COMPUTATION OF HEAT TRANSFER AUGMENTATION IN A PLATE-FIN HEAT EXCHANGER USING RECTANGULAR / DELTA WING

A THESIS SUBMITTED IN FULFILLMENT OF THE REQUIREMENTS FOR THE AWARD OF THE DEGREE OF

DOCTOR OF PHILOSOPHY
IN MECHANICAL ENGINEERING

BY

GULSHAN SACHDEVA 2K06-NITK-PhD-1099-M

UNDER THE SUPERVISION OF

PROF. K.S. KASANA

DEPARTMENT OF MECHANICAL ENGINEERING NATIONAL INSTITUTE OF TECHNOLOGY (Institution of National Importance) KURUKSHETRA-136119, INDIA March, 2010

Candidates Declaration
I hereby certify that the work which is being presented in the thesis, entitled COMPUTATION OF HEAT TRANSFER AUGMENTATION IN A PLATE-FIN HEAT EXCHANGER USING RECTANGULAR / DELTA WING for the award of the degree of Doctor of Philosophy submitted in the Department of Mechanical Engineering of National Institute of Technology, Kurukshetra, is an authentic record of my own work carried out under the supervision of Dr. K. S. Kasana, Professor, Department of Mechanical Engineering, National Institute of Technology, Kurukshetra, India. The matter presented in this thesis has not been submitted by me for the award of any degree/diploma of this or any other University/Institute.

Date:

Gulshan Sachdeva

ii

Certificate
This is to certify that the thesis entitled COMPUTATION OF HEAT TRANSFER AUGMENTATION IN A PLATE-FIN HEAT EXCHANGER USING RECTANGULAR / DELTA WING being submitted by Gulshan Sachdeva (Registration number 2K06-NITK-PhD-1099-M) to the Department of Mechanical Engineering of National Institute of Technology, Kurukshetra for the award of the Degree of Doctor of Philosophy in Mechanical Engineering, is a bona fide research work carried out by him under my supervision and guidance. His thesis has reached the standard of fulfilling the requirements of regulations relating to degree. The thesis is an original piece of research work and embodies the findings made by the research scholar himself. The results presented have not been submitted in part or in full to any other University/Institute for the award of any degree or diploma.

Dr. K. S. Kasana Professor, Mechanical Engineering Department National Institute of Technology, Kurukshetra-136119, Haryana, INDIA.

iii

Acknowledgements
At the very outset, I would like to record my heartfelt gratitude to my respected teacher and thesis supervisor Professor K. S. Kasana for his invaluable guidance, cordial advice and constant encouragement throughout this research. I feel very fortunate to have this opportunity to work with him and I shall always remain obliged to his greatness in devoting a large share of his valuable time and knowledge to this work. I would also like to express my sincere gratitude to Dr. R. Vasudevan, RCAM Labs, S.M.U., Dallas, for introducing me to the field of computational fluid dynamics. The numerous discussions, I had with Dr. R. Vasudevan instilled in me the confidence needed to crack formidable problems in CFD. I wish to thank Dr. S. S. Rattan, Professor & Head, Mechanical Engineering Department, NIT Kurukshetra for the administrative support during the execution and completion of this thesis. Thanks are also due to Dr. T. K. Garg, Dr. S. K. Sharma, Dr. S. Saxena, Dr. Dinesh Khanduja and all the staff members for their kind cooperation. I would also like to thank Dr. Punit Kumar Assistant Professor, MED, NIT, Kurukshetra for his valuable advices during the completion stage. I shall be failing in my duty if I do not express thanks to my friends especially Dharmender and Avadhesh Yadav for their ungrudging help and suggestions. I would like to mention my father Sh. Krishan Lal, mother-in-law Smt. P.Lakshmi sisters Mamta, Poonam and brothers-in-law Sameer and Ajay, whos never ending support and wholehearted help were the real impetus that continuously motivated me to do my best. I wish to dedicate this thesis to my mother late Smt. Sudesh, who has always been an endless source of inspiration and joy in my life. I find myself spellbound to acknowledge thanks to my wife Anuradha and son Avin, for their silent support, patience, encouragement and affection without which this work would never have been possible.

Gulshan Sachdeva

iv

Abstract
The improvement of the performance of heat exchangers with gas as the working fluid becomes particularly important due to the high thermal resistance offered by gases in general. In order to compensate for the poor heat transfer properties of gases, the surface area density of plate heat exchangers can be increased by making use of the secondary fins such as, off-set fins, triangular fins, wavy fins, louvered fins etc. In addition, a promising technique for the enhancement of heat transfer is the use of longitudinal vortex generators. The longitudinal vortices are produced due to the pressure difference generated between the front and back surface of the vortex generator. The longitudinal vortices facilitate the exchange of fluid near the walls with the fluid in the core and hence, the boundary layer is disturbed. It causes the increase in temperature gradient at the surface which leads to the augmentation in heat transfer. An innovative design of triangular shaped secondary fins with rectangular or a delta wing vortex generator mounted on their slant surfaces for enhancing the heat transfer rate in plate-fin heat exchanger is proposed. The performance of the proposed design is evaluated for different angles of attack of the wing and constant wall temperature boundary condition. The study is being carried out both for wing vortex generator either attached to the fin surface by welding, brazing etc. or punched out by cutting the fin surface itself. The stamped or punched wings avoid the additional cost and complexities involved in welding the wings to the plate surface and hence are easy to manufacture. The flow regime is assumed to be laminar because, usually the fin spacing is so small and the mean velocity is such that the Reynolds numbers of interest are below the critical Reynolds number. The working fluid considered herein is air. The unsteady, incompressible, viscous flow is governed by the three dimensional Navier-Stokes and energy equations. The 3-D mesh generated in the computational domain is of Cartesian cells with staggered grid arrangement. Unlike the conventional grid, the nodes for the velocities are taken at the center of the cell faces to which the respective velocity vectors are normal, whereas, the pressure and temperature nodes are defined at the center of the cell itself. The staggered grid arrangement is used to avoid the problem of checkerboard distribution of pressure and velocities.

The present analysis uses a modified version of Marker and Cell method to solve the governing equations. The solution of Navier-Stokes equations gives explicitly a provisional value of the velocity components to be used for the next time step. However, these explicitly advanced velocity components may not yield a realistic flow field. Therefore, the continuity of flow is checked using these velocity components. Since a divergence free velocity field must exist for the present case of incompressible flow, a non-zero divergence of velocity indicates the accumulation of some mass and hence, an incorrect pressure field. Therefore, the pressure field is corrected such that the divergence becomes non-zero which is followed by the recalculation of velocity components. This iterative cycle continues till a divergence free velocity field is achieved and hence, the continuity equation is satisfied. After determining the velocity field, the energy equation is solved by the successive over relaxation technique to determine the temperature field. For the numerical implementation of the present problem, a computer code has been developed in Visual-FORTRAN. The validity of the code is tested by running it for a standard 2-D lid-driven square cavity. The validity of the present model is also established by computing the combined spanwise average Nusselt number for the fully developed flow in a rectangular channel without any type of vortex generator and the results are in good agreement with those of published results. The flow structure is visualized by the cross-stream velocity vectors along and downstream the wing and the results clearly depict the generation of the vortices. In the case of stamped wing, some of the fluid gets entrained through the hole beneath the wing; thereby reducing the strength of the cross stream velocity vectors. Besides this, these figures also show the decrease in the strength of the vortices in the downstream direction. The two main vortices persist for a long distance along the flow. An increase in Reynolds number and angle of attack of the wing is found to increase the strength of the cross stream velocity vectors. The thermal performance of the wing vortex generator is predicted by the bulk temperature and the combined spanwise average Nusselt number of the fluid in axial direction. The bulk temperature increases along the length of the channel but there is a steep increase in the bulk temperature along the wing location. Similar trends are obtained for higher Reynolds number however, with lower magnitudes of bulk temperature. At higher Reynolds numbers, more fluid passes through the channel in the same interval which causes a reduction in the value of mean temperature.

vi

Combined spanwise average Nusselt number is computed by averaging the local Nusselt numbers, Nu = ( y ) (b w ) all around the periphery. In the downstream direction, the temperature of the fluid increases and so the combined spanwise average Nusselt number decreases continuously for the plate-triangular fin without any vortex generator. The churning action mixes the fluid near the surface with the comparatively colder fluid in the core region. It increases the temperature gradient near the walls and hence, the combined spanwise average Nusselt number increases. The combined spanwise average Nusselt number for the rectangular wing at = 26 and Re=100 is 35.58 percent higher than that for the case of plate-triangular fin channel at X =3.18952. The heat transfer enhancement can reduce the size of the heat exchanger for a given heat load or exit temperature. To this end the results of the computation are expressed in terms of the compactness achieved by using the proposed design and about 32% reduction in length is possible by the use of delta wing vortex generator at an angle of attack of 26. The only price to be paid for enhancing heat transfer using longitudinal vortex generators is the additional pumping power required to force the fluid through the heat exchanger. This quantity is also computed to confirm that the increase is sufficiently small. The analysis is also carried out by varying the size of the wings and using the in-line arrangement of the wings.

vii

Contents
Page No. Candidates Declaration Certificate Acknowledgement Abstract Contents List of Figures List of Tables Nomenclature ii iii iv v viii xi xvii xviii

1. INTRODUCTION 1.1 1.2 Heat Exchanger Extended Surface Heat Exchangers 1.2.1 Plate-Fin Heat Exchangers 1.2.2 1.3 1.4 Tube-Fin Heat Exchangers Heat Transfer Rate in Compact Heat Exchangers Other Heat Transfer Enhancement Techniques 1.4.1 Active Heat Transfer Enhancement Techniques 1.4.2 1.5 1.6 1.7 2.1 2.2 Passive Heat Transfer Enhancement Techniques Vortex Generator Motivation for the Present Work Layout of the Thesis Introduction Enhancement of Heat Transfer Using Vortex Generators 2.2.1 2.2.2 2.2.3 2.3 2.4 Vortex Generator Enhanced Flat Plate Flows Vortex Generator Enhanced Fin-Tube Exchanger Flows Vortex Generator Enhanced Channel Flows

1-16 1 2 3 4 6 7 7 9 12 15 15 17-36 17 17 19 20 25 33 35

2. REVIEW OF LITERATURE

Numerical Methods for Solving Navier-Stokes Equations Objectives of the Present Study

viii

3. PROBLEM FORMULATION 3.1 3.2 3.3 3.4 3.5 Introduction Statement of the Problem Governing Equations Staggered Grid Arrangement and the Meshing of the Domain Boundary Conditions 3.5.1 3.5.2 3.6 3.6.1 3.7 3.8 3.9 Initial Boundary Conditions Spatial Boundary Conditions Marker and Cell Method

37-69 37 37 41 43 44 44 45 46 46 47 51 52 53 53 54 54 55 56 56 57 58 59 60 60 61 62 65 68 70-98 70 70 74 77 ix

Method of Solution 3.6.2 MAC Algorithm Solution of Energy Equation Stability Conditions Numerical Boundary Conditions 3.9.1 Boundary Conditions for Continuity Equation 3.9.1.1 Cells Adjacent to the Horizontal No-slip Boundary 3.9.1.2 Cells Adjacent to the Inclined No-slip Boundary 3.9.1.3 Cells Adjacent to the Inclined Plane of Symmetry 3.9.2 Velocity Boundary Conditions for N-S Equations 3.9.2.1 Cells Adjacent to the Horizontal No-slip Boundary 3.9.2.2 Cells Adjacent to the Inclined No-slip Boundary 3.9.2.3 Cells Adjacent to the Inclined Plane of Symmetry 3.9.3 Thermal Boundary Conditions for Energy Equation 3.9.3.1 Cells Adjacent to the Horizontal Boundary 3.9.3.2 Cells Adjacent to the Inclined No-slip Boundary 3.9.3.3 Cells Adjacent to the Inclined Plane of Symmetry 3.9.4 Boundary Conditions for the Vortex Generator

3.10 3.11 4.1 4.2 4.3 4.4

Comparison of Results Based on Model Problem Spatial Grid Independence Introduction Streamwise Velocity Vectors by Rectangular Wing Vorticity Contours along the Rectangular Wing Heat Transfer Performance

4. PERFORMANCE OF THE RECTANGULAR WING

4.5 4.6 4.7 4.8 4.9 4.10 5.1 5.2 5.3 5.4 5.5 5.6 5.7

Performance of the Built-in Rectangular Wing Performance of the Stamped Rectangular Wing Effect of Aspect Ratio on the Performance of Rectangular Wing Performance of the In-line Rectangular Wings Pressure Loss Penalty Concluding Remarks Introduction Streamwise Velocity Vectors by Delta Wing Performance of the Built-in Delta Wing Performance of the Stamped Delta Wing Performance of the In-line Delta Wings Pressure Drop Penalty Comparison of Rectangular and Delta Wings 5.7.1 Same Chord Length and Span of the Wings 5.7.2 Same Area of the Rectangular and Delta Wings

79 85 87 90 94 97 99-124 99 99 103 108 112 117 119 119 121 124 125-126 125 126 127 128-141 142-148 149-154

5. PERFORMANCE OF THE DELTA WING

5.8 6.1 6.2

Concluding Remarks Major Findings Scope for Further Work

6. CONCLUSIONS

LIST OF PUBLICATIONS REFERENCES Appendix A: Non-Dimensional Formulation of Governing Equations Appendix B: The Program Substructure

List of Figures
Figure Fig. 1.1 Fig. 1.2 Fig. 1.3 Fig. 1.4 Fig. 1.5 Fig. 1.6 Fig. 1.7 Fig. 1.8 Fig. 1.9 Fig. 1.10 Fig. 1.11 Fig. 1.12 Fig. 2.1 Fig. 2.2 Fig. 2.3 Fig. 2.4 Fig. 2.5 Fig. 2.6 Fig. 2.7 Fig. 3.1 Fig. 3.2 Fig. 3.3 Fig. 3.4 Fig. 3.5 Title Basic components of a platefin heat exchanger Plate-fin heat exchanger and its geometries Internally finned tubes (Axial and Helical fins) Externally finned tubes (a) Individually finned; (b) Continuously finned Single pass counter flow heat exchanger Wire coil insert Displaced wire coil insert Swirl flow devices (a) Helical vane insert (b) Twisted tape insert Longitudinal vortex generators Vortex systems behind a delta winglet Orientation of a winglet pair (a) Common flow-up (b) Common flowdown Actively generated longitudinal vortices Geometrical definitions of the test specimen In-line tube fin heat exchanger Flow structure around a circular tube on a plate Geometric arrangements of tube rows and vortex generators Delta wing placed in a rectangular channel Boundary layer thinning Boundary layer thickening Plate-fin heat exchanger (a) Rectangular wings and (b) Delta wings mounted on the triangular secondary fins Two dimensional view of the geometry in the direction of flow View of the channel after rotation (a) Computational domain for the stamped wing (b) Geometry of the rectangular wing View of the channel after rotation (a) Computational domain for the stamped wing (b) Geometry of the delta wing Computational domain in case of built-in wing 41 40 38 39 15 20 21 21 23 27 28 30 38 7 10 10 11 13 14 14 Page No. 3 4 5 6

xi

Fig. 3.6 Fig. 3.7 Fig. 3.8 Fig. 3.9 Fig. 3.10 Fig. 3.11 Fig. 3.12 Fig. 3.13 Fig. 3.14 Fig. 3.15 Fig. 3.16 Fig. 3.17 Fig. 3.18 Fig. 3.19 Fig. 3.20 Fig. 3.21 Fig. 3.22 Fig. 3.23 Fig. 3.24 Fig. 3.25 Fig. 4.1 Fig. 4.2 Fig.4.3(a) Fig. 4.4 Fig. 4.5

Staggered grid Meshed computational domain for built-in wing Boundary conditions on the horizontal no-slip plane for continuity equation Boundary conditions on the inclined no-slip plane for continuity equation Boundary conditions on the plane of symmetry for continuity equation Boundary conditions on horizontal no-slip plane for N-S equations Boundary conditions on the inclined no-slip plane for N-S equations Boundary conditions on the plane of symmetry for N-S equations Isothermal boundary conditions on the horizontal surface Isothermal boundary conditions on the inclined no-slip surface Isothermal boundary conditions on the plane of symmetry Side view of the wing vortex generator Velocity nodal points on the delta wing plane Velocity nodal points on the rectangular wing plane Thermal boundary conditions on the wing plane Variation of U-velocity along the vertical mid plane for the lid driven flow in a square cavity Variation of V-velocity along the horizontal mid plane for the lid driven flow in a square cavity Combined spanwise average Nusselt number in a rectangular channel Distribution of j / f in a rectangular channel Grid independence test Streamwise velocity vectors along the built-in rectangular wing Streamwise velocity vectors beyond the built-in rectangular wing Streamwise velocity vectors along the stamped rectangular wing Streamwise velocity vectors along the channel without vortex generator Vorticity contours for the built-in rectangular wing at X=3.10 (a) Angle of attack 20 (b) Angle of attack 26

43 44 54 55 56 57 58 59 60 61 62 63 64 64 65 66 67 67 68 69 71 72 73 73 74 75

Fig.4.3(b) Streamwise velocity vectors along the stamped rectangular wing

xii

Fig. 4.6 Fig. 4.7 Fig. 4.8 Fig. 4.9

Vorticity contours for the built-in rectangular wing at X = 3.39 and angle of attack 26 Vorticity contours for the built in rectangular wing at X = 4.66 and angle of attack 26 Bulk temperature distributions for various angles of attack of the built-in rectangular wing at Reynolds number 100 Combined spanwise average Nusselt number distributions for various angles of attack of the built-in rectangular wing at Reynolds number 100

76 77 79 80

Fig. 4.10 Fig. 4.11 Fig. 4.12

Bulk temperature distributions for various angles of attack of the built-in rectangular wing at Reynolds number 200 Bulk temperature comparisons for the built-in rectangular wing at Reynolds number 100 and 200 Combined spanwise average Nusselt number distributions for various angles of attack of the built-in rectangular wing at Reynolds number 200

82 83 84

Fig. 4.13 Fig. 4.14 Fig. 4.15 Fig. 4.16

Combined spanwise average Nusselt number comparisons for the built-in rectangular wing at Reynolds number 100 and 200 Bulk temperature distributions for various angles of attack of the stamped rectangular wing at Reynolds number 100 Bulk temperature comparisons of the rectangular wing with and without stamping Combined spanwise average Nusselt number distributions for various angles of attack of the stamped rectangular wing at Reynolds number 100

85 86 86 88

Fig. 4.17 Fig. 4.18

Combined spanwise average Nusselt number comparisons of the rectangular wing with and without stamping Combined spanwise average Nusselt number distribution for different aspect ratios of the built-in rectangular wing at =20 and Reynolds number 100

88 89

Fig. 4.19

Bulk temperature distribution for different aspect ratios of the built-in rectangular wing at =20 and Reynolds number 100

89

xiii

Fig. 4.20 Fig. 4.21 Fig. 4.22

Bulk temperature distributions for various angles of attack of the inline rectangular wings at Reynolds number 100 Bulk temperature comparison of the in-line and single rectangular wing at an attack angle of 26 and Reynolds number 100 Combined spanwise average Nusselt number distributions for various angles of attack of the in-line rectangular wings at Reynolds number 100

90 91 92

Fig. 4.23

Combined spanwise average Nusselt number comparison of the in-line and single rectangular wing at an attack angle of 26 and Reynolds number 100

92

Fig. 4.24 Fig. 4.25 Fig. 4.26 Fig. 4.27 Fig. 4.28 Fig. 4.29 Fig. 5.1 Fig. 5.2 Fig. 5.3 Fig. 5.4 Fig. 5.5 Fig. 5.6 Fig. 5.7 Fig. 5.8

Bulk temperature distributions for the in-line rectangular wings at Reynolds number 100 and 200 Combined spanwise average Nusselt number distributions of the in-line rectangular wings at Reynolds number 100 and 200 Pressure drop for various angles of attack of the built-in rectangular wing Pressure drop distribution of the built-in and stamped rectangular wing Pressure drop for the in-line rectangular wings at various angles of attack of the rectangular wing Pressure drop for various aspect ratios of the built-in rectangular wing at =20 and Reynolds number 100 Generation of the secondary flow Streamwise velocity vectors along the built-in delta wing Streamwise velocity vectors after the built-in delta wing Streamwise velocity vectors along the stamped delta wing Streamwise velocity vectors after the stamped delta wing Bulk temperature distributions for various angles of attack of the built-in delta wing at Reynolds number 100 Bulk temperature distributions for various angles of attack of the built-in delta wing at Reynolds number 200 Bulk temperature comparison for the built-in delta wing at Reynolds number 100 and 200

93 94 95 96 96 97 100 101 101 102 102 103 104 105

xiv

Fig. 5.9 Fig. 5.10 Fig. 5.11 Fig. 5.12 Fig. 5.13 Fig. 5.14 Fig. 5.15

Combined spanwise average Nusselt number distributions for various angles of attack of the built-in delta wing at Reynolds number 100 Combined spanwise average Nusselt number distributions for various angles of attack of the built-in delta wing at Reynolds number 200 Combined spanwise average Nusselt number comparison for the built-in delta wing at Reynolds number 100 and 200 Bulk temperature distributions for various angles of attack of the stamped delta wing at Reynolds number 100 Bulk temperature comparison of the delta wing with and without stamping Bulk temperature distributions for various angles of attack of the stamped rectangular wing at Reynolds number 200 Combined spanwise average Nusselt number distributions for various angles of attack of the stamped rectangular wing at Reynolds number 100

106 107 107 109 109 110 111

Fig. 5.16 Fig. 5.17

Combined spanwise average Nusselt number comparison of the delta wing with and without stamping Combined spanwise average Nusselt number distributions for various angles of attack of the stamped rectangular wing at Reynolds number 200

111 112

Fig. 5.18 Fig. 5.19 Fig. 5.20 Fig. 5.21 Fig. 5.22 Fig. 5.23

In-line configured delta wings Bulk temperature distributions for various angles of attack of the inline delta wings at Reynolds number 100 Bulk temperature comparison of the in-line and single rectangular wing at an attack angle of 26 and Reynolds number 100 Bulk temperature comparisons for the in-line delta wings at Reynolds number 100 and 200 Combined spanwise average Nusselt number distributions for various angles of attack of the in-line delta wings at Reynolds number 100 Combined spanwise average Nusselt number comparison of the inline and single rectangular wing at an attack angle of 26and Reynolds number 100

113 113 114 114 115 116

xv

Fig. 5.24 Fig. 5.25 Fig. 5.26 Fig. 5.27 Fig. 5.28 Fig. 5.29 Fig. 5.30 Fig. 5.31 Fig. 5.32 Fig. 5.33

Combined spanwise average Nusselt number comparison of the inline and single rectangular wing at Reynolds number100 and 200 Pressure drop variation of the delta wing with and without stamping Pressure drop variation of the in-line and single delta wing Geometry for (a) same span and chord length of the wings (b) same area of the wings Combined spanwise average Nusselt number for the delta and rectangular wing of same chord length and span Bulk temperature variations for the delta and rectangular wing of same chord length and span. Pressure drop variations for the delta and rectangular wing of same chord length and span. Combined spanwise average Nusselt number variations for the same area of rectangular and delta wings Bulk temperature variations for the same area of rectangular and delta wings Pressure drop variations for the same area of rectangular and delta wings

116 117 118 119 120 120 121 122 123 123

xvi

List of Tables
Table Table 4.1 Table 5.1 Table 5.2 Title Percentage reduction in the length of the channel using rectangular wing Percentage reduction in the length of the channel using delta wing Pressure drop in a plate-fin channel 108 118 Page No. 80

xvii

Nomenclature
A Ac b c cp Dh e fx h H k L Nu Nu sa p P surface area cross sectional area span of the vortex generator chord length of the vortex generator specific heat of the fluid hydraulic diameter internal energy of the fluid body force in x-direction convective heat transfer coefficient characteristic length dimension (distance between the plates) thermal conductivity of the fluid length of the channel local Nusselt number based on bulk temperature of the fluid, equation 4.6 combined spanwise average Nusselt number, equation 4.7 static pressure
2 non-dimensional static pressure, P = p U av

P Pr

prime pressure correction Prandtl number, Pr = c p k

heat flux internal heat generation per unit volume Reynolds number, Re = U av H area of the vortex generator time temperature average velocity of the fluid at the channel inlet mean outflow velocity from the channel overall heat transfer coefficient axial, normal and spanwise components of velocity axial, normal and spanwise components of velocity (non-dimensional) heat exchanger volume

q
Re s t T Uav Uc UHT u, v, w U, V, W V HT

xviii

x, y, z X, Y, Z

axial, normal and spanwise dimensions of coordinates axial, normal, and spanwise coordinates (non-dimensional)

Greek symbols

upwinding factor angle of attack of the vortex generator heat transfer surface area density non- dimensional temperature, = (T T ) (Tw T ) aspect ratio of the vortex generator, = b 2 s second viscosity coefficient dynamic viscosity of the fluid kinematic viscosity of the fluid density of the fluid non-dimensional time, = t (H / U av ) any of the dependent variable, U, V, W or over-relaxation factor vorticity in x-direction non-dimensional vorticity in X-direction

HT

o x X
Subscripts

av b sa w

average bulk condition spanwise combination of channel walls wall inlet condition for the temperature

xix

Chapter 1 Introduction 1.1 Heat Exchanger

A heat exchanger is a device which is used to transfer thermal energy between two or more fluids, between a solid surface and a fluid, or between solid particulates and a fluid, at different temperatures and in thermal contact. Not only are heat exchangers often used in the process, power, petroleum, air-conditioning, refrigeration, cryogenic, heat recovery, alternative fuel, and manufacturing industries, they also serve as key components of many industrial products available in the market. The heat exchangers can be classified in several ways such as, according to the transfer process, number of fluids and heat transfer mechanism. Conventional heat exchangers are classified on the basis of construction type and flow arrangement. The other criteria used for the classification of heat exchangers are the type of process functions and fluids involved (gas-gas, gas-liquid, liquid-liquid, two phase gas etc.). The classification according to the surface compactness deals with one of the important class of heat exchangers named as compact heat exchangers. A gas-to-liquid exchanger is referred to as a compact heat exchanger if it incorporates a heat transfer surface having a surface area density greater than 700 m2/m3 or hydraulic diameter Dh 6 mm for operating in a gas stream and 400 m2/m3 or higher for operating in a liquid or phase change stream. If the surface area density is greater than 3,000 m2/m3 or 100 m Dh 1 mm, the heat exchanger is referred as a meso heat exchanger and if the surface area density is greater than about 15,000 m2/m3 or 1 m Dh 100 m, then it is known as a micro heat exchanger. A typical shell and tube heat exchanger has a surface area density less than 100 m2/m3 on one fluid side with plain tubes. Plate-fin and tube-fin regenerators are examples of compact heat exchangers for

gas flows while gasketed, welded and brazed plate heat exchangers and printed-circuit heat exchangers are examples of compact heat exchangers for liquid flows.

