always in the laminar regime. For the inclined plate with the heated surface facing upward, the heat transfer correlation has been developed withthe following considerations [23} Itisassumed that Eq. (9-83) is applicable in the laminar flow regime of Gr, Pr < Gr, Pr, ‘where Gr, isthe critical Grashof number at which the transition from laminar to turbulent flow takes place. Inthe turbulent regime, itis assumed that Eq, (9-78) or (0-79) is applicable if Gr, Pr is replaced by Gr, Pr cos 0. With this consideration,vax convection 444 ‘Table 9-4 Transition Grashot umber (see Fig. 9-4a for the definition of degree Gr, cs Sx 10° = Wo “a to os 1 From the experimental data of Fuji sand Imura (23), two expressions can be developed, one involving the coefficient 0.13 based on the results of experiments with the 30-cm test plate and the other involving the co- ‘efficient 0.16 based on a S-cm test plate, Here we present the average of these two results and give the correlation of free convection on an inclined plate with the heated surface facing upward as o14s(Gr, Pr! Gr, Pry] + 0.56(Gr, Pr cos '* | (9-84) for Gr, Pr-< 10", Gr, > Gr,, and —15° < 0 < —75°. Here, the value of the transition Grashof number Gr, depends on the angle of inclination 6, as listed in Table 9-4 In Eqs. (9-83) and (9-84), all physical properties are evaluated at the mean temperature Ty = T, — 0.25(T, ~ Ts) and f is evaluated at T, + 0.25(T, ~ T.)- Example 9-7 A square plate } m by 3 mis thermally insulated on one side and subjected to a solar radiation flux g = 600 W/m? on the other side which is considered a black surface. The plate makes an angle # = ~60° with the vertical as illustrated in Fig. 9-4a, so that the hot surface is facing upward. The heated surface dissipates heat by free convection into quiescent atmos- pheric air at 7, = 300 K, Calculate the equilibrium temperature of the plate SOLUTION This is a problem of constant wall heat flux. Therefore, the surface temperature is not known, and the mean temperature at which physical properties are to be evaluated cannot be determined, To start the calculations, ‘we choose an approximate value for h. From our previous experience we take = 6 Wim? ="C),Then a Bist approximation othe wal temperature Tis WT, — To) = 9 T= T.+9= 3004+ © - wo h 6402 ext Tease We plan to use Eq, (9-84) for which the physical properties are evaluated at a ‘mean temperature Ty = Ty ~ 028T, - Te) = 400 — 0.25(400 — 300) = 375K and fis evaluated at To + 0.25(T, — T,) = 300 + 0.25(400 ~ 300) = 325K ‘Then the physical properties of air are taken as v= 233 x 10-% m/s k = 0032 Wim -°C) Pr = 0693 (B= shy = 3.08 x 10> ‘The transition Grashof number for @ = —60°is obtained from Table 9-4 as Gr, = 108 ‘The Grashof number Gr, is BT. — __ 9.8X3.08 x 10°? 400 — 30030.5)* : (33 x 1073) = 695 « 10" Equation (9-84) is piven as Gn 145{(Gry Pr)" — (Gr, Pr)!?] + 0.56(Gr, Pr cos 0)!" SOT (Q145)[(695 x 10 x 0693)" — (LOF x 0693)*) 0.032 05 = 621 W/lm? °C) Given h, the wall surface temperature T, can be determined from + OZ (a.56(10" x 0.693 x 0.5)""" Ge 600 Ty +f = 300 + E = 3966K ‘Another iteration is not warranted, because the mean temperature used in the above calculations is suficently close to the mean temperature obtained by taking 7, = 396.6 K.ius convection 443 9-6 FREE CONVECTION ON A LONG CYLINDER ‘We now examine free convection on along cylinder for both vertical and horizontal cylinders. Vertical Cylinder The average Nusselt number for free convection on a vertical eylinder is the same 4s that for a vertical plate ifthe thickness of the thermal boundary layer is much smaller than the cylinder radius, namely, ifthe curvature effects are negligible. ‘Therefore, for an isothermal vertical cylinder, the average Nusselt number can be found from vertical plate relation, Eq (9-69), with the coefficients ¢ and m given in ‘Table 9-2. For such cases, the length Lin the definition of the Grashof or the Ray- leigh number represents the height of the eylinder. For fluids having a Prandt number 0.7 and higher, a vertical cylinder may be ‘weated asa vertical at plate when Lp Gagne <8 ‘where Dis the diameter ofthe cylinder. The physical significance ofthis criterion and the errr associated with it are better envisioned by referring to Fig, 9-Sb In the case of vertical, slender, circular cylinders, the above criterion is not satisfied; hence a vertical cylinder can no longer be treated as a vertical plate. “This matter was studied by Sparrow and Gregg (29), Minkowyez and Sparrow [30], and Cebeci [31]. Figure 9-50 shows a plot ofthe ratio ofthe local Nusselt ‘number for a vertical cylinder to that for a fat plate asa function of the parameter f= 2/2Gr!"/R) for several different values of the Prandtl number. Here R is the Tadius’ of the cylinder, Nu, = hx/k is the local number, and Gr, = ‘GACT, ~ T.)x*/»* isthe local Grashof number. Similarly, Fig 9-5b shows a plot ofthe ratio ofthe average Nusselt number fora vertical eyinder to that fora vertical plate, Note thatthe deviation increases when the Grashof number or the Prandtl hnumber decreases, For the case of a vertical eylinder subjected to uniform wall heat fux, the local Nusselt number may be determined from Eqs. 9-72) and (9-73. Horizontal Cylinder ‘Churchill and Chu [32] correlated the existing data for the average Nusselt number for free convection on an isothermal horizontal cylinder with an equation of the ssame form as Eq. (9-71) and proposed the following correlation: Nuy? = 0387 Ra! Tete 60+ ees (Or 10+ < Ray < 10" | (085)