1.2 Extended Surface Heat Exchangers

Heat exchangers, on the basis of constructional details, can be classified into tubular, plate-type, extended surface and regenerative type heat exchangers. The tubular and platetype exchangers are the primarily used surface heat exchangers with effectiveness below 60% in most of the cases. The surface area density of these heat exchangers is usually less then 700 m2/m3. In this regard, an important fact is that the thermal conductance hA on both sides of the heat exchanger should approximately be the same. Hence, the heat transfer surface on the gas side needs to have a much larger surface area as it is well known that the heat transfer coefficient h for gases is much lower than that for liquids. One of the most common methods to increase the surface area and compactness is to have extended surface (fins) with an appropriate fin density (fin frequency, fins/m) as per the requirement. This addition of fins can increase the surface area by 5 to 12 times the primary surface area. These types of exchangers are termed as extended surface heat exchangers. The heat transfer coefficient h on extended surfaces may be higher or lower than that of un-finned surfaces. The louvered fins increase both the surface area and the heat transfer coefficient, while the internal fins in a tube increase the tube surface area but may result in a slight reduction in heat transfer coefficient depending on the fin spacing. However, the overall thermal conductance increases due to the presence of extended surfaces. Heat exchanger design involves the consideration of mechanical pumping power expended to overcome fluid friction in addition to the consideration of heat transfer rate. The friction power expended with high density fluids is usually less as compared to the gain in heat transfer rate; however, for low density fluids, such as gases, the pumping power is of considerable magnitude relative to the gain in heat transfer rate. An increase in the velocity of fluid flow increases the heat transfer rate as something less than the first power of velocity, whereas, the frictional power expenditure increases as the cube of velocity and never less than the square of velocity. The frictional power limitations force the designers to keep the velocities moderately low. The flow velocities can be reduced by increasing the number of flow passages in the heat exchanger. It will reduce the frictional power much more as compared to the decrease in the heat transfer rate per unit

of surface area. This loss of heat transfer rate can be made up by an increase in the surface area which, in turn, also increase the frictional power, but only in the same proportion as the heat transfer surface area. This consideration also calls for the extended surface heat exchangers. Plate-fin and tube-fin heat exchangers are the two most common types of extended surface heat exchangers.

1.2.1 Plate-Fin Heat Exchangers

This type of extended surface heat exchanger has corrugated fins mostly of triangular or rectangular cross-sections sandwiched between the parallel plates as shown in Figure 1.1.

Figure 1.1 Basic components of a platefin heat exchanger The fins may also be incorporated in a flat tube with rounded corners. The parting sheet is usually replaced by a flat tube in the case of liquid or phase change fluid flows on the other side. Fins are die or roll formed and are attached to the plates by brazing, soldering, adhesive bonding, welding, mechanical fit, or extrusion. Plate-fins are categorized as: (1) plain i.e. uncut and straight fins, such as plain triangular and rectangular fins, (2) plain but wavy fins, and (3) interrupted fins such as offset strip, louvered fins, perforated fins etc. Figure 1.2 shows some of the most commonly used fins in parallel plate heat exchanger. The plates and the fins are made of a variety of materials - metals, ceramics and papers - with surface area density up to 5900

m2/m3. Plate-fin exchangers have been produced since the 1910s in the auto industry (copper fin-brass tubes), since the 1940s in the aerospace industry and in gas liquefaction applications since the 1950s using aluminum. They are widely used in electric power plants, propulsive power plants, systems with thermodynamic cycles i.e. heat pump, refrigeration etc and in electronic, cryogenic, gas-liquefaction, air-conditioning, waste heat recovery systems etc.

Figure 1.2 (a) Plate-fin heat exchanger and its geometries; (b) Plain rectangular fins; (c) Plain triangular fins; (d) Wavy fins; (e) Offset strip fins; (f) Perforated fins; (g) Louvered fins; after Webb [1987]

1.2.2 Tube-Fin Heat Exchangers

These heat exchangers may further be classified as (a) conventional and (b) specialized tube-fin exchangers. Tube-fin exchangers are employed when one fluid stream is at a high pressure and/or has a significantly higher heat transfer coefficient than that of the other fluid stream. In a conventional tube-fin heat exchanger, the transfer of heat takes place by conduction through the tube surface. In a specialized tube-fin exchanger i.e. heat pipe exchanger, tubes with both ends closed, act as a separating wall and heat is transferred through this separating wall by conduction and evaporation and condensation of the heat pipe fluid. This type of heat exchanger is similar to a fin-tube exchanger; however the tube is a heat pipe.

In a conventional tube-fin exchanger, round, rectangular and elliptical tubes are most commonly used. Fins are generally used outside the tube, however, they may be used on the inside of the tubes as well, if required (Figure 1.3).

Figure 1.3 Internally finned tubes (Axial and Helical fins) Depending on the fin type, tube-fin heat exchangers are further classified as (a) individually finned tube, (b) continuously finned tube and (c) longitudinally finned heat exchangers. Figure 1.4 shows two basic types of conventional tube-fin heat exchangers. Longitudinal fins are generally used in condensing applications. The fins of the tubes may be plain, wavy or interrupted. Tube-fin exchangers usually are less compact than plate-fin units. These exchangers are employed when one fluid stream is at a higher pressure and has a significantly higher heat transfer coefficient than that of the other fluid stream. Heat pipe heat exchanger consists of heat pipes which are basically the evacuated closed tubes, partially filled with a heat transfer fluid. The inner surface of the heat pipes are usually lined with a capillary wick. Hot and cold gases flow continuously in separate parts of the chamber. Heat is transferred from the hot gas to the evaporator section of the heat pipe by convection; the thermal energy is then carried away by the vapors of the heat pipe fluid to the condensation section of the heat pipe, where it transfers heat to the cold gas by convection. The heat pipe performance is influenced by the angle of orientation of the heat pipes. This tilting of the exchanger may control the pumping power and ultimately the heat transfer. These exchangers are primarily used in waste heat recovery systems. Tube-fin heat exchangers are usually less compact than plate-fin heat exchangers. A tube-fin exchanger having flat fins with 400 fins/m (10 fins / inch) has a surface area density of about 720 m2/m3.

Figure 1.4 Externally finned tubes (a) Individually finned; (b) Continuously finned

1.3 Heat Transfer Rate in Compact Heat Exchangers

The motivation to use the compact surfaces is to achieve specified heat transfer performance, q Tm within acceptably low mass and box volume constraints. The heat exchanger performance can be expressed asq = U HT A = U HT HT V HT Tm
(1.1)

where q is the heat transfer rate, UHT is the overall heat transfer coefficient based on area A, VHT is the exchanger volume, HT is the heat transfer surface area density (m2/m3) and for plate-fin exchangers, it is the ratio of heat transfer surface area for cold or hot fluid and the volume occupied by the respective heat transfer surface, Tm is the logarithmic mean temperature difference defined as Tm =

(T1
ln

T2 ) T1 T2

(1.2)

where T1 = TH2-TC1 and T2 = TH1-TC2 being the temperature differences between the hot and cold fluids respectively as shown in Figure 1.5. Clearly from Equation 1.1, a high value of HT minimizes the exchanger volume VHT for the specified q Tm .

The compact surfaces with low hydraulic mean diameter Dh generally results in higher convective heat transfer coefficient h and higher overall heat transfer coefficient

TH2 T1 TC1
UHT.

Flow of fluid

TH1 T2 TC2

Figure 1.5 Single pass counter flow heat exchanger This reduces the exchanger volume considerably for a desired heat transfer rate. Alternatively, if the exchanger volume remains same, the rate of heat transfer will increase. The compact surfaces can achieve structural stability and strength very easily even with thinner gauge material, the reduction in heat exchanger mass is more pronounced than the gain in a smaller volume.

1.4 Other Heat Transfer Enhancement Techniques

Bergles et al. [1983] identified about 14 enhancement techniques used for the heat exchangers. These enhancement techniques can be classified into active and passive techniques. Passive techniques do not require any type of external power for the heat transfer augmentation, whereas, the active techniques need some power externally, such as electric or acoustic fields and surface vibration.

1.4.1 Active Heat Transfer Enhancement Techniques

(a) Mechanical Aids

These aids consist of stirring the fluid or rotating the surfaces by mechanical means. Mechanical surface scrappers may be applied to the ducts of gases for the enhancement of heat transfer. Rotating heat exchanger ducts are commercially used to augment the heat transfer.

(b) Surface Vibration

Low or high frequency surface vibrations are used to promote single phase heat transfer augmentation. A piezoelectric device may also be used to vibrate a heat transfer surface. Heffington et al. [2001] used a piezoelectric transducer to vibrate a plate at about 2.5 kHz, which produces a shower of small diameter drops on the boiling surface. The concept used by Heffington et al. [2001] is called Vibration Induced Droplet Atomization (VIDA).

(c) Fluid Vibration

The fluid vibration is the more practical type of vibration enhancement due to the mass of the heat exchangers. The fluid vibrations range from pulsations of about 1 Hz to ultrasound and mostly applied for the single phase fluids. (d) Electrostatic Fields Electrostatic field produced by either direct current or alternating current is used for dielectric fluids to cause proper bulk mixing of the fluid in the vicinity of the heat transfer surface. The electric field applied to dielectric fluid imposes a body force on the fluid which influences the fluid motion. Yabe [1991] provided an excellent description of the fundamentals of EHD enhancement. (e) Injection Here the gas is supplied to a flow of liquid through the porous surface or the same fluid is injected upstream. The injected gas augments the single phase flow. Surface degassing of liquids may produce similar effects. (f) Suction Suction involves vapor removal, in nucleate or film boiling, or fluid withdrawal through a porous heated surface. This technique is applied to only single phase fluids.
(g) Jet impingement

Jet impingement involves spraying a liquid on the hot surface which spreads as a thin film and gets evaporated. For this purpose, single or multiple jets may be used. Spray cooling specifically involves impinging the liquid as small droplets. Pais et al. [1992] and Xia [2002] have worked on the water impingement. 8

1.4.2 Passive Heat Transfer Enhancement Techniques

(a) Coating of the Surfaces

Condensation occurs on the surface whose temperature is less than the vapor saturation temperature. This condensed liquid on the surface exists either as a wetted film or in droplets. Droplets are formed if the condensate does not wet the surface. Dropwise condensation yields a high heat transfer coefficient but it can not be sustained permanently. Non-wetting coating, such as Teflon, enhance the drop-wise condensation. A hydrophilic coating promotes the condensate drainage on evaporator fins by reducing the wet air pressure drop. Nucleate boiling can be enhanced by a fine scale porous coating. A porous coating on the base surface is an effective enhancement method for film condensation. Condensate drainage is assisted by capillary flow within the porous coating, resulting in a thinning of the condensate film thickness. The temperature drop across a laminar condensate film depends on the condensation thermal resistance and such capillary assisted film thinning reduces the condensate thermal resistance.

(b) Rough surfaces

Surfaces may be made rough by machining or restructuring the base surface or by placing some roughness adjacent to the surface e.g. a wire coil insert. So many possible roughness geometries are studied by the researchers and the three dimensional roughness geometries like cross-rifled tubes of Nakamura and Tanaka [1973], three dimensional ribs by Liao et al. [2000] etc. offer higher enhancement level. For single phase flow, mixing in the boundary layer is promoted near the surface rather than to increase the heat transfer surface area. A wire coil insert classified as wall attached roughness is shown in Figure 1.6. The knurled roughness on a vertical surface promotes the mixing in the condensate film.

(c) Extended Surfaces

It is a most common approach to enhance the heat transfer by using the extended surfaces. A plain fin may increase the surface area but a special shape extended surface may increase heat transfer coefficient in addition to the area of heat exchanger. Enhanced extended surfaces used for the gases are already discussed in section 1.2. The extended surfaces for liquids typically use much smaller fin heights than that used for gases because of the higher heat transfer coefficient for liquids. Use of high fins with liquids

would result in low fin efficiency and result in poor material utilization. Externally finned tube and internally finned tube are the examples of extended surfaces for liquids.

Figure 1.6 Wire coil insert

(d) Displaced inserts

These are the devices inserted into the flow channel to improve energy transport at the heated surface indirectly. The displaced inserts mix the main flow in addition to that in the wall region. Displaced wire coil insert (Figure 1.7) is not attached to the wall of the tube. These devices periodically mix the gross flow structure but not affecting the main flow significantly.

Figure 1.7 Displaced wire coil insert

(e) Swirl flow devices

These devices (Figure 1.8) include a number of geometrical arrangements or tube inserts for forced flow that create rotating or secondary flow. Full length twisted tape inserts or inlet vortex generator and axial coil inserts with a screw type winding are some examples of swirl flow devices.

10

(a)

(b)

Figure 1.8 Swirl flow devices (a) Helical vane insert (b) Twisted tape insert

(d) Surface tension devices

The local film thickness is determined by the force that drains the condensate. The use of surface tension forces to affect condensate drainage is an effective enhancement technique. Vertical fluted tubes are frequently used for vertical tube condensers used in desalination and are commercially available. Loosely attached axial wires (poor thermal contact) on vertical smooth tubes also provide surface tension condensation enhancement. The surface tension devices strictly do not increase the surface area of the base surface. Heat pipes typically use capillary wicking to transport liquid from the condenser section to evaporator section.

(e) Additives for liquids

Additives for single phase liquids may be solid particles or gas bubbles. Kafanov [1964] performed a detailed study of solid particles additives such as water-chalk, watercoal, water-sand, water-aluminum etc in a circular tube for varying Reynolds number. Now-a-days, however, nano-sized metallic particles are of considerable interest with regard to increase in thermal conductivity of the flowing medium. Li and Xuan [2002] reported a 24% higher heat transfer coefficient attributed to 100 nm copper particles with 1% concentration in water. Bubbling a gas through a stationary liquid stimulates the conditions for nucleate boiling because of the liquid agitation on the surface, caused by the vapor bubbles. Additives may also be such suspensions e.g. a dilute polymer-water solution, which reduce the fluid friction. Suspensions in dilute polymer and surfactants solutions reduce

11

both heat transfer and fluid friction. The use of a rough surface recovers some of the heat transfer reduction, however, at the expense of increased friction.

(f) Additives for gases

Solid additives are frequently used in fluidized beds which involve heat transfer between a bundle of tubes and a fluidized gas-solids media. Enhancement ratio for a gas solid suspensions flowing inside a tube is found to be 3.5 times high. Solid additives may be glass, sand, zinc, graphite, aluminum oxide etc. Liquid additives generally refer to water droplets added to air stream. The wetted heat transfer surface facilitates evaporation from the water film surface into the air stream. Thomas and Sunderland [1970] achieved an enhancement ratio of 20 by adding 5% water to the air stream. Moderate enhancement can also be achieved without wetting the surface. The upstream water mists cool the incoming air to its wet bulb temperature. Good enhancement can be achieved by having the water mist temperature below the air temperature.

1.5 Vortex Generator

One of the most important passive techniques to augment the heat transfer is the use of vortex generators. Transverse vortex generators produce vortices, whose axis is transverse to the main flow direction, whereas, the longitudinal vortex generators generate vortices whose axis is parallel to the main flow direction. It has been found that longitudinal vortex generators are more suitable than the transverse vortex generators when the heat transfer augmentation with pressure drop is an important consideration. The longitudinal vortices behind a slender aerodynamic object have been investigated for many years. Longitudinal vortices are found to persist for more than 100 protrusion heights downstream. A vortex generator is called a wing when its span is attached to the surface and is known as a winglet when its chord is attached to the surface. Longitudinal vortex generators may have any of the four basic shapes (Figure 1.9) i.e. delta wing, rectangular wing, delta winglet and rectangular winglet. The aspect ratio of a longitudinal vortex generator is the ratio of the square of the span b and the area of the vortex generator s i.e. b 2 s . The aspect ratio of vortex generator is an important criterion to compare the performance of the different shapes. 12

X
Y

Delta wing =2b/c

Rectangular wing =b/c

b/2 c b/2 c

Delta winglet =2b/c Rectangular winglet =b/c

Figure 1.9 Longitudinal vortex generators In case of winglet, single vortex is generated by the fluid which passes over the winglet; however, for the wing vortex generator, two vortices are produced as the obstructed fluid passes over the wing from both the side edges. Figure 1.10 shows a sketch of longitudinal vortices behind a delta winglet vortex generator placed in a laminar boundary layer on a flat plate (Torii et al. [1994]). The flow separation at the leading edge of the winglet generates a main vortex and the corner vortex is formed by the deformation of near-wall vortex lines at the pressure side of the winglet. Sometimes an induced vortex is also observed rotating opposite to the main and corner vortex. The winglet vortex generators may also be arranged forming and shaped pairs. When the direction of the secondary flow between two counter rotating vortices is away from the wall, the vortices are called common flow-up and when the direction is towards the wall, they are called common flow-down. The pairing of vortex generators produces common flow-down vortices and the pairing of vortex generators produces common flow-up vortices. Figure 1.11 shows the orientation of a winglet pair both in common flow-up and common flow-down configuration.

13

Figure 1.10 Vortex systems behind a delta winglet (Due to Torii et al. [1994]). Active vortex generation techniques are associated with control over heat transfer enhancement and pressure drop. When heat transfer augmentation is required, vortices are introduced at the expense of the power to produce vortices along with the added pressure drop. During normal operation, vortex generation is stopped. A number of ways to achieve this control are available; yet very little work is directed at active vortex methods. The use of an injected transverse jet is proved to be an effective active method to produce streamwise vortices. The jets injected are typically circular and are injected with particular pitch and skew angles with respect to the main flow as shown in Figure 1.12.

Flow Flow Flow

Flow

(a)

(b)

Figure 1.11 Orientation of a winglet pair (a) Common flow-up (b) Common flow-down 14

Vortices can be generated for a wide range of jet skew angle and studies show that a jet injected with a pitch of 45 and zero skew angle introduces two counter-rotating vortices with common out-flow.

Figure 1.12 Actively generated longitudinal vortices Electrohydrodynamics relies on an externally supplied electric field to produce an electric body force in the flow. This controlled and localized body force produces a secondary flow known as corona wind. These normal velocities introduce streamwise vorticity and the resulting secondary flow could take the form of a longitudinal vortex. Thus EHD is an active vortex induced heat transfer enhancement technique. Acoustic excitation is a somewhat different way to actively generate a secondary flow. The secondary flow may take the form of longitudinal vortices; however its manifestation is highly dependent on the geometry and flow conditions.

1.6 Motivation for the Present Work

In a plate-fin heat exchanger, the hot and cold fluids flow alternatively between the parallel plates. Physical mixing of the fluids is not permissible. Therefore, in this case, one of the economical methods i.e. punching out the longitudinal vortex generator on the plates of a plate-fin heat exchanger is ruled out. The only option available to attach the vortex generators to the plate surface is by the use of welding, brazing etc. The additional cost and complexity involved in joining the vortex generators to the plate surface are significant. A way out of this difficulty is proposed in the present work. The solution suggested is to punch out the vortex generators on the plane surfaces of the triangular

15

inserts placed between the adjacent plates. With such an arrangement there is no scope of the mixing of fluids. Thus the proposed device is an innovative combination of the plain triangular fin and the perforated fin to form a triangular fin which is provided with the delta/rectangular wing type vortex generators on its slant surfaces.

1.7 Layout of the Thesis

Subsequent chapters of the present dissertation have been organized in the following manner: Chapter-2 of the thesis provides an extensive review of the literature in the field of heat transfer enhancement by the application of vortex generators. Various techniques for solving Navier-Stokes equations are also discussed. The problem formulation is explained in Chapter-3. The complete details regarding the geometry, governing equations, boundary conditions, method of solution used to obtain the numerical solution is given here. Testing of discretization scheme, the grid independence test and the validity of the code is also presented in this chapter. Chapter-4 presents the performance characteristics of the rectangular wing vortex generator in enhancing the heat transfer as well as the associated increase in pumping power for different Reynolds numbers and various angles of attack of the wing. Results are analyzed for both the builtin and stamped wing vortex generators. Chapter-5 provides the results for the case of delta wing vortex generator and also compares the performance of delta and rectangular wings for the two cases (a) same span and chord length of the wings (b) same area of the rectangular and delta wings. Finally Chapter-6 contains the conclusions of the present work besides scope for future work in the same area.

16

Chapter 2 Review of Literature 2.1 Introduction

A considerable amount of experimental as well as analytical and computational research has been carried out on the enhancement of heat transfer. In this chapter, a brief survey of the relevant literature is presented to indicate the extent of work already reported in open literature pertaining to the enhancement of heat transfer by introducing protrusions mounted on the heat transfer surfaces. The literature is reviewed from two different view points. First section deals with the overview of the past work involving the use of various vortex generators on different surface geometries i.e. flat plate, rectangular channel, fin-tube heat exchangers etc. The flow regimes for such geometries may be laminar or turbulent. In order to analyze the flow structure and heat transfer characteristics in these domains, a detailed computational study is needed. The second section concerns with the review of the available schemes to solve the complete NavierStokes equations.

2.2 Enhancement of Heat Transfer Using Vortex Generators

Improvement of the performance of heat exchangers with gas as working fluid becomes particularly necessary because of the high thermal resistance offered by gases. To compensate for the poor heat transfer properties of the gases, the surface area density of a parallel plate heat exchanger may be increased by making use of the secondary fins i.e. offset fins, triangular fins, wavy fins, louvered fins etc, as discussed in Chapter-1. In addition, a promising passive technique for the enhancement of heat transfer is the use of vortex generators. Transverse vortex generators produce vortices with axes oriented in

17

transverse direction with respect to the main flow direction. The Karman vortex street in the wake of a circular cylinder is a thoroughly investigated transverse vortex system. For the ribbed channel geometry, transverse vortices are generated at the ribs and also induced on the plane or smooth wall. The flow becomes unsteady for this configuration at a Reynolds number - based on the mean velocity and rib height - larger than 46, while for plane channel flow, the transition Reynolds number is of higher magnitude, as reported by Grosse-Gorgemann et al. [1995]. Grosse-Gorgemann et al. [1993] showed that the enhancement mechanism by transverse vortex generators need unsteady flow and develop reversed flow regimes which further increase the resistance to flow. No enhancement in heat transfer was reported for steady flow in a periodically ribbed channel. Ghaddar et al. [1986] and Amon and Mikic [1990] investigated numerically the grooved channel flow where the grooves were so short that the separated flow was attached at the face of the next protrusion instead of the base of the groove. Hermann and Mayinger [1991] and Herman et al. [1992] made experimental investigations for similar geometries. Greinal et
al. [1986] and Fiebig et al. [1994(a)] investigated grooved and ribbed channel flow

experimentally and numerically. They all concluded that heat transfer enhance considerably when the flow is unsteady. The literature regarding transverse vortex systems is not reviewed further as the present work concerns the longitudinal vortex generators. The longitudinal vortex generators such as winglets or wings generate vortices with axes parallel to the main flow direction and always imply three dimensional flows. The longitudinal vortices produce strong swirling flow along the main flow resulting in an exchange of working fluid between the core and the heat transfer surfaces and hence, a lesser amount of energy is needed to turn the flow. Therefore, longitudinal vortex generators are preferred when pressure loss is also an important consideration along with heat transfer enhancement. Literature reveals several studies on heat transfer enhancement using longitudinal vortex generators with various types of geometrical configurations such as plate-fin heat exchangers, tube-fin heat exchangers etc. Experimental as well as numerical computations are performed to evaluate the performance of the longitudinal vortex generators by so many researchers. An extensive review of the progress in this area has been presented by Jacobi and Shah [1995] and Fiebig [1995(a), 1998].

18

2.2.1 Vortex Generator Enhanced Flat Plate Flows

Flat plate flows are approximated in many external and developing internal flow applications. Earlier work in this area was reported by Edwards and Alkar [1974]. They investigated the heat transfer enhancement on a flat plate using a row of built-in triangular and rectangular wings of small aspect ratios. Heat transfer enhancement was observed over several chord lengths behind the wing surfaces. A mean spanwise heat transfer enhancement above the flat plate value of about 40 % was measured for a Reynolds number of 61,000 based on the wing size. Turk and Junkan [1986] investigated heat transfer enhancement for laminar flow over a row of rectangular winglet pairs by varying the aspect ratio. The angle of attack was fixed at 20. It was found that the ratio of span averaged heat transfer coefficient on a flat plate with vortex generator to the corresponding value without vortex generator increased up to 3 at a distance more than 30 chord lengths downstream of the winglets. The study was carried out both for zero and favorable pressure gradients and heat transfer enhancement was found to be more with favorable pressure gradient. Torii et al. [1991] investigated local heat transfer downstream of a single delta winglet vortex generator on a flat plate. Flow visualization experiments were conducted to study the flow field and hot wire anemometer was used to measure the velocities. Naphthalene sublimation and surface thermocouples with an imposed heat flux were used to measure the heat transfer. The free stream velocity was fixed to 4 m/s. Local heat transfer enhancement of over 200% was reported in the downwash region of the flow. The velocity data provided the information about the vortex location. Yanagihara and Torii [1993] extended their earlier work by investigating multiple delta winglet vortex generators on a flat plate. All measurements were made at the same velocity. Multiple delta winglet vortex generators produce multiple vortices but the interaction between the neighboring vortices and the resulting effect on the heat transfer was not discussed. Gentry and Jacobi [1997] studied the effect of streamwise vortices induced by delta wing vortex generator. Flow visualization techniques were used to study the flow and naphthalene sublimation (Figure 2.1) was used to get the heat transfer effects. It was concluded that the maximum heat transfer enhancement is observed when a vortex is located near the edge of the thermal boundary layer. Further it was reported that the vortices should be generated in a common inflow arrangement so that the induced velocities keep the vortices near the boundary layer.

19

Figure 2.1 Geometrical definitions of the test specimen Eibeck and Eaton [1987] studied a single vortex using a Rankine vortex model for the turbulent flow and velocity data. They interpreted their data in terms of vortex circulation and boundary layer thickness. The experiments were conducted using a constant heat flux surface. The local increase in the Stanton number was attributed to a thinning of the boundary layer on the downwash side of the vortex. Pauley and Eaton [1988] extended this work for the vortex pairs. Co-rotating pairs moved together and coalesced into a single vortex while being advected downstream. This research provided useful insights about the vortex-vortex and surface-vortex interactions.

2.2.2 Vortex Generator Enhanced Fin-Tube Exchanger Flows

Fin-tubes cross flow heat exchangers are commonly used in many applications where liquid flows through the tubes and gas flows over the finned tube. The fins are a series of thin parallel plates through which the tubes pass perpendicularly. One of the fintube heat exchanger is shown in Figure 2.2. The flow field around a tube (Figure 2.3) consists of a horseshoe vortex or corner vortex in the stagnation region, a von-Karman vortex street in the middle and a dead water zone at the juncture of the tube and the plate. In order to enhance heat transfer in such a flow configuration, vortex generators are mounted in various arrangements most common being the pairs of winglet in common flow-up arrangement and the in-line wings. The tubes arrangement may also be in-line or staggered.

20

Plate-fins

Fluid flow In-line Tubes

Gas Figure 2.2 In-line tube fin heat exchanger Biswas et al. [1994(a)] numerically investigated the flow structure and heat transfer enhancement in a staggered row circular tube-fin channel with delta winglet vortex generators mounted on the fin surfaces. Steady solutions were obtained up to a Reynolds number of 500. At the channel inlet, a fully developed velocity profile for the axial velocity was assumed. The winglet vortex generator placed in the wake region of the circular tube enhances the heat transfer by as high as 240 % along with the increased overall channel heat transfer.

Figure 2.3 Flow structure around a circular tube on a plate

21

Cross-stream velocity vectors confirmed the formation of horseshoe vortex system. A modified version of the MAC method was employed to solve the governing equations for the incompressible, viscous flow. Torii et al. [2002] numerically evaluated the delta winglet pair in common flowup configuration at low Reynolds number to meet the various demands of the designers such as compactness, fan power saving and quietness etc. This configuration accelerates the fluid flow and as a consequence, the delay in separation occurs and form drag is also reduced due to narrowing of the wake and suppression of the vortex shedding. The zone of the poor heat transfer is also removed, since the fluid is accelerated in this passage. In an in-line tube arrangement, the common flow-up configuration augmented the heat transfer by 10 to 20 % and simultaneously decreased the pressure drop by 8% to 15%. A much better performance was observed in a staggered tube arrangement with the same common flow-up configuration. The in-line and staggered tube configurations are shown in Figure 2.4. For a Reynolds number of 350, a pressure loss reduction of 55 % was achieved together with a heat transfer enhancement of 30%. Torii et al. [2002] also studied the in-line tube banks with delta winglets in common flow-down configuration as proposed by Fiebig et al. [1993]. The vortex generators enhanced the heat transfer by 10 % to 25 % with 25 % to 35 % increase in pressure penalty. Further, this configuration is not so effective for low Reynolds numbers. The corresponding increase in heat transfer was also less for the staggered tube arrangement. Kwak et al. [2003] experimentally evaluated two to five rows of staggered circular tube bundles with a single transverse row of delta winglets in common flow-up configuration placed beside the front row of tubes. For three row tube bundles, the heat transfer was augmented by 10 % to 30 % and yet the pressure loss was reduced by 34% to 55% with an increase in Reynolds number from 350 to 2100. Pesteei et al. [2005] performed the experiments to study the effect of the winglet location on the heat transfer enhancement and pressure drop in a fin-tube heat exchanger. Height of the delta winglet was the same as of the channel and the aspect ratio and the Reynolds number was fixed at 1.33 and 2250 respectively. The winglet was mounted at an angle of attack of 45o, which is the best angle for the fin-tube arrangement as reported by Fiebig et al. [1990]. The study showed that for the highest local heat transfer coefficient, the winglet pair should be placed at a distance of half of the tube diameter both in X and Y directions. The winglet pairs were found to be most effective when placed on the downstream side. Mounting the delta winglet pair on the upstream side did not 22

produced any significant effect on the heat transfer coefficient and it was argued that at this location, winglet pair produces very strong horse-shoe vortices because of the presence of the tube.

In-line tube arrangement

Staggered tube arrangement

Figure 2.4 Geometric arrangements of tube rows and vortex generators Tiwari et al. [2003] studied numerically the various combinations of the delta winglet pairs in a rectangular channel with a built in oval tube. A finite volume method due to Eswaran and Prakash [1998] was used to discretize and solve the governing equations. The spanwise average Nusselt number for the case of four winglet pairs was found to be about 100% higher as compared to the no winglet case at a Reynolds number of 1000. Some different combinations of the pair of delta winglet vortex generators were also analyzed by Prabhkar et al. [2003] in a rectangular channel with built-in oval tube. Tiwrai et al. [2005] studied numerically the effect of wake splitter placed behind the circular tubes in a cross flow configuration. Chen et al. [1998] predicted the influence of the angle of attack and the aspect ratio of a punched delta winglet pair placed near the leading edge of the finned oval tube. Here the non-isothermal boundary condition of the fin was considered due to conjugate heat transfer in the finned tube. The computational domain was discretized into a finite number of control volumes and the winglet was

23

approximated by the interface between two control volumes. A slight change in the location of the winglet pair produced little change in the heat transfer rate. Fiebig et al. [1993] experimentally evaluated the effect of delta winglets in a tubefin heat exchanger. For the in-line tube arrangement, the winglets caused a 55-65% increase in heat transfer with a 20-45% increase in friction factor within the Reynolds number range 600-270. Fiebig et al. [1994(b)] performed the experiments to compare the round and flat tubes with longitudinal vortex generators. For the staggered fin and tube arrangement, the heat transfer was increased by 10% for round tubes as against a much more significant 100% for flat tubes. The loss in pressure was also half of that for round tubes. Lawson and Thole [2008] investigated both the built-in and punched delta winglets on the tube surface of louvered fin heat exchangers. Louvered fin heat exchangers are commonly used heat exchanger designs in automobile industries. These types of designs are preferred as the higher efficiencies involved allow the heat exchangers to be smaller and lighter with an acceptable increase in the pressure losses. A good attempt has been made to investigate the combined effects of winglets and louvered fin-tube heat exchangers. Heat transfer augmentation of about 47% with a corresponding increase of 19% in pressure losses was observed. Allison and Dally [2007] conducted experiments to analyze the combined effects of the delta winglet vortex generator in common flow-up configuration and a louver fin surface in a fin-tube radiator. Water tunnel dye visualization technique was used to study the flow structure. Mounting the vortex generators in a flow-up or converging manner in front of the tubes give rise to a significant portion of the flow impinging on the tube stagnation zone, thus increase Nusselt numbers along the tube side. Delta winglet geometry improved the heat transfer by 87 % of the capacity of the louver fin surface. On the other hand, a substantially lower pressure drop of approximately 53% of the louver surface was reported. Zhang [2008] compared the tube-fin with mounted and punched vortex generators. Both staggered and in-line arrangements of the tube bank were investigated. Experimental techniques used were naphthalene sublimation and condensation test. Joardar and Jacobi [2008] experimentally assessed the overall heat transfer and pressure drop performance by mounting the delta winglet pairs in common flow-up configuration on the plain fins. The spacing between the fins (3.5 mm) was typical to those used in aircooling and refrigeration applications and approximately 3500 winglets were used in an in-line fashion for the three row alternate tube configuration and one third of that for the 24

single row leading tube arrangement. For Reynolds numbers between 220 and 960, the air side heat transfer coefficient was shown to increase by 16.5-44 % for single row vortex generator pair and 30-68.8% for three row vortex generator pairs. Chang et al. [2009] numerically studied the relationship between heat transfer enhancement and the intensity of the secondary flow produced by the delta winglet vortex generators mounted on a three row flat tube bank fin surfaces. It was revealed that the cross-averaged absolute vorticity flux in the main direction reflects the intensity of the secondary flow and exhibited the same trend with Nusselt number as well. Wen and Ho [2009] experimentally studied the fin-tube heat exchanger by varying the design of the fins namely plain fin, wavy fin, and compounded fin. The results of the compounded fin compared to the plain fin showed that the pressure drop, heat transfer coefficient, f factor and j factor increased by about 33.563.1%, 27.045.5%, 6.971.1% and 9.413.2%, respectively. Eiamsa-ard et al. [2010] investigated experimentally the heat transfer, flow friction and thermal performance factor characteristics in a tube fitted with delta winglet twisted tape, using water as working fluid. Chu et al. [2009] performed the numerical study on the heat transfer characteristics and flow structure of fin-and-ovaltube heat exchangers with punched delta winglet vortex generators placed in common flow-up configuration. The analysis of the heat transfer enhancement was also done with the view of field synergy principle. The average Nusselt number and the friction factor decreased with the increase in tube row number. The lesser the tube row number, the better the heat transfer rate and the better the field synergy. The vortex generators located downstream of oval tubes were more effective than those located upstream of oval tubes for heat transfer enhancement. Webb. [1980] established a broad range of Performance Evaluation Criteria (PEC) applicable to single phase flow in tubes. Detailed procedures are outlined to select the optimum surface geometry.

2.2.3 Vortex Generator Enhanced Channel Flows

Channel flow is the basic flow for many internal flows, especially for many compact heat exchangers. Heat transfer in laminar channel flow is distinct from the corresponding flat plate flow, at least with respect to two aspects. First, the channel has two walls and when vortex generator is attached to one wall; it affects the heat transfer on both the walls. Second, a favorable pressure gradient always persists in channel flows and it has a tendency to become fully developed in the downstream direction. The vortex

25

generators may reduce the critical Reynolds number of transition drastically. A channel flow with vortex generators may be highly unsteady or in a state of transition to turbulence at very low Reynolds numbers. Fiebig et al. [1986] studied experimentally delta and rectangular wing and winglet vortex generators in a flat plate channel. The Reynolds number was assigned the values of 1360 and 2270 depending on the spacing between the plates. Aspect ratios of the wings and winglets were fixed at 1.25 and 1.0 respectively. Delta wing vortex generator enhanced the heat transfer by as high as 200% and overall Colburn factor was increased by 20 to 60 % at the Reynolds number 1360 by increasing the angle of attack from 10 to 50. Fiebig et al. [1991] further extended their work by considering the punched triangular and rectangular wings and winglets. Unsteady liquid crystal thermography was used to get the heat transfer coefficient and the drag - a measure of flow losses - was measured by balance. The flow visualization was done by laser light sheet. Local heat transfer enhancement of about 300% was achieved just one chord length behind the delta wing for an angle of attack of 30. It was concluded that per unit vortex generator area, delta wings are most effective, closely followed by delta winglets and delta winglet pairs. Fiebig et al. [1995(b)] further extended the work with built-in rows of rectangular winglets. Eight different winglet arrangements resulted from the combination of in-line or staggered, symmetric or parallel relative positioning, and attachment, either on one side or alternating both sides were thoroughly analyzed. Maximum average heat transfer enhancement was achieved for the in-line, symmetric winglet vortex generator configuration when attached alternatively to both the plates. Laminar flow and heat transfer characteristics in a rectangular channel with builtin delta wing and winglet pair was analyzed by Biswas et al. [1994(b)]. Figure 2.5 shows a delta wing vortex generator in a rectangular channel. A modified version of MAC algorithm was used to obtain the solution of governing equations. Combined upwinding and central differencing was employed for the discretization of the convective terms. An evaluation of the effect of vortex generator was also done from a thermodynamics viewpoint. Both the evaluation criteria confirmed the use of winglets to be a more effective augmentation technique. Biswas et al. [1996] determined the flow structure developed by delta winglet vortex generator placed in a fully developed channel flow. Experiments were performed to corroborate the numerical predictions of the flow structure. Vorticity contours confirmed the formation of main vortex, induced vortices and the corner vortex. A much higher value of (j/f) for an angle of attack of 15 was 26

computed numerically as compared to those for 22.5, 30 and 37.5. Deb et al. [1995] proposed a numerical model to compute both laminar and turbulent flow through a rectangular channel containing built-in winglet vortex generators. The flow was described by the unsteady Reynolds averaged Navier-Stokes equations and the k- model of turbulence. At a non-dimensional distance of 2.5 from the inlet, for a Reynolds number of 2000, an enhancement of 30% in the combined spanwise average Nusselt number is observed over the corresponding value for a channel without any obstacle. Wall function approach was used in turbulent flow computations which greatly reduced the storage and computational time. Height of the winglets was reduced while studying for the turbulent flow with a view not to enhance much pressure penalty.
B

Delta wing
L

T=TW Upper plate

H

Lower plate

X Z Y

Main flow direction

Figure 2.5 Delta wing placed in a rectangular channel Numerical simulations of turbulent flows in a rectangular channel with mounted vortex generators on the bottom wall were carried out by Zhu et al. [1993(a)], [1993(b)]. Zhu et al. [1995] extended their work by considering rib-roughness elements on the top plate and rectangular pair of winglets on the bottom wall. The flow field was calculated by solving Reynolds-averaged Navier-Stokes and energy equations, and the turbulence was taken into account by solving standard k- model of Launder and Splading [1974]. More than 450 % enhancement of the Nusselt number was reported as the combined effect of rib-roughness and rectangular winglets at a Reynolds number of 1.5105. Common flow-up and common flow-down configurations were numerically investigated in a rectangular channel three-dimensional incompressible viscous flow by

27

Yang et al. [2001]. The pseudo-compressibility method was introduced into the Reynolds-averaged Navier Stokes equation. A two layer turbulence model was adopted. One equation model was used for the inner layer and the standard k- model was applied on the outer layer. Turbulent kinetic energy was reported to be higher near the wall. The shape of the vortices changed to an ellipse expended in the spanwise direction in case of common flow-down; however, in the case of common flow-up, the vortices expended in the vertical direction. Kataoka et al. [1977] indicated that heat transfer was locally enhanced in the region where two neighboring vortices induced the flow towards the heat transfer surface (downwash region). Local thinning of the boundary layer associated with the secondary flow was found to be responsible for the heat transfer augmentation. Figure 2.6 and 2.7 show the region of boundary layer thinning and boundary layer thickening for the common flow-down and common flow-up configurations respectively. Yang et al. [2008] further predicted the effects of delta winglet pair in common flow-up configuration in a rectangular channel flow. In the case of common flow-up pair, the distortion of the thermal boundary layer was not as strong as the distortion of the hydraulic boundary layer. The heat transfer enhancement was maintained at the downstream location 30 times as large as the chord length of the vortex generator.

Figure 2.6 Boundary layer thinning

28

Tigglebeck et al. [1992] experimentally predicted the flow structure and heat transfer enhancement by using single and double rows of punched delta winglets. For an aligned arrangement of two rows of vortex generators, the flow structure in the wake of the second row is qualitatively similar to that of the first row. The peak value of the spanaveraged Nusselt number at the wake of the second row is strongly dependent on the spacing of the two rows. Tiggleback [1994] further compared the four basic forms of vortex generators i.e. delta wing, rectangular wing, pair of delta winglets and pair of rectangular winglets. These vortex generators were punched out form the parallel plates. Winglets performed better than the wings and a pair of delta winglet performed slightly better than the rectangular winglet at higher angles of attack and at higher Reynolds numbers. Fiebig [1998] presented a survey on triangular and rectangular protrusions in boundary layer and channel flows. Wings and winglets were considered either by themselves, or in a single row transverse to the flow direction, or in a two dimensional array. Heat transfer enhancement was higher in laminar flow than in turbulent flow and for single vortex generators, the heat transfer enhancement increased with increase in angle of attack and vortex generator area. For single row of vortex generators, heat transfer enhancement increased further with decreased transverse spacing of the vortex generators. A zero spacing of the winglet tips was found optimal for counter rotating delta winglet rows. Dense configurations with small angles of attack and small winglet to channel height ratios led to relatively high values of heat transfer enhancement as compared to the flow loss penalty. In this analysis, the Reynolds number was varied from 2000 to 9000 and the angle of attack was varied from 30 to 90. The performance of the winglets was found better than the wings and with regard to the winglet shape, a pair of delta winglets was better than rectangular winglets at higher angles of attack and higher Reynolds number. Jacobi and Shah [1998] studied the behavior of air flows in complex heat exchanger passages with a focus on the heat transfer effects of boundary-layer development, turbulence, spanwise and streamwise vortices, and wake management. Each of these flow features has been discussed for the plain, wavy, and interrupted passages found in contemporary compact heat exchanger designs. The interrupted passages such as spine and pin-fin geometries rely heavily on the boundary-layer restarting and vortex shedding mechanisms. However, since the wake region is much more prone to

29

turbulence, these geometries tend to make an early transition to turbulence. Thus, high heat transfer along with higher pressure drop was reported.

Figure 2.7 Boundary layer thickening Maughan and Incropera [1991] experimentally investigated winglet pairs with perforated ribs mounted on the top surface of the channel to take advantage of the buoyancy driven flows. The buoyancy forces produce secondary flows which in turn enhance heat transfer from the bottom surface of a heated horizontal channel. In some of the applications, an enhanced heat transfer at one surface may be of little advantage if an equivalent enhancement does not exist at the opposite surface. At low Rayleigh numbers, the vortex generators augment heat transfer at the top surface by mechanically inducing vortices. Flow visualization showed the mechanically driven secondary flow at the top surface to be essentially an inverted image of the buoyancy driven flow at the bottom surface. At higher Rayleigh numbers, vortex generators were no longer effective.

30

Perforated ribs, on the other hand, were effective throughout the entire range of Rayleigh numbers. Brockmeier et al. [1993] compared the performance of a parallel plate-fin channel using delta wing vortex generators with that of four standard heat exchanger surfaces-two plain fins, an offset strip and louvered fin geometry. For the standard surfaces, the basic performance characteristics in the form of heat transfer and friction data versus Reynolds number were taken from published experimental results; however, in the case of vortex generator surface, numerical prediction was done. For the quantitative comparison of the different surfaces, the arithmetic expressions of the characteristics of j and f versus Reynolds number were derived. The vortex generator surface allowed a reduction of 76% in the heat transfer surface area for fixed heat duty and for fixed pumping power. The comparative assessment involved some uncertainties related to the experimental and numerical basic performance data and their interpretation. Heat transfer can be enhanced by increasing the surface area density and the plain triangular fins as the inserts between the parallel plates of a plate-fin heat exchanger are widely used to increase the surface area density. Plate-fin isosceles triangular ducts were investigated for the hydrodynamically developed laminar forced flow by Zhang [2007]. Conductance of the fin from 0 to infinity and convection of the fluid was also considered which made it a conjugate problem. Apex angle of the triangular inserts was varied from 30 to 120. The Nusselt number increased with the increase in fin conductance. Vasudevan et al. [2000] investigated such geometry of a parallel plate heat exchanger with the triangular fins placed between the plates. The delta winglet vortex generators were mounted on the slant surfaces of the triangular inserts. Thickness of the vortex generator was not considered. Heat transfer enhancement of 20% to 25 % was achieved at the expense of a moderate pressure drop. Flow was considered to be laminar and due to the low hydraulic mean diameter, Reynolds number was varied from 100 to 200. Hiranvar et al. [2007] considered a delta winglet pair of non-zero thickness in a hydrodynamically developed and thermally developing laminar channel flow. It was concluded that spanwise average Nusselt number increases with an increase in the thickness of the winglets and it was reasoned that the finite thickness of winglet provides more cross-sectional area for energy transfer from the bottom plate and hence, results in increased heat transfer. To study the influence of the thickness of winglet, the computations were carried out with W/H = 0.0622, 0.1244, 0.1866 and 0.2486. Here W 31

is the width of the winglet and H is the distance between two parallel plates of the channel. As compared to the case of W/H of zero, the increase in the overall heat transfer of channel is 0.83%, 4.34%, 7.62% and 12.49% for W/H = 0.0622, 0.1244, 0.1866 and 0.2485, respectively. The comparison of single winglet and a winglet pair was also done and was found that the enhancement in heat transfer due to a winglet pair is twice that of a single winglet. Wu and Tao [2008] numerically investigated the influence of various parameters i.e. location of the winglet pair, space between the winglet pair, area and geometry of the vortex generator etc, on the heat transfer enhancement and flow resistance in a rectangular channel. They concluded that overall Nusselt number of the channel was found to decrease with increasing distance of the rectangular winglet pair from the inlet of the channel as well as with decreasing space between the pair. The location of the pair had no significant influence on the total pressure drop of the channel. With the area of the rectangular winglet pair fixed, increasing the length of the vortex generator resulted in an enhancement of heat transfer which was more pronounced as compared to that observed on increasing the height of the vortex generator. Delta winglet pair was proved to be more effective than rectangular winglet pair with regard to heat transfer enhancement for a given area of the vortex generator. For the punched delta winglet pair, validation was also done by conducting experiments in a wind tunnel. Some of the different shaped vortex generators are also studied by various researchers. Sohankar and Davidson [2001] investigated an inclined block shaped vortex generator mounted on one side of a channel flow at different Reynolds numbers. An incompressible finite volume code based on a fractional step technique with a multigrid pressure poisson solver and a non-staggered grid arrangement was used. The convective terms were discretized by either second-order central or third order QUICK differencing scheme. Influence of the thickness was also investigated at different Reynolds numbers. It was concluded that dissipative schemes like QUICK should not be used when studying transitional flows and the increase in the thickness of vortex generator makes strong and bigger streamwise vortices, which give rise to a higher Nusselt number. Sohankar [2007] extended the previous work of Sohankar and Davidson [2001] and Sohankar [2004] for a larger range of Reynolds number in a rectangular channel with a pair of angled ribs as a vee-shaped vortex generator. The angle of the vortex generator with respect to the main flow was between 10and 30. Both the DNS and LES obtained the similar results using LES simulation with fine grid and a DNS simulation with finer grid. The fluid flow and 32

heat transfer was unsteady for Reynolds number larger than 1000 as reported earlier by Sohankar and Davidson [2001]. The performance parameter j/f, increased with increasing Reynolds number or the incidence angle. Saha et al. [1999] numerically simulated the flow past a square cylinder placed centrally in a parallel plate channel at a Reynolds number of 21400. An effort had been made to capture the essence of time-averaged flow quantities through the turbulence models in two dimensions. Three turbulence models, namely, the standard k-, KatoLaunder k- and the RNG k- had been taken up for this purpose. All the three models clearly revealed the vortex shedding phenomena at nearly identical Strouhal numbers. Saha et al. [2000] predicted the vortex structure and kinetic energy budget in two dimensional flow past a square cylinder. Saha et al. [2003] extended the study of flow past a square cylinder in three dimensional but at low Reynolds number.

2.3 Numerical Methods for Solving Navier-Stokes Equations

Incompressible viscous flow simulations suffer from the problem that there is no direct equation for the prediction of the pressure terms; however, the spatial coupling of the pressure and the velocity exists. Pressure for the incompressible flow problems is a relative variable which adjusts itself instantaneously for the condition of zero divergence to be satisfied at all the computational cells. So the explicit determination of the pressure field is not feasible. The complications with the determination of pressure field have led to the development of methods which eliminate the pressure from the governing equations. The stream function formulation is a powerful method for solving incompressible Navier- Stokes equations. Here the pressure is completely eliminated by cross differentiation of the momentum equations. The stream function and the vorticity together constitute a second order elliptic system which is very well explained in literature. But the major drawback of this approach is that it cannot be applied to three dimensional problems and also the pressure terms have to be recovered from the computed stream function and vorticity calculations, as required at least on the solid boundary for calculating the drag forces. A partially implicit technique which takes into account the spatial coupling between the velocity and pressure fields requires a primitivevariable approach in which 33

the primitive variables u, v, w and p are mentioned as a function of x, y, z, t and Reynolds number. Here it is implicitly assumed that the momentum equation in x direction determines the x component velocity and so on. Since the mass balance equation is applied for pressure calculation, a separate equation for pressure is not required unlike in the stream function-vorticity approach. Harlow and Welch [1965] have used a staggered grid instead of conventional grid in the well-known explicit method of Marker and Cell which is a two step procedure. In the first step, provisional values of the velocity components are explicitly computed using the advection, diffusion and pressure terms of the previous time step. The velocity field obtained in this manner is not free form the divergence of mass. So in the second step, these velocity components are corrected to satisfy the continuity equation. Chorin [1967] developed a related technique which involves the simultaneous iteration of pressure and velocity components. Vicelli [1971] showed the equivalence of these two methods. Harlow and Amsden [1970], Nichols and Hirt [1971] and Hirt and Cook [1972] modified the MAC method for application to free surface flows. The MAC method is extensively used by many researchers. The modified MAC algorithm was effectively used by Biswas et al. [1990], [1992] and [1994(a), (b)] to compute the flow structures in a rectangular channel with various vortex generators. Saha et al. [2002] analyzed the flow past a square cylinder by the MAC method modified by Hoffman and Benocci [1994]. The MAC method is successively used by Deb et al. [1995], Robinchaux et al. [1992] and by so many researchers to simulate unsteady turbulent flows. However, this method has stability problems which slow down the calculations for the steady flow considerably. The implicit methods do not require the stability considerations and hence, these are more attractive methods. The algorithm known as SIMPLE (Semi- Implicit Method for Pressure Linked Equations) is based on finite volume discretization of the governing equations on a staggered grid and was introduced by Patankar and Spalding [1972]. The SIMPLE algorithm and its variants were extensively used by the researchers for the numerical simulation of incompressible flows. The SIMPLER algorithm of Patankar [1981] and the SIMPLAC algorithm of Van Doormaal and Raithby [1984] are the modifications of the original SIMPLE method. A comparative illustration of the operator splitting algorithm, viz., the PISO of Issa [1986] and the SIMPLE family of algorithms was reported by Jang et al. [1986]. Karki [1986] extended the SIMPLE algorithm to investigate compressible and incompressible viscous flow problems with shock waves in complex geometries. The collocated grid arrangements for boundary fitted coordinate was 34

reported by Rhie and Chow [1983] and Peric [1985]. Thompson et al. [1982] made considerable contributions to the development of numerical grid generation techniques for solving elliptic partial differential equations for both the external and internal flow problems. Kost et al. [1991] and Majumdar et al. [1992] modified the SIMPLE method for solving incompressible flows in arbitrary geometries. Mukhopadhyay et al. [1993] developed a SIMPLRElike algorithm for viscous flows in irregular geometries. The ideas of element-wise interpolation and transformation of non-orthogonal element geometry into a square computational element, as in finite element methods, are employed while solving the integral conservation equations. Kobayashi and Pererira [1991] modified the momentum interpolation method given by Peric [1985] and named it as Pressure Weighted Interpolation Method. This method solved explicitly the non-orthogonal terms in the momentum equation and dropped these terms in the pressure correction equations. Eswaran and Prakash [1998] proposed a finite volume based algorithm for solving incompressible fluid flow equations in a complex geometry. Prabhkar et al. [2003] successfully applied this algorithm in a channel with built in oval tube and vortex generator.

2.4. Objectives of the Present Study

Literature survey shows the potential of longitudinal vortex generators to augment heat transfer in various types of geometries. However, only little work is done with a combination of longitudinal vortex generators and secondary fins as inserts between the parallel plates of a plate-fin heat exchanger. The present study is carried out on triangular shaped secondary fins, with delta/rectangular wings on their slant faces, sandwiched between the parallel plates of plate-fin heat exchanger. The following key issues have been considered as the objectives of the present research: 1. To visualize the cross-stream velocity vectors along and beyond the wing vortex generator. Secondary flow structure is analyzed as the fluid passes over the vortex generator. 2. To simulate the vorticity contours to get the idea of the strength of the crossstream velocity vectors. 3. To determine the heat transfer characteristics in terms of combined spanwise average Nusselt number and the bulk temperature. Nusselt number is a measure of

35

the efficacy of heat transfer and the bulk temperature is a direct measure of thermal energy. 4. To compute the reduction in the length of heat exchanger for a particular bulk temperature at the exit of the exchanger. To study the effect of varying the Reynolds number on the heat transfer enhancement and the pressure drop for delta and rectangular wings. 5. To study the effect of varying the Reynolds number on the heat transfer enhancement and the pressure drop for delta and rectangular wings. 6. To simulate the effect of stamped-wing vortex generator and compare it with the results pertaining to built-in wing. 7. To compare rectangular and delta wings for the same span and chord length and also for the same area of both the wings.

36

Chapter 3 Problem Formulation 3.1 Introduction

The present work focuses on the numerical study of the flow structure and heat transfer characteristics of the fluid flowing in a plate-fin heat exchanger with delta and rectangular wing vortex generator mounted on the slant surfaces of the triangular fins. Since numerical simulation requires the problem to be defined clearly and quantitatively, this chapter dwells upon the formulation of the present problem in detail. This includes the description of the computational domain with properly generated grid, specific boundary conditions, discretization of the governing equations, method of solution etc.

3.2 Statement of the Problem

The geometry of the proposed design is shown in Figure 3.1. Two different shapes of the longitudinal vortex generator i.e. rectangular wing and delta wing are investigated for heat transfer enhancement potential in a plate-fin heat exchanger. The elaborated view shows the rectangular and delta wing vortex generators on the slant surfaces of triangular fins. These fins may also act as spacers between the parallel plates of the heat exchanger. Vortex generators are the protruded surfaces which can be made in two ways: First, a desired form of the vortex generator made of a thin slander sheet is attached to the fin surface by any of the joining techniques like welding, brazing etc, known as built-in vortex generator. In the second method, a desired shape is punched out by cutting the heat transfer surface itself and leaving a hole beneath the protruded surface, known as stamped or punched vortex generator. These protruded surfaces are kept at some angle with the direction of the fluid flow which is known as the angle of attack of the vortex generator. A two dimensional view of the geometry in the flow direction is shown in Figure 3.2. The

37

angle of the slant surfaces of the triangular fins with the horizontal is 45. A vertical plane of symmetry is imagined through the apex of every triangle as shown in the Figure 3.2.

Triangular fin with rectangular wing Triangular fin with delta wing (a) Lower base plate

Gases (b) Lower base plate Gases

Figure 3.1 Plate-fin heat exchanger (a) Rectangular wings and (b) Delta wings mounted on the triangular secondary fins These vertical planes divide the complete domain into symmetrical sections. One of the symmetrical sections ADBC is considered as the computational domain. This computational domain is rotated 45 in anticlockwise direction so that the slant surface ABAB along with the wing vortex generator becomes horizontal as shown in Figure 3.3. The rotation of the domain provides a better view of the wing and it helps to apply the boundary conditions accurately while programming. No slip plane Plane of symmetry
A D

Wing

450

450
C B

Figure 3.2 Two dimensional view of the geometry in the direction of flow

38

D ADDA, ABBA & BBCC No-slip surfaces ACCA & DBBD- Surface of symmetry QMNR -Stamped rectangular wing QPSR - Punched area L=8H

M P

Q N S R

B (a) Y Chord (c) =.7771H X Q R N Z (b) Figure 3.3 View of the channel after rotation (a) Computational domain for the stamped wing (b) Geometry of the rectangular wing A rectangular cut QPSR is made to protrude the stamped rectangular wing QMNR. The leading and trailing edges of the wing are at non-dimensional distances of 2.91 and 3.69 respectively, from the inlet for all the computations at various angles of attack. The aspect ratio of the rectangular wing is kept constant for all the angles of attack of the wing. Dimensions of the rectangular wing are shown in Figure 3.3. In a similar way the computational domain for the delta wing is shown in Figure 3.4. Here a triangular cut RMQ is made to protrude the stamped delta wing RNQ. The chord length of the delta wing is 1.06 and the trailing edge is located at 3.69 for all the angles of attack of the delta wing. Span (b) =.2121H

Aspect ratio =b/c=.2729

39

ADDA, ABBA & BBCC No-slip surfaces ACCA & DBBD- Surface of symmetry RNQ -Stamped delta wing RMQ- Punched area L=8H A D N M A B R Q

X R .565 H 1.06H N = 15 Q

Z Figure 3.4 View of the channel after rotation (a) Computational domain for the stamped wing (b) Geometry of the delta wing The computational domain as shown in Figures 3.3 & 3.4 is of trapezoidal crosssection which can be viewed as symmetrical upper triangular section and lower triangular section. In the case of built-in wing, the fluid from the upper triangular section does not mix with that in the lower triangular section. Moreover, as the boundary condition of constant wall temperature has been assumed here, the effect of wing will not be observed in the lower triangular section. Hence, the computational domain is reduced to the upper triangular section for the built-in wing as shown in Figure 3.5. Similarly upper triangular section of the computational domain is considered for the built-in delta wing (not shown in figure).

40

No-slip surfaces ADDA & ABBA D

Built-in rectangular wing MNQR

A M Q N A B R

X=3.69 Surface of symmetry BDBD

Figure 3.5 Computational domain in case of built-in wing The primary objective of the study is to compute the velocities, pressure and temperature distribution of the working fluid in the proposed geometry. The working fluid considered herein is air. The vortex generator, rectangular and delta wing is assumed to be a surface of zero thickness. Considering the thickness of the fins and wing would make it a conjugate heat transfer problem and moreover the fins are made of very slender sheets, hence the thickness of the fins and wings vortex generators are neglected. Based on the results obtained, the performance of the plate-fin heat exchanger with wings vortex generator is evaluated.

3.3 Governing Equations

The continuity equation, Navier-Stokes equations and the energy equation are the fundamental equations in the present model. These equations are reduced to the case of laminar incompressible flow by assuming to be constant, setting the divergence of velocity equal to zero and neglecting the dissipative terms in the energy equation. The number of unknowns is reduced and the need for specifying a particular system of units is eliminated by using dimensionless parameters. The velocities in all three directions are non-dimensionalized by the average incoming velocity Uav at the duct inlet as follows.

41

U=

u U av

V=

v U av

W=

w U av

The lengths are non-dimensionalized by the spacing between the plates, H i.e.
X = x H Y= y H Z= z H

2 The pressure is non- dimensionlized by U av i.e.

P=

p 2 U av

The non dimensional Reynolds number (Re) is

Re =

(U av H )

The non dimensional temperature ( ) is

(T

T ) Tw T

The non dimensional time () is

t H / U av

The governing equations are non-dimensionalized by the above parameters and in weak conservative form are written as follows. Continuity equation U V W + + =0 X Y Z X-momentum equation
U U 2 UV UW P 1 2U 2U 2U + + + = + + + X Y Z X Re X 2 Y 2 Z 2 Y-momentum equation
V VU V 2 VW P 1 2V 2V 2V + + + = + + + X Y Z Y Re X 2 Y 2 Z 2

(3.1)

(3.2)

(3.3)

Z-momentum equation W UW VW W 2 P 1 2W 2W 2W + + + = + + + X Y Z Z Re X 2 Y 2 Z 2
(3.4)

42

Energy Equation U V W 1 2 2 2 + + + = + + X Y Z Re Pr X 2 Y 2 Z 2 The dimensionless governing equations are derived in Appendix A.

(3.5)

3.4 Staggered Grid Arrangement and the Meshing of the Domain

The 3-D mesh generated in the computational domain is of Cartesian cells with staggered grid arrangement. Unlike the conventional grid, the nodes for the velocities are taken at the center of the cell faces to which the respective velocity vectors are normal, whereas, the pressure and temperature nodes are defined at the center of the cell itself, as shown in Figure 3.6. The U velocity component is determined on the faces of the cells which are normal to X-axis and similarly V and W velocity components are determined on the cell faces normal to Y and Z directions respectively. The staggered grid arrangement is used to avoid the problem of checkerboard distribution of pressure and velocities which may result when the equations are discretized by the central differencing. Another important advantage of the staggered grid is that the interpolation of velocity components is not required to compute the transport rates across the faces of the control volume, as the velocities themselves are computed on the cell faces.

V (i, j, k) U (i, j, k)

P (i, j, k) (i, j, k) X Z Figure 3.6 Staggered grid 43

W (i, j, k)

The three dimensional mesh generated in the computational domain and the staggered grid arrangement as applied on the cells is shown in Figure 3.7.The inclined planes pass through the centre of the cell faces.

D Plane of symmetry and inclined no slip plane passing through V-velocity and W-velocity nodal points. Y B D

X A B Z

V -Velocity points W-Velocity points U -Velocity points

Figure 3.7 Meshed computational domain for built-in wing

3.5 Boundary Conditions

The Navier-Stokes equations are elliptic in space and parabolic in time. Hence, the spatial boundary conditions are defined all over the computational domain and marching in time is performed. It is a steady state problem solved by false transient approach. An initial guess is assumed and the steady state solution is predicted for a finite number of time steps. The procedure used to implement these conditions numerically is described in Section 3.8.

3.5.1 Initial Boundary Conditions

The forward marching in time requires initial guess of the velocity, pressure and temperature values all over the computational domain excluding the boundaries. These guessed values are modified in further time steps by the solution algorithm and are specified as under. U = 1.0, V = 0, W = 0, P = 0 and = 0

44

3.5.2 Spatial Boundary Conditions

The flow is assumed to be axial at the entrance of the channel i.e. U = 1.0, V = 0 and W = 0 Since a smooth transition is required at the exit of the channel, the boundary condition imposed here is the same as given by Orlanski [1976]: + Uc =0 X where represents any of the dependent variables U, V, W or while Uc is the mean channel outflow velocity. The computational domain for the built-in wing or stamped wing is confined by the no-slip planes and the planes of symmetry. The computational domain for the analysis of stamped wing has three no-slip planes and two planes of symmetry as shown in Figures 3.3 and 3.4. Figure 3.5 depicts the computational domain for the case of built-in wing and shows two no-slip planes and one plane of symmetry. The velocities on the no-slip planes are zero as the fluid is viscous, i.e. U = 0, V = 0, and W = 0. It needs a mention that the plane of wing is also considered as a no-slip plane. However, in the case of stamped wing, the plane ABBA is a no-slip surface excluding the area corresponding to the rectangular or triangular hole. Since the fluid may flow through these punched areas, the no-slip boundary conditions can not be applied here. This fact is given due consideration while implementing the boundary conditions numerically in the computer program. Furthermore, the plane of symmetry passes through the V and W velocity nodes of the staggered grid. Therefore, V and W velocities are taken as zero on the plane of symmetry as a non-zero value will result in an asymmetric velocity field. The axial velocity U is considered to be symmetric across the plane of symmetry, i.e. U-component of velocity has identical values at nodal points which lie on the other side of the plane of symmetry at equal distance. Hence, V = 0, W = 0 and U is symmetric. Even though the slant surfaces of the triangular fins and the surface of the wing vortex generator are not in direct contact with the hot fluid, the uniform wall temperature on these surfaces is assumed similar to the top and bottom plates of the plate-fin heat exchanger. If the heat transfer from the parallel plates to the triangular fins or wing vortex 45

generator is considered, it would be a problem considering conjugate heat transfer. The isothermal boundary condition on these surfaces i.e. fins and wings will not be applicable. The thermal boundary condition on the no-slip walls and on the wing surface is w = 1.0. The thermal boundary condition is not implemented on the punched out portion of the horizontal no-slip plane in case of stamped wing vortex generator. The nondimensional temperature () is symmetric across the planes of symmetry.

3.6 Method of Solution

The main difficulty in solving the problem of incompressible flow is the absence of any explicit equation for the evaluation of pressure, while the spatial coupling of pressure and velocity does exist. A 3-D solution of the flow field requires a primitivevariable formulation of the incompressible Navier-Stokes equations without encountering non-physical wiggles in the pressure distribution. A modified version of Marker and Cell method due to Harlow and Welch [1965] and Hirt and Cook [1972] has been used by many researchers to solve the fundamental equations by employing staggered grid arrangement.

3.6.1 Marker and Cell Method

The solution of the momentum Equations 3.2-3.4 give explicitly a provisional value of the velocity components to be used for the next time step. However, these explicitly advanced velocity components may not yield a realistic flow field. Therefore, the continuity of flow is checked using these velocity components. Since a divergence free velocity field must exist for the present case of incompressible flow, a non-zero divergence of velocity indicates the accumulation or annihilation of some mass and hence, an incorrect pressure field. Therefore, pressure at any cell is directly linked with the value of divergence of velocity at that cell. Now, on one hand, the pressure field is corrected with the help of non-zero divergence and on the other, the velocity components are adjusted. This iterative cycle continues till a divergence free velocity field is achieved and hence, the continuity Equation 3.1 is satisfied. The boundary conditions are applied to the momentum equations after every explicit evaluation of velocities and also, for the pressure-velocity iterations. The present simulations have been carried out for an accuracy of 0.001 in the values of the velocity

46

components. After determining the velocity field, the energy Equation 3.5 is solved by the successive over-relaxation technique to determine the temperature field. For the numerical implementation of the present problem, a computer code has been developed in Visual-FORTRAN. The flow chart of the program substructure and the brief description of the indices are given in Appendix-B.

3.6.2 MAC Algorithm

The convective terms of the Navier-Stokes equations are discretized by a weighted average of second upwind and space centered scheme and the convective terms of the X-momentum Equation 3.2 in discretized form are as follows.
UU 1 = X 4X U i,j,k + U i + 1,j,k U i,j,k + U i + 1,j,k + U i,j,k + U i + 1,j,k U i,j,k U i + 1,j,k U i 1,j,k + U i,j,k U i 1,j,k + U i,j,k U i 1,j,k + U i,j,k U i 1,j,k U i,j,k

= DUUDX

(3.7)

UV 1 = Vi,j,k + Vi + 1,j,k Ui,j,k + Ui,j + 1,k + Vi,j,k + Vi + 1,j,k Ui,j,k Ui,j + 1,k Y 4Y V i,j 1,k + Vi + 1,j 1,k Ui,j 1,k + Ui,j,k Vi,j 1,k + Vi + 1,j 1,k Ui,j 1,k Ui,j,k = DUVDY (3.8)
UW 1 = Wi,j,k + Wi + 1,j,k Ui,j,k + Ui,j,k + 1 + Wi,j,k + Wi + 1,j,k Ui,j,k Ui,j,k + 1 Z 4Z W i,j,k 1 + Wi + 1,j,k 1 Ui,j,k 1 + Ui,j,k Wi,j,k 1 + Wi + 1,j,k 1 Ui,j,k 1 Ui,j,k = DUWDX
(3.9)

The diffusive terms are discretized by a central difference scheme and the diffusive terms of the X-momentum equation in discretized form are as under.

2U X 2
2U Y 2

=
=

U i + 1, j , k 2U i , j , k + U i 1, j , k X 2
U i , j + 1, k 2U i , j , k + U i , j 1, k Y 2

= D 2UDX 2
= D 2UDY 2

(3.10)

(3.11)

47

 2U Z 2

U i , j , k + 1 2U i , j , k + U i , j , k 1 X 2

= D 2UDZ 2

(3.12)

The pressure gradient in X direction is discretized as

Pi + 1, j , k Pi , j , k P = = DPDX X X

(3.13)

The quantity is an upwinding factor and with = 1, the scheme is second order upwind and if = 0, the scheme is space centered. Some factor of upwinding is incorporated to have the stability by choosing the value of between 0.2 and 0.3.
~ If the explicitly advanced predicted velocity is termed as U n + 1 , the transient Ui, j , k

velocity term of the X-momentum equation is discretized as ~ U in, +,1k U in j , k U j , =

(3.14)

Substituting the discretized Equations 3.7 to 3.14 into the X-momentum Equation 3.2 and manipulating the terms, explicitly advanced velocity component is determined as follows.

~ +1 + [SDISCU]n Uin j, k = Uin j, k (DPDX)n , , i, j, k i, j, k

Where

(3.15)

[SDISCU ]in,

1 (D2UDX 2 + D2UDY 2 + D2UDZ 2) = (DUUDX DUVDY DUWDZ ) + j, k Re (3.16)

Equation 3.15 can be written as

Pn Pin+ 1, j , k ~n + 1 i, j , k + [SDISCU ]n n U i, j , k = U i, j , k + i, j , k X
In a similar way, Equation 3.3 can be written in discretized from as below.

(3.17)

~ 1 + [SDISCV ]n Vin j+ k = Vin j , k (DPDY )n , , , i, j , k i, j , k

where

(3.18)

[SDISCV ]in,

Pi , j + 1, k Pi , j , k P = = DPDY Y Y

1 (D2VDX 2 + D2VDY 2 + D2VDZ 2) = (DVUDX DVVDY DVWDZ ) + j, k Re

48

Similarly the Z- direction momentum Equation 3.4 can be written as

~ W n +1 = W n (DPDZ )n + [SDISCW ]n i, j , k i, j , k i, j , k i, j , k
where

(3.19)

[SDISCW ]in,

Pi , j , k + 1 Pi , j , k P = = DPDZ Z Z

1 (D2WDX 2+ D2WDY 2+D2WDZ 2) = (DWUDX DVWDY DWWDZ ) + j, k Re

At this stage, the pressure distribution is not correct. Therefore, pressure in each cell is to be corrected in such a way that there is no net mass flow into or out of any cell. In the original MAC method, the Poisson equation for pressure was solved to obtain corrected pressures. Chorin [1967] employed a procedure involving simultaneous iterations of pressure and velocity components. Vicelli [1971] demonstrated that the two methods, as applied with the MAC algorithm, are equivalent. This modified MAC method by Chorin [1967] is the solution algorithm. The corrected velocity components (unknown) are related to the corrected pressures, which are system unknowns as well, in the following way:

Pn +1 Pn +1 i + 1, j , k i, j , k U n +1 = U n + i, j , k i, j , k X
Subtracting Equation 3.20 from 3.17 P P i + 1, j , k ~ n + 1 U n + 1 = i, j , k U i, j , k i, j , k X

n + [SDISCU ]i, j , k

(3.20)

(3.21)

where the pressure corrections Pi ,' j ,k are defined as

1 Pi', j, k = Pin j+ k Pin j, k , , ,

The pressure corrections and the exact velocities - both are unknown at this stage and are calculated simultaneously in an iterative process. Equation 3.21 can be written as

~ Uin + 1 = Uin + 1 + P Pi + 1, j, k , j, k , j, k X i, j, k
Similarly for V and W velocity components

(3.22)

1 1 P Vin j+ k = Vin j+ k + Pi, j + 1, k , , , , Y i, j, k

49

(3.23)

 ~ 1 1 P Pi, j, k + 1 Win j+ k = Win j+ k + , , , , Z i, j, k

]
]

(3.24)

The continuity equation is discretized by the backward differencing scheme, so it requires the velocity components at the previous nodes. The velocity components at these nodal points are written as under.

~ + + U in 11 j , k = U in 11 j, k P Pi 1, j , k , , X i, j, k

(3.25) (3.26)

~ 1 1 P Vin j+ 1, k = Vin j+ 1, k Pi, j 1, k , , Y i, j, k

]
]

~ 1 1 Win j+ k 1 = Win j+ k 1 P Pi, j, k 1 , , , , Z i, j, k

Un i, ~ Un i,

(3.27)

Now substituting the Equations 3.22 to 3.27 into the continuity equation 3.1, we get
+ 1 Un + 1 1 1 1 1 Win j+ k Win j+ k 1 Vin j+ k Vin j+ 1, k , , , , , , , i 1, j , k j, k + + = Z Y X ~n + 1 ~n + 1 ~n + 1 ~n + 1 ~n + 1 +1 U Wi, j, k Wi, j, k 1 Vi, j , k Vi, j 1, k i 1, j , k j, k + + Z Y X

(3.28)

Pi, j + 1, k 2 Pi, j, k + Pi, j 1, k Pi + 1, j, k 2 Pi, j, k + Pi 1, j, k + X 2 Y 2 Pi, j, k 2 Pi, j, k + Pi, j, k 1 + Z 2

Since the terms in the square bracket on the left hand side of the above equation is the divergence of exact velocity, it should vanish. Hence,
Pi, j + 1, k 2 Pi, j, k + Pi, j 1, k Pi + 1, j, k 2 Pi, j , k + Pi 1, j, k + X 2 Y 2 Pi, j, k + 1 2 Pi, j, k + Pi, j, k 1 + Z 2 ~ ~ ~ ~ ~ ~ 1 1 1 Un + 1 Un + 1 Vin j+ k1 Vin j+ 1, k Win j+ k Win j+ k 1 , , , , , , , i 1, j, k i, j , k = + + X Y Z (3.29)

The pressure corrections in the neighboring cells i.e. U i+1, j , k , U i1, j , k , U i, j +1, k U i, j 1, k , U i, j , k +1 , U i, j , k 1 are neglected. The right side of the above equation is the

50

divergence of the evaluated velocities and is denoted by ( Div) i , j , k . Now the Equation 3.29 can be written as under. Pi, j , k = (Div )i , j , k 1 1 1 2 + + 2 2 Y Z 2 X
(3.30)

Over-relaxation factor o accelerates the calculation of the pressure correction Equation 3.30 and a typical value of 1.7 is used as over-relaxation factor.

Pi, j , k =

(Div )i , j , k 1 1 1 + + 2 2 2 Y Z 2 X

(3.31)

After calculating the pressure correction Pi', j , k , the pressure in the each cell is corrected as
Pin j+1 Pin j , k + Pi', j , k , ,k ,
(3.32)

Now the pressure and velocity components for each cell are corrected through an iterative procedure in such a way that for the final pressure field, the divergence free velocity field with an upper bound value of 0.001 is achieved. The perfect velocity boundary conditions and a divergence free converged velocity field will eventually determine the correct pressure in all the cells including the cells at the boundary. Thus, this method avoids the application of pressure boundary conditions.

3.7 Solution of Energy Equation

After evaluating the steady state velocities, the energy equation is solved with a Successive Over-Relaxation (SOR) technique to determine the temperature field. The steady state energy equation, neglecting the dissipative term, may be written in the conservative form as

2 U V W 2 2 1 + + = + + X Y Z Re Pr X 2 Y 2 Z 2
Equation 3.33 is written as

(3.33)

2 X2

2 Y 2

2 Z 2

V W U Re Pr = + + Y Z X

(3.34)

51

The velocities U, V, and W are known from the solution of momentum equations and hence, Equation 3.34 is now a linear equation. Initial guess for the temperature values is made in the computational domain. After discretizing and evaluating the right hand side of Equation 3.34, a Poisson equation for temperature is obtained. Now, SOR technique is used to solve this equation with the right hand side being updated after each iterative sweep. The convective terms on the left hand side are discretized by the weighted average scheme as follows.

U 1 [U i, j,k + i +1, j,k + a.Ui, j,k i, j,k i +1, j,k = 2X i, j,k X Ui 1, j,k i 1, j,k + i, j,k a.Ui 1, j, k i 1, j,k i, j,k
discretized.

)
(3.35)

)]

where a is the weighted average coefficient. Similarly V Y and W Z are

The terms on the right hand side are discretized by the central differencing as under.
2 X 2 = i +1, j , k 2i , j , k + i , j 1.k

X 2

(3.36)

Similarly 2 Y 2 and 2 Z 2 are discretized.

3.8 Stability Conditions

The mesh size must be chosen small enough to resolve the expected spatial variations in all the dependent variables. Once a mesh has been chosen, the choice of the time increment is governed by two restrictions, namely, the Courant-Friedrichs-Lewy (CFL) condition and the restriction on the basis of grid-Fourier numbers. According to the CFL condition, material can not move through more than one cell in one time step. Therefore, the time increment must satisfy the inequality
X Y Z < min , , |U | |V | |W |
(3.37)

where, the minimum is with respect to every cell in the mesh. Typically, is chosen equal to one fourth to one-third of the minimum cell transit time. When the viscous diffusion terms are more important, the condition necessary to ensure stability is dictated by the restriction on the grid Fourier numbers, which results in the following equation.

52

 1 ( X ) 2 ( Y ) 2 ( Z ) 2 < 2 Re ( X ) 2 ( Y ) 2 + ( Y ) 2 ( Z ) 2 + ( Z ) 2 ( X ) 2

(3.38)

The final for each increment is the minimum of the s obtained from Equations 3.37 and 3.38.

3.9 Numerical Boundary Conditions

The boundary conditions discussed earlier need to be numerically specified to implement in a computer program. The staggered grid arrangement used by most of the researchers employ a layer of fictitious cells all around the boundary of the computational domain. These fictitious cells have been assigned the appropriate values of the velocities to implement the required conditions at the boundaries. In this study, the use of fictitious cells can not be employed while solving for the stamped wing vortex generator as one of the no-slip boundary ABBA shown in Figures 3.4 & 3.5, exists inside the computational domain. Thus the fictitious cells would be occupying a region of real flow. This difficulty is overcome by applying the boundary conditions directly on the appropriate grid cells as discussed further.

3.9.1 Boundary Conditions for the Continuity Equation

The velocity components estimated from the NavierStokes equations are to be corrected while evaluating the divergence in each cell. The divergence is used to modify the pressure values which further correct the velocity distribution. The boundary conditions on the concerned cells are applied directly in the expressions used for the calculation of divergence. The continuity equation is discretized by the backward difference and the divergence in any cell is computed by the following expression as discussed in Section 3.6.2. ~ ~ ~ ~ ~ ~ U i , j , k U i 1, j , k Vi , j , k Vi , j 1, k Wi , j , k Wi , j , k 1 + + Divi , j , k = X Y Z

(3.39)

Velocity components are taken at the centre of the cell faces to which they are normal and boundary conditions are applied numerically on all the cell faces which lie on any of the boundaries i.e. no-slip boundaries, plane of symmetry. Each and every boundary is treated separately for the determination of divergence as follows.

53

3.9.1.1 Cells Adjacent to the Horizontal No-Slip Boundary

The cell faces for the vertical velocity V nodal points lie on the horizontal surface. The velocity component V i, j-1, k is made zero for all the cells lying just above the horizontal no-slip boundary and the expression for the divergence is as follows. ~ ~ ~ ~ ~ U i , j , k U i 1, j , k Vi , j , k Wi , j , k Wi , j , k 1 + + Divi , j , k = X Y Z slip plane as shown in Figure 3.8 and for these cells V equation reduces to Divi , j , k = ~ ~ U i , j , k U i 1, j , k

(3.40)

The computational domain for the stamped wing has the cells below the horizontal noi, j, k

will be zero. The divergence

X
V i, j, k

Vi , j 1, k

~ ~ Wi , j , k Wi , j , k 1

(3.41)

Vi, j-1, k = 0

V i, j, k = 0

Horizontal no-slip boundary

Y j i k X Z

V i, j-1, k

Figure 3.8 Boundary conditions on the horizontal no-slip plane for continuity equation

3.9.1.2 Cells Adjacent to the Inclined No-Slip Boundary

The upper inclined no-slip plane as shown in Figure 3.9 passes through the cell faces on which the velocity components V
i, j, k

and W

i, j, k-1

are computed. No-slip

boundary condition is applied and the concerned velocity components are made zero. The expression for the divergence becomes ~ ~ ~ ~ U i , j , k U i 1, j , k Vi , j 1, k Wi , j , k + Divi , j , k = X Y Z

(3.42)

Similarly lower inclined no slip plane passes through the Vi,j-1,k and Wi,j,k nodal points and so these values are set zero in the divergence equation for the lower inclined no-slip plane and the divergence equation is modified as follows.

54

Divi , j , k =

~ ~ U i , j , k U i 1, j , k

~ Vi , j , k

~ Wi , j , k 1

(3.43)

V i, j, k = 0 Upper inclined no-slip plane W i, j, k-1 = 0 Y Z Lower inclined no-slip plane W i, j, k = 0 Horizontal no-slip plane

V i, j-1, k = 0

Figure 3.9 Boundary conditions on the inclined no-slip plane for continuity equation

3.9.1.3 Cells Adjacent to the Inclined Plane of Symmetry

The upper inclined plane of symmetry passes through the V
i, j, k

and W

i, j, k

velocity components nodal points and the velocity components Vi,j-1,k and Wi,j,k-1 lie on the lower plane of symmetry as shown in Figure 3.10, therefore, these velocity components are set equal to zero in the divergence equation as a non-zero value will result in an asymmetric velocity field. The U-velocity nodal points which are required in the solution of divergence equation do not lie on the plane of symmetry; hence these are not modified in the divergence equation. The divergence equation for the cells adjacent to the upper plane of symmetry becomes Divi , j , k = ~ ~ U i , j , k U i 1, j , k ~ Vi , j 1, k ~ Wi , j , k 1

(3.44)

And the divergence equation for the cells adjacent to the lower plane of symmetry is as under. Divi , j , k = ~ ~ U i , j , k U i 1, j , k ~ Vi , j , k ~ Wi , j , k

(3.45)

55

Vi, j, k = 0 Upper plane of symmetry Horizontal no-slip plane W i, j, k = 0

Y Lower plane of symmetry W i, j, k-1 = 0 Vi, j-1, k = 0 j i k

X Z

Figure 3.10 Boundary conditions on the plane of symmetry for continuity equation

3.9.2 Velocity Boundary Conditions for N-S Equations

To solve the discretized form of NavierStokes equations, the algebraic calculation of the following velocity components at various locations are required. Ui-1,j,k , U i, ,j, k , Ui+1,j,k , Ui,j-1,k , Ui,j+1,,k , Ui,j,k-1 , Ui,j,k+1 , Ui-1,j,k+1 , Ui-1,j+1,,k , Vi,j,k , Vi+1,j,k , Vi-1,j,k , Vi,j-1,k , Vi,j+1,k , Vi+1,j-1,k , Vi,j,k-1 , Vi,j,k+1 , Vi,j-1,k+1 , W Wi,j,k-1 , Wi+1,j,k-1 , Wi,j+1,k , Wi,j+1,k-1 , Wi,j-1,k. The velocity components which lie on or outside any of the boundary surfaces are identified. The appropriate values of these velocity components are then specified either directly or in terms of the velocity of an interior cell. The velocity components which lie directly on the boundaries are set equal to zero and those outside the computational domain are treated as follows.
i, j ,k

, Wi+1,j,k ,Wi-1,j,k , Wi,j,k+1 ,

3.9.2.1 Cells Adjacent to the Horizontal No-Slip Boundary

The cell faces for the V-velocity nodal points lie on this boundary and are assigned zero value directly in the computer program. Thus the velocity components Vi,j-1,k , Vi,j-1,k+1 are set equal to zero and the U and W velocity components of the cells below the horizontal plane are set equal and opposite to the respective velocity components just above the horizontal no-slip plane.

56

Hence Ui,j-1,k = - U i, j, k and Wi,j-1,k = - W i ,j, k Implementing this condition for the cells above the horizontal plane satisfies the no-slip boundary conditions on the horizontal plane. Cells above the horizontal no-slip boundary W i, j, k V i, j-1, k+1 = 0 Y U i, j+1, k V i, j, k = 0 W i, j+1, k V i, j, k+1 = 0

U i, j, k V i, j-1, k = 0

U i, j-1, k

W i, j-1, k

U i, j, k

W i, j, k Z

Cells below the horizontal no-slip boundary Figure 3.11 Boundary conditions on the horizontal no-slip plane for N-S equations Similarly for the cells lying just below the horizontal no-slip plane, vertical velocity components V horizontal plane. Ui,j+1,k = - U i , j, k and Wi,j+1,k = - W i, j, k , Wi,j+1,k-1 = - W i, j, k-1
i, j, k,

i, j, k+1

are set equal to zero and the U and W velocity

components are manipulated as under to satisfy the no-slip boundary condition on the

3.9.2.2 Cells Adjacent to the Inclined No-Slip Boundary

Inclined no-slip planes pass through the nodal points of V and W velocity components as shown in Figure 3.12. For the cells adjacent to the upper inclined no-slip plane V
i, j, k,

i, j, k-1,

Wi,j+1,k velocity components are set zero directly and by setting

U i,j,k-1 = - U i, j, k and U i,j+1,k = - U i ,j , k, the condition U = 0 satisfies on the upper inclined no-slip plane. Similarly the velocity components Vi,j-1,k , Vi,j,k+1 are set equal to zero and by implementing U i,j,k+1 = - U i, j, k and U i,j-1,k = - U i, j, k, the condition U = 0 satisfies on the lower inclined no-slip plane.

57

Upper inclined no- slip plane

DD

U i, j+1, k V i, j, k = 0 U i, j, k-1 U i, j, k W i, j, k-1 = 0 W i, j+1, k = 0

Y AA Z

Horizontal no-slip plane

BB

V i, j, k+1 = 0 U i, j, k V i, j-1, k = 0 U i, j, k+1 W i, j, k = 0

U i, j-1, k

Lower inclined no-slip boundary

Figure 3.12 Boundary conditions on the inclined no-slip plane for N-S equations

3.9.2.3 Cells Adjacent to the Inclined Plane of Symmetry

On both the upper and lower inclined planes of symmetry, V and W velocity components are set zero directly, as the inclined planes of symmetry pass through these nodal points as shown in Figure 3.13. Therefore, V
i, j, k,

i, j-1, k+1

and Wi,j+1,k-1 , W V
i, j, k-1

i, j ,k

velocity components for the upper plane of symmetry and V

i, j-1, k,

and Wi,j,k-1 ,

Wi,j-1,k for the lower plane of symmetry are set equal to zero. The U-velocity components across this plane are symmetric. And for the upper inclined plane of symmetry, the condition of symmetry is enforced by setting the velocities. U i , j , k +1 = U i , j.k + U i , j 1, k +1

2
U i , j.k + U i , j +1, k 1

And
U i , j +1, k =

58

DD

Upper plane of symmetry

U i, j+1, k U i, j+1, k-1

(U i, j, k + U i, j+1, k-1) / 2 U i, j, k (U i, j, k + U i, j-1, k+1) / 2

Y

U i, j, k+1

Horizontal no-slip plane

AA Z

U i, j-1, k+1
BB

U i, j+1, k-1

(U i, j, k + U i, j+1, k-1) / 2

U i, j, k-1

U i, j, k

(U i, j, k + U i, j-1, k+1) / 2

U i, j-1, k

U i, j-1, k+1
CC

Lower plane of symmetry

Figure 3.13 Boundary conditions on the plane of symmetry for N-S equations Similarly for the lower inclined plane of symmetry
U i , j , k 1 = U i , j 1, k = U i , j.k + U i , j +1, k 1

2
U i , j.k + U i , j 1, k +1

3.9.3 Thermal Boundary Conditions for Energy Equation

The temperature nodal points are considered at the centre of the cells and any of the confined planes do not pass through these locations. Hence, the thermal boundary conditions can not be specified directly on any plane. To solve the discretized form of the energy equation, the algebraic calculation of the temperature components at the following locations are required.
i , j, k , i+1 , j, k , i-1 , j, k , i , j+1, k , i , j-1, k , i , j, k+1 , i , j, k-1

59

The temperature nodal points lying outside the computational domain are identified and then specified in such a way that the required thermal boundary condition is satisfied on these planes.

3.9.3.1 Cells Adjacent to the Horizontal Boundary

The isothermal boundary condition i.e. w = 1.0 is enforced by assuming the linear variation of temperature across the plane. For the cells above the horizontal surface, isothermal boundary condition is satisfied by the following relation.
i , j 1, k + i , j , k

= w = 1.0

i , j 1, k = 2 i , j , k

Cells above the horizontal surface i, j, k W = 1.0 i, j-1, k i, j, k i, j+1, k

Cells below the horizontal surface Figure 3.14 Isothermal boundary conditions on the horizontal surface

Similarly for the cells below the horizontal no-slip plane

i , j +1, k = 2 i , j , k

3.9.3.2 Cells Adjacent to the Inclined No-Slip Boundary

The temperature nodal points i,j,k-1 and i,j+1,k are located outside the upper inclined no-slip plane (Figure 3.15) and the thermal boundary condition =1 is employed by specifying the temperature as follows.
i , j , k 1 = 2 i , j , k

And
i , j + 1, k = 2 i , j , k

60

The computational domain has the cells below the horizontal no-slip plane in case of stamped wing and the temperature nodal points i,j,k-1 and i,j+1,k are located outside the lower inclined no-slip plane. In a similar way, the thermal boundary condition = 1 is employed by specifying
i , j , k + 1 = 2 i , j , k i , j 1, k = 2 i , j , k

Upper inclined no slip plane = 1.0 i, j+1, k

i, j, k-1 i, j, k

Y Z

i, j, k i, j, k+1 Lower inclined no slip plane = 1.0

i, j-1, k

Figure 3.15 Isothermal boundary conditions on the inclined no-slip surface

3.9.3.3 Cells Adjacent to the Inclined Plane of Symmetry

For the upper plane of symmetry i,j+1,k and i,j,k+1 lies outside the upper plane of symmetry (Figure 3.16) and it must be specified in terms of the temperatures at the interior points. The temperatures across this plane are symmetric. The interpolation is used to obtain the temperature distribution at the interior nodal points in the

61

computational domain. These nodal points are equidistant across the plane of symmetry and are computed as under. Hence
i , j +1, k = i , j , k +1 = i , j , k + i , j +1, k 1

2
i , j , k + i , j 1, k +1

Plane of symmetry i, j+1, k ( i, j, k + i, j+1, k-1) / 2 i, j, k ( i, j, k + i, j+1, k+1) / 2 i, j, k+1

i, j-1, k+1

Figure 3.16 Isothermal boundary conditions on the plane of symmetry Similarly for the lower inclined plane of symmetry, the temperature nodal points lying outside the domain are i,j-1,k and temperatures at the interior points as under.
i,j,k-1 and are specified in terms of the

i , j 1, k =
i , j , k 1 =

i, j , k + i , j 1, k +1 2
i , j , k + i , j +1, k 1

3.9.4 Boundary Conditions for the Vortex Generator

Rectangular / Delta wing vortex generator are considered as a plane of no-slip isothermal surface. In this study, the angle of attack has been varied form 15 to 26. The dimensions of the staggered grid are chosen in such a way that there is no need to redefine the numerical boundary conditions for any angle of attack of the wing. The relation used to compute the grid size is tan = Y X . Figure 3.17 depicts the view of the wing vortex generator in X-Y plane. The plane of the wing, shown as an inclined line, passes 62

through centre of the cell faces. Here Y is kept constant and X varies according to the angle of attack of the wing. In other words, the angle of the wing vortex generator is varied by changing the dimensions of X accordingly. Figure 3.17 also shows that plane of the wing passes through U and V velocity nodal points. The plane of the wing vortex generator is considered to be a no-slip plane and hence all these velocities are numerically set to zero. The W velocity nodal points do not lie on the plane of the wing and the no-slip condition for the W velocity components on this plane is managed by interpolating the values and setting them equal to zero. Hence the W velocity components on one side of the plane are set equal and opposite to the other side of the wing plane. Cells above the wing plane V (i, j, k) Y W i-1,j, k Y W i, j, k X + Location for W U (i, j, k)

W i,j-1, k

W i,j+1, k W i, j, k W i+1, j, k

Y X X Cells below the wing plane

Figure 3.17 Side view of the wing vortex generator For the cells just below the wing, by setting Wi+1,j,k = -W i, j, k and Wi,j+1,k = -W i, j, k , the required no-slip condition on the wing plane is achieved. Similarly for the cells just above the wing plane, by setting Wi-1,j,k = -W
i ,j, k

and Wi,j-1,k = -W

i ,j, k

, the condition

W = 0 on the wing surface is implemented. These conditions are extended in Z direction according to the shape of the wing. These boundary conditions are numerically applied in Navier-Stokes equations at the location of the wing vortex generator. Figures 3.18 and 3.19 shows the three dimensional view of the delta and rectangular wing vortex generator for the proper understanding of the boundary conditions. 63

Y -V- Velocity points -U-Velocity points X X

Y Y Z X

Figure 3.18 Velocity nodal points on the delta wing plane

Y -V- Velocity points -U-Velocity points X X

Y Y Z X

Figure 3.19 Velocity nodal points on the rectangular wing plane Isothermal boundary condition is applied to energy equation for the wing surface. The wing surface does not pass through the temperature nodal points as shown in 64

Figure 3.20. The temperature nodal points

i, j+1, k

and i, j, k+1 lie above the wing surface

and here the isothermal boundary condition on the wing plane is obtained by defining these values in terms of the values of the temperature below the wing. Isothermal condition for the cells below the plane of the wing is enforced by setting
i +1, j , k = 2 i , j , k i , j + 1, k = 2 i , j , k

Similarly for the cells just above the wing plane

i 1, j , k = 2 i , j , k i , j 1, k = 2 i , j , k

W = 1.0

Cells above the wing plane

i-1,j,k Y i, j-1, k

i, j, k

i, j +1, k i, j, k i +1, j, k

Y X X Cell below the wing plane

Figure 3.20 Thermal boundary conditions on the wing plane

3.10 Comparison of Results Based on Model Problem

The discretization scheme used for computation was tested on a model problem of flow in a lid-driven square cavity. The computational domain consists of a two dimensional lid-driven square cavity with no-slip and impervious boundary conditions at the bottom and side walls except at the top, where U is a non zero constant and equal to ULID. Here ULID is the velocity at the top surface and taken equal to 1.0. In the governing equations the velocities have been non-dimensionlized with respect to ULID. Figure 3.16 shows the variation of U velocity along the vertical mid plane of the square

65

cavity and the Figure 3.17 shows the variation of V velocity along the horizontal midplane of the square cavity for a Reynolds number of 100. The results are compared with the computational results of Ghia et al. [1982]. It may be mentioned that Ghia et al. adopted a 129129 mesh with multigrid technique. However, the present computation on a 5151 grid shows good agreement with the results of Ghia et al. as seen in Figures 3.21 and 3.22. The minimum U-velocity position and the variation of U-velocity match well. The variation of V velocity is also in good agreement with Ghia et al. [1982].
1 0.9 0.8

0.6 0.5 0.4 0.3 0.2 0.1 0 -0.4 -0.2 0

Vertical mid plane

0.7

Ghia et al. [1982] Present Scheme

0.2

0.4

0.6

0.8

U velocity

Figure 3.21 Variation of U-velocity along the vertical mid plane for the lid driven flow in a square cavity The validity of the code is further established by comparing the combined spanwise average Nusselt number predicted by the present scheme with that of Biswas et
al. [1994(b)] as shown in Figure 3.23. It is a rectangular channel without any vortex

generator and has developed profile at the inlet. The computations are performed at Reynolds number 500, Prandtl number 0.7 and channel aspect ratio equal to 2.0. The results computed by the present scheme are in good agreement with the published numerical results of Biswas et al. [1994(b)].

66

0.2 0.15 0.1 0.05

Ghia et al. [1982] Present Scheme

V-Velocity

0 0 -0.05 -0.1 -0.15 -0.2 -0.25 -0.3 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Horizonatl Mid-Plane

Figure 3.22 Variation of V-velocity along the horizontal mid plane for the lid driven flow in a square cavity

30

Re =500, Pr =0.7
25 20

Nu sa

15 10 5 0 0 1 2 3 4

Biswas et al. [1994(b)] Present computation

Non-Dimensioanl Length X

Figure 3.23 Combined spanwise average Nusselt number in a rectangular channel

67

The ratio of the mean-Colburn factor to the apparent friction-factor (j/f) is also determined for a rectangular channel. The Reynolds number of the flow is 1580, channel aspect ratio is 3 and air is considered as the working fluid. The j/f ratio is compared (Figure 3.19) with the published results of Biswas et al. [1996] and found in good agreement especially after the non dimensional length of 3.

Biswas et al. [1996] Present computation

j/f

0 0 2 4 6 8 10

Non-Dimensional Length X

Figure 3.24 Distribution of j / f in a rectangular channel

3.11 Spatial Grid Independence

An effort is undertaken to obtain grid independent results. For Reynolds number = 100 in a channel with a built-in rectangular wing at an angle of attack of 20, reasonable agreement between the results obtained for 2041 and 3061 cross-stream grids is found. In the forgoing specification of cross-stream grids, 20 and 30 refer to the number of grids in the Y-direction. Similarly, 41 and 61 refer to the number of grids in the Z-direction. The size of the cell in X-direction depends upon the angle of attack and the cell size in Ydirection. Hence, it may be mentioned that for a 2041 cross stream grids, 82 grid nodes and for 3061 cross-stream grids, 123 grid nodes are taken in X-direction for a nondimensional length of 8. A mesh of 204182 nodes produces combined spanwise average Nusselt number which differs from the extrapolated grid-independent average

68

Nusselt number by less than 2% (Figure 3.25). Hence, the computations are done by using a 2041 cross-stream grid.

20 18 16 14

Reynolds number 100, Angle of attack 20 Extrapolated independent grid Present grid

Nu sa

12 10 8 6 4 2 0 0 1 2 3 4

Non-Dimensional Length X

Figure 3.25 Grid independence test

69

Chapter 4 Performance of the Rectangular Wing 4.1 Introduction

Three dimensional velocity distributions and pressure fields are obtained in a plate-triangular fin channel with built-in and stamped rectangular wing vortex generator mounted on the slant surfaces of the triangular fins. Temperature field in the computational domain is obtained by further solving the energy equation. However, the performance of the rectangular wing is expressed in terms of the parameters in which the heat exchanger designers are primarily interested. This chapter illustrates the procedure used for evaluating the performance parameters such as combined spanwise average Nusselt number, bulk temperature, static pressure loss etc. These parameters are compared with the values in a channel without any type of vortex generator to quantify the effect of rectangular wing vortex generator. The laminar flow considered in this study is not just for the computational simplification. It is rather necessary to assume as the fin spacing is usually very less and the mean velocity range is such that the flow considered is often laminar. Therefore, the Reynolds number of the flow is limited to 200 in order to maintain laminar flow.

4.2 Streamwise Velocity Vectors by Rectangular Wing

The physics of the flow over a rectangular wing vortex generator in a channel is explained by the flow structure. The incoming fluid strikes the rectangular wing vortex generator and travels from both the side edges towards the rear side of the wing where the pressure is lower. It creates two main longitudinal vortices, one rotating counterclockwise on the right side and the other rotating in clockwise direction on the left side. The velocity vectors in YZ plane are plotted at different axial locations to represent the secondary flow. The cross-stream velocity vectors at various axial locations along the

70

built-in wing for Reynolds number 100 and an angle of attack of 20 are shown in Figure 4.1.The symbol X represents the non-dimensional distance from the inlet of the channel.

X=3.69

F AIN

W LO

X=3.39

X=3.10 D

X=2.91

Figure 4.1 Streamwise velocity vectors along the built-in rectangular wing At X=2.91, the fluid just strikes the leading edge of the wing and the rotation of the fluid particles takes place. After striking the wing, the bulk of the fluid moves downwards. The fluid rolls up forming vortices which are finally swept around the wing and carried downstream. Further axial locations along the wing clearly depict two main vortices, one rotating in clockwise and the other rotating in anti-clockwise direction. The horizontal line formed by the velocity vectors represents the location of intersection of fluid at the wing surface. In the downstream, this intersection of the fluid with the wing surface take place towards the horizontal surface and at the trailing edge of the wing i.e. X= 3.69, the fluid strikes the junction of the wing and fin surface and then rolls up. Figure 4.2 shows the cross-stream velocity vectors at various axial locations beyond the built-in wing for Reynolds number 100 and an angle of attack of 20. This figure clearly shows the weakening of the vortices along the channel length beyond the vortex generator even though the two main vortices persist for a long distance along the flow.

71

X=5.05

OW FL AIN

X=4.66

X=4.27 D

X=3.98

Figure 4.2 Streamwise velocity vectors beyond the built-in rectangular wing In the case of stamped wing, some of the fluid gets entrained through the hole beneath the wing, thereby reducing the strength of the cross-stream velocity vectors. The velocity vectors for the stamped rectangular wing at Reynolds number 100 and = 20 is shown in Figures 4.3(a) and 4.3(b) and it is observed that the vortices in the upper half of the domain produces weak vortices in the lower triangular section. At the location, X = 3.69, where the wing is attached to the fin surface, the velocity vectors are clearly seen to pass through the punched area and generating weak vortices in the lower triangular section. An increase in Reynolds number and angle of attack of the wing is found to increase the strength of the cross-stream velocity vectors. Figure 4.4 shows the velocity field at various sections normal to the channel axis for the case of flow through a plate-fin channel without any vortex generator, for a Reynolds number of 20. It is observed that at sections next to the inlet, fluid moves away from the no-slip boundaries AB and AD towards the channel axis of symmetry BD. The reason behind this flow is the assumption of uniform U-velocity profile at the inlet which gets modified along the length and the fully developed U-velocity profile is roughly parabolic with the maximum value occurring at the centre.

72

w Flo ain M

X=3.10

X=2.91

Figure 4.3(a) Streamwise velocity vectors along the stamped rectangular wing

w Flo ain M

X=3.69

X=3.39

Velocity vectors passing through the punched area

Figure 4.3(b) Streamwise velocity vectors along the stamped rectangular wing

73

For this change in the U-velocity profile, from a constant profile at the inlet to a parabolic profile at a downstream section, mass must shift away from the walls of the channel towards its axis of symmetry BD which is shown in Figure 4.4.

IN MA

OW FL

X=0.39

X=0.26

Figure 4.4 Streamwise velocity vectors along the channel without vortex generator

4.3 Vorticity Contours along the Rectangular Wing

In order to understand the vortex structures along the wing, the constant vorticity contours have been drawn. The vorticity is a measure of the rotation of the fluid particles which provides the overview of the strength of the vortices. The vorticity in x-direction is given by the relation-

x =

w v y z
x H U av

The non-dimensional relation for the vorticity is as under.

X =

(4.1)

Figure 4.5 compares the vorticity contours in a channel for the two different angles of attack- 20 and 26 of the built-in wing at the same location X= 3.10. Counter rotating vortices are represented by the negative sign. Peak vorticity is observed in the centre of the vortices and it is higher for larger angles of attack of the wing. The values of the vorticity contours are higher for the angle of attack of 26 as compared to that for the = 20 at the same location and same Reynolds number.

74

Angle of attack 20
0.7 0.6 0.5 0.4 0.3 0.2 0.2 0.1 0.2 0.2 5.2 0.2 -4.8 -4.8 -4.8-9.8 -4.8

X = 3.10

0.2 15.2 5.2 10.2 5.2 0.4 0.6 0.2 -4.8

-9.8

-4.8

0.8

1.0

1.2

1.4

Z (a)

Angle of attack 26
0.7 0.6 0.5 0.4 -4.8 -9.8 -4.8 -9.8

X = 3.10

0.2 -9.8 -14.8 5.2 5.2 0.2 20.2 15.2 10.2 5.2 0.2 0.2 0.4 0.6 0.8 -9.8 -4.8 0.2 -4.8 -9.8 -9.8 -14.8

Y
0.3 0.2 0.1

-4.8 -9.8 5.2 0.2

-4.8 1.0 1.2 1.4

Z (b)

Figure 4.5 Vorticity contours for the built-in rectangular wing at X = 3.10 (a) Angle of attack 20 (b) Angle of attack 26

75

The vorticity contours does not show the symmetry between the two main vortices rotating in opposite direction. The reasons are (1) the plane of symmetry does not impose any viscous resistance to the counter-rotating vortex (2) The velocity of the flow is higher along the axis of symmetry. The inclined no-slip plane directs the fluid towards the low pressure side in the downstream, which is coming after striking the rectangular wing, thus the strength of the vortices along the no-slip surface is higher. Figure 4.5 (b) and Figure 4.6 reveal that not only the peak vorticity of the vortex, but also the strength of the vorticity around the core of the vortex is increased along the length of the wing. Peak vorticity of 25.2 is observed in the clockwise rotating vortices for an angle of attack of 26 and at location X = 3.39. Figure 4.7 shows the vorticity contours after the built-in wing at an angle of attack of 26 and reveals that the one of the main vortex rotating in the clockwise direction along the inclined no-slip plane loses its strength sharply beyond the wing vortex generator. Elliptical deformation of the vortices also takes place due to the reduction in the strength of the vortices which is brought about by the viscous resistance of the no-slip surfaces. At higher Reynolds numbers, the vortices generated are much stronger and hence the contours for the vorticity have greater magnitudes.

o Angle of attack 26
0.7 0.6

X =3.39
0.5 -4.8 0.4 -9.8 0.2 0.3 0.2 0.1 0.2 0.2 0.2 5.2 5.2 10.2 15.2 5.2 10.2 0.4 -4.8 -9.8 10.2 15.2 -4.8 -9.8 -14.8

0.2 0.2 -4.8 25.2 -4.8 20.2 5.2 -14.8 -4.8 -9.8 0.6 0.8 1.0 1.2 1.4

Figure 4.6 Vorticity contours for the built-in rectangular wing at X = 3.39 and Angle of attack 26

76

Angle of attack 26
0.7 0.6 0.5 0.4 -0.8 -0.8 0.2 0.2 -1.8 0.1 0.2 0.2 0.4 1.2 0.6 -0.8 0.8

X = 4.66

-0.8 -1.8

0.3 0.2

-1.8 -0.8 -2.8

1.0

1.2

1.4

Figure 4.7 Vorticity contours for the built in rectangular wing at X = 4.66 and Angle of attack 26

4.4 Heat Transfer Performance

The efficacy of heat transfer is characterized by the Nusselt number which is a dimensionless representation of the heat transfer coefficient. It is defined as the ratio of convective conductance h to the thermal conductance k/L.
Nu = h (k / L )
(4.2)

The characteristic dimension L in the present problem is considered equal to H i.e. the vertical spacing between the two plates of the heat exchanger. The rate of heat transfer per unit surface area is given by
T q = k = h(Tw Tb ) y y =0

(4.3)

where Tw is the wall temperature and Tb is the bulk temperature of the flowing fluid. From these two Equations 4.2 and 4.3, it can be written
T H Nu = T T y w b

(4.4)

77

Non-Dimensional form of the

T T T = w H y

T is y
(4.5)

where T is the temperature at the inlet. Now the non-dimensional local Nusselt number becomes
Nu = Y b w
(4.6)

Instead of determining the local Nusselt number, combined spanwise average Nusselt number is computed by taking the average of local Nusselt number all around the periphery. The combined spanwise average Nusselt number is given by the relation as under.
qdP H P = kP (Tb ( x ) Tw (x ))

Nu sa

(04.7)

where P is the perimeter of the channel cross-section and dP is a line element along the periphery. The second parameter for the heat transfer analysis is the bulk temperature which is directly a measure of thermal energy. The cold fluid travels in the exchanger and takes heat from the walls of the exchanger. Therefore, the mean or bulk temperature of the fluid increases along the length. The bulk temperature relation is given by
U dA c A c b (x ) = U d Ac
Ac

(4.8)

The bulk temperature is numerically computed by the following relation.

b (x ) =

( U ) ( U )

(4.9)

where, the summation is done along the cross section of the channel.

78

4.5 Performance of the Built-in Rectangular Wing

The aspect ratio () of the rectangular wing is kept constant at 0.2729 for the present analysis; however, the angles of attack and the Reynolds numbers are varied. Heat transfer characteristics of the channel are predicted by the bulk temperature and the combined spanwise average Nusselt number as discussed earlier. Figure 4.8 shows the axial growth of bulk temperature for the three angles of attack of the built-in-wing i.e. 15, 20, 26 while the fluid is flowing at Reynolds number 100. The bulk temperature of the fluid flowing in a plate-fin channel even without any vortex generator increases along the channel length due to the transfer of heat from the heated fin surfaces to the cold fluid. This transfer of heat is further enhanced by the use of rectangular wing and Figure 4.8 depicts the sharp increase in the bulk temperature of the fluid along the wing vortex generator. The boundary layer of the fluid gets disturbed by the secondary flow of the fluid. The hot fluid near the fin surfaces mixes with relatively colder fluid flowing in the core region, which causes an increase in bulk temperature.

0.95 0.9 0.85

Reynolds Number 100

Bulk Temperature

0.8 0.75 0.7 0.65 0.6 0.55 0.5 0.45 2 3 4 5 6 7 8

Plate-fin channel without wing Wing without stamping =15 Wing without stamping =20 Wing without stamping =26

Non Dimensional Length X

Figure 4.8 Bulk temperature distributions for various angles of attack of the built-in rectangular wing at Reynolds number 100

79

The bulk temperature at the trailing edge of the wing vortex generator at an angle of attack of 26 is 14.79 % higher than that for the plate-fin channel without wing vortex generator. The bulk temperature increases with an increase in the angle of attack of the wing as the strength of the secondary flow generated by the wing increases with the angle of attack. On the basis of the required bulk temperature at the exit of the channel, an engineer may design a more compact heat exchanger by using the vortex generators. Table 4.1 shows the compactness achieved in terms of the reduction in length of the channel to attain the exit bulk temperature of 0.80. In a plate-fin channel, the bulk temperature 0.80 is achieved at a non-dimensional length of 6.15 while the same bulk temperature in a plate-fin channel with rectangular wing vortex generator at an attack angle of 26 can be obtained at a length of 4.71 which is about 23% shorter than the required length without using any vortex generator. Table 4.1 Percentage reduction in the length of the channel using rectangular wing Geometry Non-Dimensional Length of the Exchanger Plane Fin Channel Rectangular Wing =15 Rectangular Wing =20 Rectangular Wing =26 6.15 5.80 5.32 4.71 Percentage Reduction in the Length of the Exchanger __ 5.69 13.49 23.41

To the downstream of the channel, the heat transfer from the isothermal walls decreases continuously due to increase in the temperature of the flowing fluid. Further, it is found that the thermal boundary layer decreases the temperature gradient near the isothermal surfaces and hence, the combined spanwise average Nusselt number decreases. Along the length for X< 2.0, the temperature difference between the incoming cold fluid and the walls of the exchanger is high; therefore, the combined spanwise average Nusselt number is very high in this region. Since the thermal boundary layer is also thin in this zone, the thermal effect of rectangular wing is mainly observed downstream of the wing vortex generator. The wing vortex generator churns and mixes the fluid near the fin surface with the comparatively colder fluid in the core region. Hence, the temperature of the fluid adjacent to the fin surface decreases. Therefore, the temperature gradient near

80

the walls increases and consequently, the combined spanwise average Nusselt number also increases. The combined spanwise average Nusselt number at various angles of attack of the wing and Reynolds number 100 is shown in Figure 4.9. It can be clearly seen from this figure that the combined spanwise average Nusselt number increases steeply along the wing vortex generator accompanied with a sharp decrease in the combined spanwise average Nusselt number near the trailing edge of the wing. This decrease in combined spanwise average Nusselt number ( Nu sa ) is due to the wake region generated near the trailing edge. In this wake region, temperature of the fluid increases which decreases the temperature gradient. Therefore, decrease in the combined spanwise average Nusselt number is observed in this region.

13

Reynolds Number 100

12

11

Nu sa

10

Plate-fin channel without wing Wing without stamping =15 Wing without stamping =20 Wing without stamping =26

6 0 1 2 3 4 5 6 7 8 9

Non Dimensional Length X

Figure 4.9 Combined spanwise average Nusselt number distributions for various angles of attack of the built-in rectangular wing at Reynolds number 100 The combined spanwise average Nusselt number increases further as the longitudinal vortices prolongs farther downstream. It is interesting to note that even at the exit of the channel; the combined spanwise average Nusselt number is higher than its value for the case of the plate-fin channel without wing. Increasing the angle of attack of the wing further increases the combined spanwise average Nusselt number. The combined 81

spanwise average Nusselt number for the rectangular wing at an angle of attack of = 26 and Reynolds number 100 is 35.58% higher than that of the plain plate-fin channel at location X= 3.18. The bulk temperature variation in a channel at various angles of attack of the wing and Reynolds number 200 is shown in Figure 4.10. Similar trends are obtained for Re=200; however, with lower magnitudes of bulk temperature. At higher Reynolds numbers, more fluid passes through the channel in the same interval which causes a reduction in the value of bulk temperature. The bulk temperature for the built-in wing at an angle of attack of 20 is compared for the Reynolds number 100 and 200 as shown in Figure 4.11. A considerable difference in the bulk temperature is observed with the increase in Reynolds number. In a plate-fin channel without any vortex generator at Reynolds number 200, the exit temperature achieved is 0.688 which is 19.22% less than the bulk temperature obtained in the same channel at Reynolds number 100.

0.9 0.8

Reynolds Number 200

Bulk Temperature

0.7 0.6 0.5 0.4 0.3 2 3 4 5 6 7 8

Plate-fin channel without wing Wing without stamping =15 Wing without stamping =20 Wing without stamping =26

Non Dimensional length X

Figure 4.10 Bulk temperature distributions for various angles of attack of the built-in rectangular wing at Reynolds number 200

82

Using a rectangular wing at an angle of attack of 26, an increment of 17% is observed at the exit for the higher Reynolds number. The combined spanwise average Nusselt number distribution in a channel at various angles of attack of the wing and Reynolds number 200 is shown in Figure 4.12. Increase in Reynolds number implies the increase of mass flow rate in the channel. This increased mass flow rate decreases the mean temperature of the fluid as the heat from the isothermal walls is now transferring to more amount of fluid. The extra amount of fluid flowing in the channel takes more heat from the channel surfaces, thereby increasing the combined spanwise average Nusselt number. Also, the use of rectangular wing vortex generator with greater mass flow rate produces vortices of higher strength which ensures better heat removal rate. The combined spanwise average Nusselt number for the rectangular wing at an angle of attack of 26 is about 54% higher at X= 3.334 than that for the plate-fin channel without wing. It may be noted that a considerable enhancement in combined spanwise average Nusselt number is observed even beyond the wing.

0.9 0.8

Bulk Temperature

0.7 0.6 0.5 0.4 0.3 2 3 4 5 6 7 8

Plate-fin channel without wing, Re=100 Wing without stamping =20, Re=100 Plate-fin channel without wing, Re=200 Wing without stamping =20, Re=200

Non Dimensional length X

Figure 4.11 Bulk temperature distributions for the built-in rectangular wing at Reynolds number 100 and 200

83

Figure 4.13 compares the combined spanwise average Nusselt number obtained in a channel with rectangular wing vortex generator at an attack angle of 20 for the Reynolds number 100 and 200. The peak value of the combined spanwise average Nusselt number achieved at Reynolds number 200 is 25.59 % higher than the same at a Reynolds number of 100. After taking a plunge, the combined spanwise average Nusselt number in the channel with rectangular wing vortex generator and for the Reynolds number 200 increases by about 38% than that for the channel without wing, whereas, at Reynolds number 100, the combined spanwise average Nusselt number rises after the plunge but it is even lesser than that for the channel without wing. Therefore, increasing the Reynolds number is appreciated when the heat transfer between the fluids is important consideration instead of the change in temperature of the fluids.

16 15 14

Reynolds Number 200

Plate-fin channel without wing

Nu sa
13 12 11 10 9 8 7 0 1 2 3 4 5 6 7 8 9

Wing without stamping =15 Wing without stamping =20 Wing without stamping =26

Non Dimensional Length X

Figure 4.12 Combined spanwise average Nusselt number distributions for various angles of attack of the built-in rectangular wing at Reynolds number 200

84

15 14 13 12 11

Nu sa

Plate-fin channel without wing, Re=100 Wing without stampng =20, Re=100 Plate-fin channel without wing, Re=200 Wing without stamping =20, Re=200

10 9 8 7 6 0 1 2 3 4 5 6 7 8 9

Non Dimensional Length X

Figure 4.13 Combined spanwise average Nusselt number distributions for the built-in rectangular wing at Reynolds number 100 and 200

4.6 Performance of the Stamped Rectangular Wing

As described earlier, the easiest way to manufacture wing vortex generator on a surface is to punch the desired shape (rectangular or delta) out of the surface itself. This punched portion of the surface is then pulled out till the required angle of the vortex generator is achieved. This way of making a vortex generator leaves a hole underneath the vortex generator. Figure 4.14 shows the bulk temperature enhancement by varying the angles of attack of the stamped rectangular wing in a channel at Reynolds number 100. Along the stamped wing, the strength of the vortices is less due to the drainage of the fluid; the bulk temperature does not increase as in the built-in wing case. The enhancement in bulk temperature with stamped wing and built-in wing is compared (Figure 4.15) at an angle of attack of 20 and Reynolds number 100. The bulk temperature for the built-in wing at the location X=3.3026, is about 10% higher than that for the plate-fin channel without wing, while it is only 5.5% higher for the stamped wing. At the exit, the bulk temperature for the built-in wing is slightly higher than the same for the stamped wing.

85

0.95 0.9 0.85

Reynolds Number 100

Bulk Temperature

0.8 0.75 0.7 0.65 0.6 0.55 0.5 0.45 2 3 4 5 6 7 8

Plate-fin channel without wing Wing with stamping =15 Wing with stamping =20 Wing with stamping =26

Non Dimensional Length X Figure 4.14 Bulk temperature distributions for various angles of attack of the stamped rectangular wing at Reynolds number 100

0.95 0.9 0.85

Reynolds Number 100

Bulk Temperature

0.8 0.75 0.7 0.65 0.6 0.55 0.5 0.45 2 3 4 5 6 7 8

Plate-fin channel without wing Wing without stamping =20 Wing with stamping =20

Non Dimensional Length X

Figure 4.15 Bulk temperature distributions of the rectangular wing with and without stamping 86

The combined spanwise average Nusselt number (Figure 4.16) for the stamped rectangular wing shows substantial enhancement as compared to plate-fin channel without wing and it increases further with the increase in angle of attack of the wing. The strength of the vortices is reduced by the hole beneath the wing due to which the increase in combined spanwise average Nusselt number is less than that obtained in the case of built in wing as shown in Figure 4.17. The maximum enhancement achieved compared to plate-fin channel without vortex generator is 36.67 % with the built in wing and is 17.27% with stamped wing at an angle of attack of 26. The wake region generated by the stamped wing is weak due to the flow of fluid through the hole under the wing. Therefore, the decrease in the combined spanwise average Nusselt number around the trailing edge of the stamped wing is less as compared to that with the built-in wing.

4.7 Effect of Aspect Ratio on the Performance of Rectangular Wing

The aspect ratio of the vortex generator is defined as the ratio of (span)2 / area of vortex generator. For the rectangular wing, the aspect ratio formulation will simply be the ratio of b/c i.e. span / chord length of the wing. In this section, the effect of varying the aspect ratio of the rectangular wing at an angle of attack of 20 is studied. The aspect ratio of the rectangular wing can be increased by increasing the span of the wing. Here the chord length of the wing is kept fixed at 0.7771 and the span is increased. Another way of increasing the aspect ratio by decreasing the chord length does not produce considerable effect on the heat transfer enhancement as the literature shows. Therefore, the area of the wing vortex generator is increased by varying the span which obstructs the flow to a greater extent. Figure 4.18 shows the effect of varying the aspect ratio on the combined spanwise average Nusselt number. The combined spanwise average Nusselt number is enhanced by the increase in the span of the rectangular wing, however, as the wake region generated is more pronounced, the decrease in the combined spanwise average Nusselt number is also higher with the increase in the aspect ratio. Figure 4.19 shows the bulk temperature variation with the increase in aspect ratio of the rectangular wing. At X=3.39, marginal increase in the bulk temperature is observed and for the wing having aspect ratio 0.637, it is only 4.8% higher than the bulk temperature for the wing at aspect ratio of 0.2729.

87

11

Reynolds Number 100

10

Nu sa

Plate-fin channel without wing Wing with stamping =15 Wing with stamping =20 Wing with stamping =26

6 1 2 3 4 5 6 7 8

Non Dimensional Length X

Figure 4.16 Combined spanwise average Nusselt number distributions for various angles of attack of the stamped rectangular wing at Reynolds number 100

13 12 11 10

Reynolds Number 100

Nu sa

9 8 7 6 5
0 1 2 3 4

Plate-fin channel without wing Wing without stampng =26 Wing with stampng =26

Non-Dimensional Length X

Figure 4.17 Combined spanwise average Nusselt number distributions of the rectangular wing with and without stamping

88

13 12 11 10

Reynolds Number 100, =20 Plate-fin channel without wing Wing with aspect raio of 0.2729 Wing with aspect ratio of 0.455 Wing with aspect ratio of 0.637

Nu sa

9 8 7 6 5 0 1 2 3 4

Non-Dimensional Length X

Figure 4.18 Combined spanwise average Nusselt number distribution for different aspect ratios of the built-in rectangular wing at =20 and Reynolds number 100

0.95 0.9 0.85

Reynolds number 100, =20

Bulk Temperature

0.8 0.75 0.7 0.65 0.6 0.55 0.5 0.45 2 3 4 5 6 7 8

Plate-fin channel without wing Wing with aspect ratio 0.2729 wing with aspect ratio 0.455 wing with aspect ratio 0.637

Non-Dimensional Length X

Figure 4.19 Bulk temperature distribution for different aspect ratios of the built-in rectangular wing at =20 and Reynolds number 100

89

4.8 Performance of the In-line Rectangular Wings

A series of vortex generators may be placed one behind the other on the fins of the plate-fin heat exchanger. It results in repeated interruption in the growth of the thermal boundary layer and a further reduction in the resistance to heat transfer. This section deals with such an in-line arrangement of the rectangular wing vortex generators. The trailing edge of the first rectangular wing is placed at X = 2.71 and that of the second is kept at X = 4.66. The chord length and the span of the in-line wings are kept same and the aspect ratio of the wings is 0.2729 for all the angles of attack of the wing. The increase in bulk temperature along the first wing is 25.76%, while along the second inline wing, a further increment of 9% is achieved (Figure 4.20) while both the wings are kept at an angle of attack of 26. The bulk temperature of the fluid striking the second inline wing is already about 76% of the value of the bulk temperature at the exit. Figure 4.21 shows the bulk temperature comparison of the in-line and single rectangular wing vortex generator for an angle of attack of 26 and Reynolds number 100. This figure clearly shows the increase in bulk temperature by the in-line arrangement of the rectangular wing over the single wing vortex generator.

0.9 0.8

Reynolds Number 100

Bulk Temperature

0.7 0.6 0.5 0.4 0.3 1 2 3 4 5 6 7 8

Plate-fin channel without wing In-Line wings without stamping =15 In-Line wings without stamping =20 In-Line wings without stamping =26

Non-Dimensional Length X

Figure 4.20 Bulk temperature distributions for various angles of attack of the in-line rectangular wings at Reynolds number 100

90

Reynolds Number 100

0.9 0.8 0.7 0.6 0.5 0.4 0.3 1 2 3 4 5 6 7 8

Bulk Temperature

Plate-fin channel without wing Inline wing without stamping =26 Single wing without stamping =26

Non-Dimensional Length X

Figure 4.21 Bulk temperature comparison of the in-line and single rectangular wing at an attack angle of 26 and Reynolds number 100 Figure 4.22 shows the combined spanwise average Nusselt number for the in-line arrangement of the rectangular wings at various angles of attack and Reynolds number 100. Similar trends are obtained for the second in-line wing; however, with lower magnitudes of combined spanwise average Nusselt number. The peak enhancement in combined spanwise average Nusselt number achieved by the first wing is 36% and that by the second wing is 20% as compared to the plate-fin channel without any vortex generator. The combined spanwise average Nusselt number comparison of the in-line wings and the single wing is shown in Figure 4.23. This figure shows the superiority of the in-line rectangular wings over the single wing in terms of heat transfer enhancement. Hence the continuous disruption of the boundary layer achieved by the in-line arrangement of the rectangular wings considerably effects the heat transfer augmentation. Along the length of the channel, a series of the in-line wing vortex generators can be mounted and this will increase the heat transfer performance significantly.

91

15 14 13 12

Reynolds Number 100 Plate-fin channel without wing Inline wing without stamping =15 Inline wing without stamping =20 Inline wing without stamping =26

Nu sa

11 10 9 8 7 6 5 0 1 2 3 4

Non-Dimensional Length X

Figure 4.22 Combined spanwise average Nusselt number distributions for various angles of attack of the in-line rectangular wings at Reynolds number 100

13 12 11 10 9 8 7 6 0 1

Reynolds Number 100

Plate-fin channel without wing Single wing without stamping =26 Inline wing without stamping =26

Nu sa

Non-Dimensional Length X

Figure 4.23 Combined spanwise average Nusselt number distribution of the in-line and single rectangular wing at an attack angle of 26 and Reynolds number 100

92

Figure 4.24 compares the variation of bulk temperature for in-line rectangular wings at an angle of attack of 20 and Reynolds number 100 and 200. Similar trends are obtained for the Reynolds number 200 but with lower magnitudes. At higher Reynolds number, the bulk temperature decreases due to the increased mass flow rate. In order to maintain the exit bulk temperature at 0.85 with the Reynolds number at 100, the required length of the exchanger without any vortex generator is 7.78492, which is 35.77% larger than that for the case with in-line rectangular wings at an angle of attack of 26. Similarly reduction of 21.51% and 11.36% in the length of the channel is possible by having the inline rectangular wings at 20 and 15 respectively. The exit bulk temperature of 0.85 can not be achieved by the plain plate-fin heat exchanger while the fluid is flowing at Reynolds number 200. Figure 4.25 shows the variation of combined spanwise average Nusselt number for the in-line rectangular wings at an angle of attack of 20 and Reynolds number 200. At higher Reynolds number, the combined spanwise average Nusselt number increases above that of Reynolds number 100, as greater mass of fluid takes away more heat from the isothermal surfaces.

0.9

Angle of Attack=20

0.8

Bulk Temperature

0.7

0.6

0.5

0.4

Plate-fin channel without wing, Re=100 Wing without stamping, Re=100 Plate-fin channel without wing, Re=200 Wing without stamping, Re=200
1 2 3 4 5 6 7 8

0.3

Non-Dimensional Length X

Figure 4.24 Bulk temperature distributions for the in-line rectangular wings at Reynolds number 100 and 200

93

14

12

Plate-fin channel without wing, Re=100 Wing without stamping Re=100 Plate-fin channel without wing, Re=200 Wing without stamping Re=200

Nu sa

10

Angle of Attack 20

4 0 1 2 3 4 5 6 7 8 9

Non Dimensional Length X

Figure 4.25 Combined spanwise average Nusselt number distributions of the in-line rectangular wings at Reynolds number 100 and 200

4.9 Pressure Loss Penalty

Most of the heat exchangers require fluids to be pumped and so it is essential to determine the fluid pumping power needed as part of the system design. The fluid pumping power is proportional to the fluid pressure drop, which is associated with fluid friction and other pressure drop contributions along the fluid flow. In a plate-fin heat exchanger, the fluid mainly experiences skin friction. Hence the drop in pressure of fluid is experienced by the heat exchanger. In addition, form drag is also experienced at the leading and trailing edges of an interrupted fin surface, which further drop the pressure of the fluid. Hence the augmentation in heat transfer is at the expense of the increase in pumping power requirement. In the present study, the non-dimensional pressure drop along the length of the channel is computed. The average pressure at any section is determined from the integral value of pressure at that location. Pressure drop is computed by subtracting the exit pressure from the average axial pressure. Figure 4.26 shows the drop in pressure for the plate-fin channel using built-in rectangular wing at various angles of attack and Reynolds number 100. 94

5 4.5 4 3.5

Reynolds Number 100

Plate-fin channel without wing Wing without stamping =15 Wing without Stamping =20 Wing without Stamping =26

Pressure Drop

3 2.5 2 1.5 1 0.5 0 0 1 2 3

Non Dimensional Length X

Figure 4.26 Pressure drop for various angles of attack of the built-in rectangular wing A sharp increase in the rate of pressure drop is observed along the location of the wing. The wing vortex generator acts as an obstacle to the flow and therefore, sudden pressure drop takes place. Increasing the angle of attack of the wing demands more pumping power for the same exit conditions. The pressure drop at the inlet in a plate-fin channel using rectangular wing at an angle of attack of 26 is 11% more than the same for a plate-fin without any wing vortex generator. Figure 4.27 compares the pressure drop between a built- in and stamped rectangular wing mounted at an angle of attack of 20 in a plate-fin channel. The requirement of the pressure for the case of stamped wing lies between the values pertaining to plain plate-fin and built in wing. The reason may be the drop in back pressure because of the hole beneath the wing vortex generator. Pressure drop for the in-line arrangement of the rectangular wings at various angles of attack is shown in Figure 4.28. This arrangement further increases the drop in pressure. The pressure drop for single wing is 11.6% lower at the inlet as compared to the in-line wings at an angle of attack of 26o. The effect of increasing the aspect ratio of the wing on the pressure drop is shown in Figure 4.29 and it is found that pressure drop is 10.8% higher for an aspect ratio of 0.637 as compared to the value for the wing with aspect ratio of 0.2729.

95

4.5 4 3.5

Reynolds Number 100

Pressure Drop

3 2.5 2 1.5 1 0.5 0 0 1 2 3 4

Plate-fin channel without wing Wing without stamping =20 Wing with stamping =20

Non Dimensional Length X

Figure 4.27 Pressure drop distribution of the built-in and stamped rectangular wing

6 5 4 3 2 1 0 0 1

Reynolds Number 100 Plate-fin channel without wing In-line wing without stamping =15 In-line wing without stamping =20 In-line wing without stamping =26

Pressure Drop

Non-dimensional Length X

Figure 4.28 Pressure drop for the in-line rectangular wings at various angles of attack of the rectangular wing

96

5 4.5 4 3.5

Reynolds Number 100 Plate-fin channel without wing Wing with aspect ratio .2729 Wing with aspect ratio .455 Wing with aspect ratio .637

Pressure Drop

3 2.5 2 1.5 1 0.5 0 0 1 2 3 4

Non-Dimensional Length X

Figure 4.29 Pressure drop for various aspect ratios of the built-in rectangular wing at =20 and Reynolds number 100

4.10 Concluding Remarks

A secondary flow is observed with its axis parallel to the main flow. The rectangular wing vortex generator produces two main longitudinal vortices along the channel axis and the strength of the vortices is higher for the built-in wing as compared to stamped wing. The strength of the vortices further increases with the larger angles of attack of the wing and higher Reynolds numbers. The churning of the fluid by these vortices is found to enhance the mean temperature of the fluid significantly. The bulk temperature of the fluid increases sharply along the length of the rectangular wing. Increasing the Reynolds number decreases the bulk temperature due to the increased mass flow rate. Compactness of the heat exchanger up to 19.92 % for the exit bulk temperature of 0.85 is achieved by using a rectangular wing with an angle of attack of 26. The stamped wings are easy to produce on the inclined surfaces of the triangular fins and the combined spanwise Nusselt number for the stamped wing at an attack angle of 20 and Reynolds number 100 is 12.74% higher as compared to plate-fin channel without any 97

vortex generator. Moreover the pressure drop is lower for the stamped rectangular wing. So the triangular fins with stamped rectangular wings have a lot of promise for the enhancement of heat transfer. A series of rectangular wings enhance the heat transfer considerably with an extra drop in pressure. Still it is a promising method where the requirement of pumping power is not a constraint for the enhancement of heat transfer.

98

Chapter 5 Performance of the Delta Wing 5.1 Introduction

In the previous chapter, a detailed study pertaining to the performance of rectangular wing was presented. Similarly, this chapter dwells upon the performance evaluation of delta wings mounted on the triangular shaped fins in a plate-fin heat exchanger. Here also the three dimensional velocity distributions and the pressure fields are obtained for built-in and stamped delta wings by numerically solving the continuity and momentum equations. Temperature field in the computational domain is obtained by solving the energy equation. The performance parameters such as combined spanwise average Nusselt number, bulk temperature, static pressure loss etc. are compared with the values in a plate-fin channel without any type of vortex generator to clearly visualize the effect of the delta wing vortex generator. The flow considered is laminar and the Reynolds number is limited to 200 for the same reason as stated in Chapter-4.

5.2 Streamwise Velocity Vectors by Delta Wing

A secondary flow is generated after striking the wing, as the fluid from the space below the wing moves to the space above the wing. The horizontal line MN shown in Figure 5.1 predicts the position of the intersection of transverse plane and the plane of wing at the corresponding axial location. Here the induced spiraling motion takes the fluid from the underside of the wing and swirls it around the upper side and then impinges on the low pressure surface of the wing. This horizontal line may also be considered as location of the intersection of fluid and the inclined plane of wing. The cross-stream velocity vectors at various axial locations along the built-in delta wing for Reynolds number 100 and an angle of attack of 20 are shown in Figure 5.2.

99

Figure 5.1 Generation of the secondary flow The symbol X represents the non-dimensional distance from the inlet of the channel. At the leading edge of the delta wing, a weak spiraling flow is generated as compared to the rectangular wing. The leading edge is just a point in the case of delta wing, whereas rectangular wing has a leading edge equal to the span of the wing. At other locations along the wing, the generation of the vortices is clearly shown in Figure 5.2. The length and the position of the trailing edge of the delta wing are fixed irrespective of the angle of attack of the wing. The non-dimensional length of the delta wing is taken as 1.06 and trailing edge is positioned at X=3.69. Figure 5.3 shows that the longitudinal vortices exist even after crossing the delta wing but the strength of the vortices is reduced. The left hand side vortex loses its strength more rapidly due to the viscous resistance offered by the inclined no-slip plane. The magnification of the velocity vectors shows the longitudinal vortices far away from the delta wing vortex generator. The gradual development of the vortex along the length of the stamped delta wing is shown in Figure 5.4. Some of the fluid moves form the triangular hole beneath the wing and disturbs the fluid flowing below the surface AB. In the lower half domain ABC, the fluid is moving with the same Reynolds number, therefore, the vortex generated in the lower domain by the fluid flowing through the hole is very weak. Figure 5.5 shows the cross-stream velocity vectors downstream of the stamped delta wing vortex generator. The strength of the vortices is decreased due to the viscous resistance of the no-slip surfaces after the vortex generator and hence the boundary layer thickening takes place which provide the thermal resistance to the heat transfer.

100

X=3.69

in Ma

Flo

X=3.3

X=2.91 D

X=2.63

Figure 5.2 Streamwise velocity vectors along the built-in delta wing

X=7.71

in Ma

Flo

X=6.6

X=5.81 D

X=4.18

Figure 5.3 Streamwise velocity vectors after the built-in delta wing

101

in Ma

Flo

X=3.49

X=2.82

Figure 5.4 Streamwise velocity vectors along the stamped delta wing

in Ma

Flo

X=5.81

X=4.18

Figure 5.5 Streamwise velocity vectors after the stamped delta wing

102

5.3 Performance of the Built-in Delta Wing

The heat transfer characteristics of the channel are expressed by the bulk temperature and the combined spanwise average Nusselt number which has already been explained in the previous chapter. The aspect ratio of the longitudinal vortex generator is defined as the square of the span to the area of vortex generator. In case of delta wing, the aspect ratio formulation is 2b c . In the present analysis, the aspect ratio is varied with the angle of attack of the wing by varying the span of the wing. The trailing edge location and the chord length of the delta wing are fixed. The span of the delta wing is 0.5656, 0.7778, and 0.9899 for an angle of attack of 15, 20 and 26 respectively. Figure 5.6 shows the variation of bulk temperature using built-in delta wing vortex generator at different angles of attack while the fluid is flowing at Reynolds number 100.

0.95 0.9 0.85

Reynolds number 100

Bulk Temperature

0.8 0.75 0.7 0.65 0.6 0.55 0.5 0.45 2 3 4 5 6 7 8

Plate-fin channel without wing Wing without stamping =15 Wing without stamping =20 Wing without stamping =26

Non Dimensional Length X

Figure 5.6 Bulk temperature distributions for various angles of attack of the built-in delta wing at Reynolds number 100

103

A considerable enhancement in bulk temperature is observed due to the use of delta wing vortex generator and it increases with the increase in angle of attack of the wing. The use of delta wing vortex generator mounted at an angle of attack of 26 increases the bulk temperature by 21% at X=3.43 as compared to the value for the platefin channel without any vortex generator at the same location. Even at the exit of the channel, the bulk temperature is 12% higher than that for the case of plain plate-fin channel. Further, as expected, increasing the Reynolds number decreases the bulk temperature of the fluid as shown in Figure 5.7. The bulk temperature decreases with an increase in Reynolds number because of the greater mass of fluid flowing through the channel. While the magnitude of bulk temperature achieved is less than the same at Reynolds number 100, a significant increase in bulk temperature is observed and it is higher for the higher angles of attack. The exit bulk temperature while using the delta wing at an angle of attack of 26 is 20.35% higher than the plate-fin channel without wings. The comparison of the bulk temperature variation at the same angle of attack, 26 but at different Reynolds number is shown in Figure 5.8.

0.9 0.8

Reynolds Number 200

Bulk Temperature

0.7 0.6 0.5 0.4 0.3 2 3 4 5 6 7 8

Plate-fin channel without wing Wing without stamping =15 Wing without stamping =20 Wing without stamping =26

Non-Dimensional Length X

Figure 5.7 Bulk temperature distributions for various angles of attack of the built-in delta wing at Reynolds number 200

104

0.9 0.8

Bulk Temperature

0.7 0.6 0.5

Plate-fin channel without wing, Re= 100

0.4 0.3 0.2 1 2 3 4 5 6 7 8 9

Wing without stamping =26, Re=100 Plate-fin channel without wing, Re= 200 Wing without stamping =26, Re=200

Non-Dimensional Length X

Figure 5.8 Bulk temperature comparisons for the built-in delta wing at Reynolds number 100 and 200 Figure 5.9 shows the combined spanwise average Nusselt number using built-in delta wing vortex generator at different angles of attack. The fluid is flowing at Reynolds number 100. It is apparent from the figure that increasing the attack angle greatly enhances the potential of vortex generator and it can be seen that for = 26, the combined spanwise average Nusselt number is 74.53% higher than that for the plain plate-fin channel at X=3.1892. Along the wing, the combined spanwise average Nusselt number rises to a peak value of 13.998 for the attack angle of 26 and then take a plunge near the trailing edge of the wing. A small dead zone exists in the immediate neighborhood of the wing and fin surface junction which causes poor heat transfer at that location. However, to the downstream of the wing, heat transfer is larger as compared to the plain plate-fin channel flow. The Reynolds number here also has a significant effect on the heat transfer characteristics. Higher Reynolds number signifies a greater mass flow rate and consequently the strength of the vortex is higher which ensure better heat removal.

105

16

Reynolds number 100

14

12

Nu sa

10

Plate-fin channel without wing Wing without stamping =15 Wing without stamping =20 Wing without stamping =26

4
0 1 2 3 4 5 6 7 8

Non Dimensional Length X

Figure 5.9 Combined spanwise average Nusselt number distributions for various angles of attack of the built-in delta wing at Reynolds number 100 The effect of Reynolds number on combined spanwise average Nusselt number is clearly evident from Figure 5.10. The peak of the combined spanwise average Nusselt number for the Reynolds number 100 is at 13.998 while it is at 21.89 for Reynolds number 200, which is 56.3% higher. If compared with the plain plate-fin channel at the same location, the combined spanwise average Nusselt number is enhanced by 130% for the Reynolds number 200. At higher angle of attack and higher Reynolds number, the combined spanwise average Nusselt number is more than that for the plain plate-fin channel throughout the length. Due to higher Nusselt number, heat transfer is increased which inherently augments the bulk temperature, therefore, the percentage increase in bulk temperature for the higher Reynolds number is higher. The comparison of the combined spanwise average Nusselt number variation at the same angle of attack, 26, but at different Reynolds number is shown in Figure 5.11.

106

25 23 21 19

Reynolds Number 200

17 15 13 11 9 7 5 0 1 2 3 4 5

Plate-fin channel without wing Wing with stamping =15 Wing with stamping =20 Wing with stamping =26

Nu sa

Non-Dimensional Length X Figure 5.10 Combined spanwise average Nusselt number distributions for various angles of attack of the built-in delta wing at Reynolds number 200

24

19

Nu sa

14

Plate-fin channel without wing, Re=100 Wing without stamping =26, Re=100 Plate-fin channel without wing, Re=200 Wing without stamping =26, Re=200

4
0 1 2 3 4 5 6 7 8 9

Non Dimensional Length X

Figure 5.11 Combined spanwise average Nusselt number comparisons for the built-in delta wing at Reynolds number 100 and 200 107

5.4 Performance of the Stamped Delta Wing

A triangular hole exists beneath the delta wing through which the fluid can flow to the other side. The ease in manufacturing the stamped wings motivates the researchers to study this type of wings. The bulk temperature for the stamped delta wing at various angles of attack is shown in Figure 5.12. The Reynolds number of the flow is 100. The strength of the vortices is less due to the drainage of the fluid and hence, the bulk temperature does not increase as in the built-in wing case. The enhancement in bulk temperature with stamped wing and built-in wing is compared (Figure 5.13) for an angle of attack of 20 and Reynolds number 100. The bulk temperature variation for the stamped wing lies in between the values corresponding to plate-fin channel without wing and built-in delta wing. The length of the channel can be reduced if the heat exchange rate or the exit temperature is held constant. Table 5.1 shows the compactness achieved in terms of reduction in the length of the channel to attain the bulk temperature of 0.85. In a plate-fin channel, the bulk temperature 0.85 is achieved at a non-dimensional length of 7.784 while the same bulk temperature in a plate-fin channel with built-in delta wing at an attack angle of 26 can be obtained at a length of 5.291 which is about 32 % shorter than the required length of the channel without using any vortex generator. Similarly, for the stamped delta wing, about 16.5% reduction in length is possible for the same exit temperature of 0.85. It may, therefore, be concluded that the stamped delta wings are inferior to the built-in wings but better than the plate-fin channel without wing. Figure 5.14 compares the bulk temperature of the fluid at different angles of attack of the wing for the Reynolds number 200 and it shows that the increase in Reynolds number decreases the bulk temperature of the fluid. Table 5.1 gives a comparison of the compactness for the built-in and stamped delta wing vortex generators at various angles of attack of the wing. Table 5.1 Percentage reduction in the length of the channel Geometry Plane Duct Delta Wing =150 Delta Wing =200 Delta Wing =260 Exchanger Length Built in wing Stamped Wing 7.784 7.784 7.257 7.48 6.313 6.99 5.291 6.499 % Reduction Built in wing Stamped Wing 6.77 3.390 18.89 10.20 32.02 16.5

108

0.9 0.85 0.8

Reynolds number 100

Bulk Temperature

0.75 0.7 0.65 0.6 0.55 0.5 0.45 2 3 4 5 6 7 8

Plate-fin channel without wing Wing with stamping =15 Wing with stamping =20 Wing with stamping =26

Non-Dimensional Length X Figure 5.12 Bulk temperature distributions for various angles of attack of the stamped delta wing at Reynolds number 100

0.95 0.9 0.85

Reynolds number 100

Bulk Temperature

0.8 0.75 0.7 0.65 0.6 0.55 0.5 0.45 2 3 4 5 6 7 8

Plate-fin channel without wing Wing without stamping =20 Wing with stamping =20

Non Dimensional Length X Figure 5.13 Bulk temperature comparison of the delta wing with and without stamping

109

0.8 0.75 0.7

Reynolds Number 200

Bulk Temperature

0.65 0.6 0.55 0.5 0.45 0.4 0.35 0.3 2 3 4 5 6 7 8

Plate-fin channel without wing Wing without stamping =20 Wing without stamping =26

Non-Dimensional Length X

Figure 5.14 Bulk temperature distributions for various angles of attack of the stamped rectangular wing at Reynolds number 200 The combined spanwise average Nusselt number (Figure 5.15) for the stamped delta wing shows a significant enhancement as compared to the plate-fin channel without wing. Since the strength of the vortices is reduced by the hole beneath the wing, the increase in combined spanwise average Nusselt number is less than that obtained for the built in delta wing. In case of stamped wing, the dead zone generated around the trailing edge is weak and the combined spanwise average Nusselt number rises sharply after the wing and then gradually decreases along the length. The heat transfer performance in terms of combined spanwise average Nusselt number is compared (Figure 5.16) for the plate-fin channel with and without stamped delta wing at an angle of attack of 26. Increasing the Reynolds number considerably augment the heat transfer performance. The combined spanwise average Nusselt number in case of stamped delta wing at Reynolds number 200, as shown in Figure 5.17, is higher throughout the length than that for the plate-fin channel without wing.

110

12 11 10

Reynolds number 100 Plate-fin channel without wing Wing with stamping =15 Wing with stamping =20 Wing with stamping =26

Nu sa

9 8 7 6 0 1 2 3 4 5

Non Dimensional Length X

Figure 5.15 Combined spanwise average Nusselt number distributions for various angles of attack of the stamped rectangular wing at Reynolds number 100

18

Reynolds Number 100

15

12

Plate-fin channel withour stamping Wing without stamping =26 Wing with stamping =26

Nu sa

3 0 1 2 3 4 5 6 7 8

Non-Dimensional Length X

Figure 5.16 Nusselt number comparison of the delta wing with and without stamping

111

17 15

Reynolds Number 200 Plate-fin channel without wing Wing with stamping =26 Wing with stamping =20

Nu sa

13 11 9 7 5 0 1 2 3 4 5

Non-Dimensional Length X

Figure 5.17 Combined spanwise average Nusselt number distributions for various angles of attack of the stamped rectangular wing at Reynolds number 200

5.5 Performance of the In-line Delta Wings

A series of delta wing vortex generators is placed one behind the other on the fins of the plate-fin heat exchanger. It results in repeated interruption of the growth of thermal boundary layer and reduction in the resistance to heat transfer. The trailing edge of the first delta wing is placed at X = 3.69 and that of the second is kept at X = 5.67. The arrangement is shown in Figure 5.18. The increase in bulk temperature along the first wing is 21.8%, whereas, along the second in-line wing a further increment of 9.6% is achieved (Figure 5.19). Both the wings are kept at an angle of attack of 26. The bulk temperature of the fluid striking the second in-line wing is already about 88.2% of the value of the bulk temperature at the exit. Figure 5.20 depicts the bulk temperature comparison of the in-line and single delta wing at an attack angle of 26 and Reynolds number 100. Figure 5.21 compares the increase in bulk temperature while varying the Reynolds number of the flow for an angle of attack of 20 of the wing. At lower Reynolds number, the increase in bulk temperature along the second in-line wing is comparatively less than the same at higher Reynolds number. At the exit of the channel, the percentage increase

112

in bulk temperature is 22.67% as compared to the plate-fin channel without wing for the Reynolds number 200.
D

In-line Delta Wing

S Q N A
Plane of Symmetry B

A T

M R X=5.67

Q Y
1.06H

X 0.565H R =150

X=3.69

Figure 5.18 In-line configured delta wings

1 0.9 0.8 0.7 0.6

Reynolds number 100

Bulk Temperature

Plate-fin channel without wing

0.5 0.4 0.3 1 2 3 4 5 6 7 8

Wing without stamping =15 Wing without stamping =20 Wing without stamping =26

Non-Dimensional Length X

Figure 5.19 Bulk temperature distributions for various angles of attack of the in-line delta wings at Reynolds number 100

113

0.95 0.9 0.85

Reynolds Number 100

Bulk Temperature

0.8 0.75 0.7 0.65 0.6 0.55 0.5 0.45 2 3 4 5 6 7 8

Plate-fin channel without wing In-line wing without stamping =26 Single wing without stamping =26

Non-Dimensional Length X

Figure 5.20 Bulk temperature comparison of in-line and single rectangular wing at an attack angle of 26 and Reynolds number 100

Angle of Attack 20
0.9 0.8

Bulk Temperature

0.7 0.6 0.5 0.4 0.3

2 3 4 5 6 7 8

Plate-fin channel without wing,Re=100 In-line wing without stamping, Re=100 Plate-fin channel without wing,Re=200 In-line wing without stamping,Re=200

Non-Dimensional Length X

Figure 5.21 Bulk temperature comparisons of the in-line delta wings at Reynolds number 100 and 200

114

Figure 5.22 shows the combined spanwise average Nusselt number for the in-line arrangement of the delta wings at various angles of attack and Reynolds number 100. The peak enhancement in combined spanwise average Nusselt number achieved by the first wing is 75% and that by the second wing is 82% as compared to the plate-fin channel without any vortex generator. The enhancement is considerably higher for larger angles of attack of the wing and even at exit, the combined spanwise average Nusselt number for the case when a wing is used with an attack angle of 26 is 18% higher than the value at the exit for the plate-fin channel without wing. Figure 5.23 compares the combined spanwise average Nusselt number for the in-line arrangement and single delta wing mounted at an angle of attack of 26. Figure 5.24 compares the combined spanwise average Nusselt number for an inline delta wings kept at an angle of attack of 20 for different Reynolds numbers. At Reynolds number 200, the peak combined spanwise average Nusselt number obtained along the second wing is about 90% higher than the value for the plain plate-fin channel.

17 15 13

Plate-fin channel without wing Inline wing without stamping =15 Inline wing without stamping =20 Inline wing without stamping =26

Nu sa

11 9 7 5 3 0 1 2 3 4 5 6 7 8 9

Reynolds number 100

Non-Dimensional Length X
Figure 5.22 Combined spanwise average Nusselt number distributions for various angles of attack of the in-line delta wings at Reynolds number 100 115

18 16 14 12

Reynolds Number 100

Plate-fin channel without wing single delta wing =26 In-line delta wing =26

Nu sa

10 8 6 4 2 0 1 2 3 4 5 6 7 8 9

Non-Dimensional Length X
Figure 5.23 Combined spanwise average Nusselt number comparison of the in-line and single rectangular wing at an attack angle of 260 and Reynolds number100

19 17 15

Plate-fin channel without wing, Re=200 In-line wing without stamping, Re=100 Plate-fin channel without wing, Re=100 In-line wing without stamping, Re=200

Nu sa

13 11 9 7 5 3
0 1 2 3 4 5 6 7 8

Angle of Attack 20

Non-Dimensional Length X
Figure 5.24 Combined spanwise average Nusselt number comparison of the in-line and single rectangular wing at Reynolds number100 and 200

116

5.6 Pressure Drop Penalty

As discussed in Chapter-4, the heat exchangers require fluids to be pumped and so it is essential to determine the fluid pumping power needed as part of the system design. The fluid pumping power is proportional to the fluid pressure drop, which is associated with fluid friction and other pressure drop contributions along the fluid flow. Table 5.2 shows the variation of average pressure drop in a plate-fin channel using delta wings at various angles of attack and at Reynolds number 100. Increase in the angle of attack of the wing calls for more pumping power for the same exit conditions. Figure 5.25 shows the comparison of pressure drop in a plate-fin channel with and without stamping at an angle of attack of 20 and Reynolds number 100. While calculating the pressure drop distributions, the exit pressure is subtracted from the average pressure at any cross-section and the average pressure is determined from the integral value of pressure at that crosssection.

5 4.5 4 3.5

Reynolds Number 100 Plate-fin channel without wing Wing without stamping =20 Wing with stamping =20

Pressure Drop

3 2.5 2 1.5 1 0.5 0 0 1 2 3

Non Dimensional Length X

Figure 5.25 Pressure drop variation of the delta wing with and without stamping The vortex generator acts as an obstacle to the flow and therefore, sudden pressure drop exists there. The drop in pressure is more pronounced along the location of the wing.

117

Increasing the angle of attack of the wing demands more pumping power for the same exit conditions. As stated in Chapter 4, the requirement of pressure with the stamped wing lies in between the plain plate-fin and built in wing values. The in-line arrangement of the delta wings further increases the drop in pressure. At the inlet of the channel, the pressure drop of the in-line wings as compared to plate-fin channel without wing is 39.8% higher while for the single wing the increase in pressure drop is 20.04%.

7 6 5

Reynolds number 100 Plate-fin channel without wing In-line wing without stamping =20 SIngle wing without stamping =20

Pressure Drop

4 3 2 1 0 0 1 2 3

Non Dimensional Length X

Figure 5.26 Pressure drop variation of the in-line and single delta wing Table 5.2 Pressure drop in a plate-fin channel Geometry Plate-fin channel without wing Delta wing at = 15 Delta wing at = 20 Delta wing at = 26 Pressure Drop (Non-Dimensional) 3.83 4.12 4.68 6.01

Table 5.2 shows a considerable increase in pressure drop at the inlet for the builtin delta wing at various angles of attack. The non-dimensional pressure drop for the platefin channel without any vortex generator is 3.83 which is 22% less than using the delta wing vortex generator at = 20.

118

5.7 Comparison of Rectangular and Delta Wings

The effect of the shape of wing on the heat transfer enhancement and pressure loss is studied in this section. For the comparisons to be meaningful, two different view-points are taken: (a) same span and chord length of the wings (b) same area of the wings as shown in Figure 5.27.

b c

b/2

(a)

(b)

Figure 5.27 Geometry for (a) same span and chord length of the wings (b) same area of the wings

5.7.1 Same Chord Length and Span of the Wings

The study is carried out by keeping the span and chord length to be equal for both the wings. In this way, the area of the rectangular wing is double the area of delta wing and the aspect ratio of the delta wing is twice that of rectangular wing. The nondimensional span and chord length of the wings are 0.4949 and 0.7778 respectively. The angle of attack of the wings is taken as 20 for the comparison discussed here. The drop in combined spanwise average Nusselt number (Figure 5.28) near the trailing edge for both the wings is almost same, but the increase in combined spanwise average Nusselt number for the rectangular wing is higher than that for the delta wing. The distribution of combined spanwise average Nusselt number shows the superiority of the rectangular wing over delta wings. The bulk temperature (Figure 5.29) achieved by using the rectangular wing is also higher than that obtained using delta wing vortex generator. In case of rectangular wing, the shielded part of the flow cross-section is more which indicates that the drop in pressure is considerably higher as shown in Figure 5.30. Pressure drop for the case of rectangular wing is 11.9% higher than the same by the delta wing.

119

15 13

Reynolds Number 100 Plate-fin channel without wing Delta wing without stamping =20 Rectangular wing without stamping =20

Nu sa

11 9 7 5 0 1 2 3 4

Non Dimensional Length X

Figure 5.28 Combined spanwise average Nusselt number for the delta and rectangular wing of same chord length and span

1 0.9 0.8

Reynolds number 100

Bulk Temperature

0.7 0.6 0.5 0.4 0.3 0.2 1 2 3 4 5 6 7 8

Plate-fin channel without wing Delta Wing without stamping =20 Rectangular Wing without stamping =20

Non Dimensional Length X

Figure 5.29 Bulk temperature variations for the delta and rectangular wing of same chord length and span. 120

5 4.5 4 3.5

Plate-fin channel without wing Delta wing without stamping =20 Rectangular wing without stamping =20

Pressure Drop

3 2.5 2 1.5 1 0.5 0 0 1 2 3 4 5 6 7 8 9

Non Dimensional Length X

Figure 5.30 Pressure drop variations for the delta and rectangular wing of same chord length and span.

5.7.2 Same Area of the Rectangular and Delta Wings

The area of the rectangular wing is made equal to that of the delta wings. The area of the rectangular wing can be made equal to that of the delta wing by reducing the chord length of the rectangular wing. However, it is a futile exercise as we know that under the condition of same area, the longer the side edge, the more the longitudinal vortices generated and of course, the stronger the disturbance of longitudinal vortices to flow, therefore, more the heat transfer enhancement. Hence, to do the comparison, the area of both the wings is kept same by taking the span of the rectangular wing equal to half of the span of the delta wing. The chord length of the wings is kept the same. The wings are kept at an angle of attack of 26 and the Reynolds number of the flow is 100. The chord length of both the wings is 0.725 and the span of the rectangular wing is 0.3535 which is half of the delta wing. Figure 5.31 shows the combined spanwise average Nusselt number and predicts a higher heat transfer efficacy for the rectangular wing than the delta wing. The rectangular wing offers less area of resistance at its trailing

121

edge as compared to the delta wing. Therefore, the plunge of the combined spanwise average Nusselt number in case of rectangular wing is less than that for the delta wing. Moreover, the rectangular wing produces stronger vortices at its leading edge and hence, the combined spanwise average Nusselt number is higher for the case of rectangular wing.

15

Reynolds number 100

13

Plate-fin channel without wing Delta wing without stamping =26 Rectangular wing without stamping =26

Nu sa

11

5 0 1 2 3 4 5 6 7 8 9

Non Dimensional Length X

Figure 5.31 Combined spanwise average Nusselt number variations for the same area of rectangular and delta wings. The comparison of bulk temperatures pertaining to rectangular and delta wings is shown in Figure 5.32. It can be seen that higher values of bulk temperature are predicted for the case of rectangular wing. It is because of the higher heat transfer achieved due to the rectangular wing for the same mass flow rate in the channel. Although the chord length of both the rectangular and delta wings is same, the pressure drop in the case of rectangular wing is marginally higher than that for the delta wing vortex generator. Figure 5.33 shows the pressure drop variations for both the rectangular and delta wings.

122

1 0.9 0.8

Reynolds number 100

Bulk Temperature

0.7 0.6 0.5 0.4 0.3 0.2 1 2 3 4 5 6 7 8 9

Plate-fin channel without wing Delta wing without stamping =26 Rectangular wing without stamping =26

Non-DImensional Length X

Figure 5.32 Bulk temperature variations for the same area of rectangular and delta wings
6

Plate-fin channel without wing Delta wing without stamping =26 Rectangular wing without stamping =26

Pressure Drop

0 0 1 2 3 4 5 6 7 8 9

Non-Dimensional Length X
Figure 5.33 Pressure drop variations for the same area of rectangular and delta wings

123

5.8 Concluding Remarks

Similar to rectangular wings, the secondary flow is observed with its axis parallel to the main flow. The churning of the fluid is found to enhance the mean temperature of the fluid. Increasing the span of the delta wings considerably augment the heat transfer, though the pressure drop is also increased. The increased span and higher angles of attack of the delta wing produces significant enhancement in heat transfer and bulk temperature. In addition, increasing the Reynolds number further increase the heat transfer but a decrease in bulk temperature is observed due to higher mass flow rate. Hence, an optimum value of Reynolds number may be selected according to the preferences of the designer. The comparison of the rectangular and delta wings for the same wing span and chord length and also for the same area predicts the higher heat transfer enhancement and bulk temperature for the rectangular wings.

124

Chapter 6 Conclusions 6.1 Major Findings

The potential use of longitudinal vortex generators either wings or winglets has long been established through experimental and numerical investigations in tube-fin and rectangular channel flows. In plate-triangular fin heat exchangers, the use of delta and rectangular wings also provide additional heat transfer enhancement on the extended surfaces. The stamped wings circumvent the difficulty of manufacturing the wings on the fin surfaces and further avoid the intermixing of the cold and hot fluids. Although the stamped wings are inferior to the built-in wings, these are always better than the plate-fin channel without any vortex generators. Increasing the angle of attack of the wings considerably increases both the combined spanwise average Nusselt number and the bulk temperature of the fluid. The loss in pressure also increases marginally with the increase in angle of attack of the wing; however, the compactness and the increased heat transfer strongly compensate the pressure loss penalty. Furthermore, the increase in Reynolds number in the proposed geometry augments the heat transfer but the bulk temperature is lower than that at a smaller value of Reynolds number. It is also found that using a series of wings provides a better combined spanwise average Nusselt number and bulk temperature of the fluid with only a little increase in the cost of the pumping power. While comparing the delta and rectangular wings for the same wing span and chord length, the rectangular wing vortex generator is found to be a better choice in terms of heat transfer enhancement and increase in bulk temperature. The area restricted by the rectangular wing is more and hence, the pressure loss is higher than that in the case of delta wings. In another comparison with both the wings having the same area, the

125

rectangular wings again proved to be a better option. However, the drop in pressure is marginally higher than that for the delta wings.

6.2 Scope for Further Work

In the present study, the thickness of the wing vortex generator is not considered. A more accurate model would be to consider the finite thickness of the parallel plates of the plate-fin channel and of the wings vortex generator. It would be a conjugate heat transfer problem and the solution will yield more exact predictions. The computations have been done assuming the flow regime to be laminar. Although turbulent flow is not frequently encountered in plate-fin heat exchangers, in some special applications involving very high velocities, the flow regime can become turbulent. Therefore, the present study can be extended by considering turbulent flows. Using an appropriate turbulence model, the performance of the proposed design can be computed for higher Reynolds numbers.

126

List of Publications from Present Work National & International Journals

1. Sachdeva, G., Vasudevan, R., and Kasana, K. S., 2009, Computation of Heat Transfer Enhancement in a Plate-Fin Heat Exchanger with Triangular Inserts and Delta Wing Vortex Generator, International Journal for Numerical Methods in
Fluids, Published On-Line, DOI 10.1002/fld 2113, Wiley Publications.

2. Sachdeva, G., Kasana, K. S., and Vasudevan, R., 2009, Numerical Analysis of a Plate-Fin Cross Flow Heat Exchanger having Plain Triangular Secondary Fins and In-line Arrangement of Rectangular Wing Vortex Generator, International
Journal of Applied Energy Research,Vol.4, No.9, pp. 1705-1713.

3. Sachdeva, G., Kasana, K. S., and Vasudevan, R., 2010, Analysis on Heat Transfer Enhancement by Using Triangular Shaped Inserts as Secondary Fins in Cross Flow Plate Fin Heat Exchanger, International Journal of Dynamics of
Fluids, Vol. 6, No 4, pp. 41-47.

4. Sachdeva, G., Kasana, K. S., and Vasudevan, R., 2010, Heat Transfer Enhancement by Using Rectangular Wing Vortex Generator on the Triangular Shaped Fins of Plate-Fin Heat Exchanger, Heat Transfer - Asian Research, Published On-Line, DOI 10.1002/htj. 20285, Wiley Publications.

National & International Conferences

1. Sachdeva, G., Kasana, K. S., and Vasudevan, R., 2010, Effect of Varying the Aspect Ratio of Rectangular Wing Mounted on the Triangular Shaped Fins of Plate-Fin Heat Exchanger, Proceedings of the International Conference on
Advances in Mechanical Engineering, S.V. National Institute of Technology,

Surat- Gujrat. 2. Sachdeva, G., Kasana, K. S., and Vasudevan, R., 2010, Numerical Analysis of Plate-Fin Heat Exchanger having In-line Delta Wing Vortex Generator on the Slant Surfaces of the Triangular Shaped Inserts, Proceedings of the International
Conference on Advances in Mechanical Engineering, S.V. National Institute of

Technology, Surat- Gujrat.

127

References
Achaichia, A., and Cowell, T. W., 1988, Heat Transfer and Pressure Drop Characteristics of Flat Tube and Louvered Plate-Fin Surfaces, Experimental Thermal

and Fluid Science, Vol. 1, pp. 147-157.

Allison, C. B., and Dally, B. B., 2007, Effect of Delta-Winglet Vortex Pair on the Performance of a Tube-Fin Heat Exchanger, International Journal of Heat and Mass

Transfer, Vol. 50, No. 25-26, pp. 5065-5072.

Amon, C. H., and Mikic, B. B., 1990, Numerical Prediction of Convective Heat Transfer in Self-Sustained Oscillatory Flows, Journal of Thermophysics and Heat Transfer, Vol. 4, No. 2, pp. 239-246. Bergles, A.E., Augmentation of Heat Transfer, Heat Exchanger Design Handbook, Hemisphere Publishing Company, Washington DC, 1983. Bergles, A. E., Nirmalan, V., Junkhan, G.H., and Webb, R. L., 1983, Bibliography on Augmentation of Convective Heat and Mass Transfer II, Heat Transfer Laboratory

Report HTL-31, ISU-ERI, Iowa State University, Ames, IA.

Biswas, G., Mitra, N. K., and Fiebig, M., 1989, Computation of Laminar Mixed Convection Flow in a Channel with Wing type Built-in Obstacles, Journal of

Thermophysics and Heat Transfer, Vol. 3, pp. 447-453.

Biswas, G., Laschefski, H., Mitra, N. K., and Fiebig, M., 1990, Numerical Investigation of Mixed Convection Heat Transfer in a Horizontal Channel with a Built-in Square Cylinder, Numerical Heat Transfer, Part A, Vol. 18, pp. 173-188. Biswas, G., and Chattopadhyaya, H., 1992, Heat Transfer in a Channel with Built-in Wing Type Vortex Generators, International Journal of Heat and Mass Transfer, Vol. 35, pp. 803-814.

128

Biswas, G., Mitra, N. K. and Fiebig, M., 1994(a), Heat Transfer Enhancement in FinTube Heat Exchangers by Winglet Type Vortex Generators, International Journal of

Heat and Mass Transfer, Vol. 37, pp. 283-291.

Biswas, G., Deb, P. and Biswas, S., 1994(b), Generation of Longitudinal Streamwise Vortices: A Device for Improving Heat Exchanger Design, Journal of Heat Transfer, Vol. 116, pp. 588-597. Biswas, G., Torii, K., Fujii, D., and Nishino, K., 1996, Numerical and Experimental Determination of Flow Structure and Heat Transfer Effects of Longitudinal Vortices in a Channel Flow, International Journal of Heat and Mass Transfer, Vol. 39, No.16, pp. 3441-3451. Biswas, G., and Mitra, N. K., 1998, Longitudinal Vortex Generators for Enhancement of Heat Transfer in Heat Exchanger Applications, Proceedings of the Eleventh

International Heat Transfer Conference, Vol. 5, pp. 339-344.

Brockmeier, U., Guntermann, T., and Fiebig, M., 1993, Performance Evaluation of a Vortex Generator Heat Transfer Surface and Comparison with Different High Performance Surfaces, International Journal of Heat and Mass Transfer, Vol. 36, No. 10, pp. 2575-2587. Chang, L. M., Wang, L.B., Song, K. W., Sun, D. L., and Fan, J. F., 2009, Numerical Study of the Relationship Between Heat Transfer Enhancement and Absolute Vorticity Flux along Main Flow Direction in a Channel Formed by a Flat Tube Bank Fin with vortex Generators, International Journal of Heat and Mass Transfer, Vol. 52, pp. 1794-1801. Chen, Y., Fiebig, M., and Mitra, N.K., 1998, Conjugate Heat Transfer of a Finned Oval Tube with a Punched Longitudinal Vortex Generator in form of a Delta WingletParametric Investigations of the Winglet, International Journal of Heat and Mass

Transfer, Vol. 41, pp. 3961-3978.

129

Chorin, A. J., 1967, A Numerical Method for Solving Incompressible Viscous Flow Problems, Journal of Computational Physics, Vol. 2, pp. 12-26. Chu, P., He, Y. L., Lei, Y.G., Tian, L.T., and Li, R., 2008, Three-Dimensional Numerical Study on Fin-and-Oval-Tube Heat Exchanger with Longitudinal Vortex Generators, Applied Thermal Engineering, Vol. 29, pp. 859-876. Date, A. W., 1974, Prediction of Fully-Developed Flow in a Tube Containing a Twisted Tape, International Journal of Heat and Mass Transfer, Vol. 17, pp. 845-859. Deb, P., Biswas, G. and Mitra, N. K., 1995, Heat Transfer and Flow Structure in Laminar and Turbulent Flows in a Rectangular Channel with Longitudinal Vortices,

International Journal of Heat and Mass Transfer, Vol. 38, pp. 2427-2444.
Edwards, F. J., and Alker, G. J. R., 1974, The improvement of Forced Convection Surface Heat Transfer using Surface Protrusions in the Form of (a) Cubes and (b) Vortex Generators, Proceedings of the Fifth International Heat Transfer Conference, Tokyo, Vol. 2, pp. 2244-2248. Eiamsa-ard, S., Wongcharee, K., Eiamsa-ard, P., and Thianpong, C., 2010, Heat Transfer Enhancement in a Tube using Delta-Winglet Twisted Tape Inserts, Applied

Thermal Engineering, Vol. 30, pp. 310-318.

Eibeck, P. A., and Eaton, J. K., 1987, Heat Transfer Effects of a Longitudinal Vortex Embedded in a Turbulent Shear Flow, Journal of Heat Transfer, (ASME), Vol. 109, pp.16-24. Eswaran, V., and Prakash, S., 1998, A Finite Volume Method for Navier-Stokes Equations, Proceedings of the Third Asian CFD Conference, Bangalore, Vol. 1, pp. 251-269. Fiebig, M., Kallweit, P., and Mitra, N.K., 1986, Wing Type Vortex Generators for Heat Transfer Enhancement, Proceedings of the Eighth International Heat Transfer

Conference, San Francisco, Vol. 6, pp. 2909-2913.

130

Fiebig, M., Brockmeier, U., Mitra, N. K., and Guntemann, T., 1989, Structure of Velocity and Temperature Fields in Laminar Channel Flows with Longitudinal Vortex Generators, Numerical Heat Transfer- Part A, Vol. 15, pp. 103-114. Fiebig, M., Mitra, N., and Dong, Y., 1990, Simultaneous Heat Transfer Enhancement and Flow Loss Reduction of Fin-Tubes, Proceedings of the Ninth International Heat

Transfer Conference, Jerusalem, Vol. 4, pp. 51-55.

Fiebig, M., Kallweit, P., Mitra, N. K., and Tiggelbeck, S., 1991, Heat Transfer Enhancement and Drag by Longitudinal Vortex Generators in Channel Flow,

Experimental Thermal and Fluid Science, Vol. 4, pp. 103-114.

Fiebig, M., Valencia, A., and Mitra, N.K., 1993, Wing type Vortex Generators for Finand-Tube Heat Exchangers, Experimental Thermal and Fluid Science, Vol. 7, pp. 287-295. Fiebig, M., Grosse-Gorgemann, A., Hahne, W., Leiner, W., Mitra, N.K., and Weber, D., 1994(a), Local Heat transfer and Flow Structure in Grooved Channels, Measurements and Computations, Proceedings of the Tenth International Heat Transfer Conference, Brighton, UK, Vol. 4, pp. 237-242. Fiebig, M., Valencia, A., and Mitra, N.K., 1994(b), Local Heat Transfer and Flow Loses in Fin- and-Tube Heat Exchangers with Vortex Generators: A Comparison of Round and Flat Tubes, Experimental Thermal and Fluid Science, Vol. 8, pp. 35-45. Fiebig, M., 1995(a), Embedded Vortices in Internal Flow: Heat Transfer and Pressure Loss Enhancement, International Journal of Heat and Fluid Flow, Vol. 16, pp. 376-388. Fiebig, M., Guntermann, T., and Mitra, N. K., 1995(b), Numerical Analysis of Heat Transfer and Flow Loss in a Parallel Plate Heat Exchanger Element with Longitudinal Vortex Generators as Fins, Journal of Heat Transfer (ASME), Vol. 117, pp. 1064-1067. Fiebig, M., 1998, Vortices, Generators, and Heat Transfer, Transactions of Institute of

Chemical Engineers, Part-A, Vol. 76, pp. 108-123.

131

Gentry, M. C., and Jacobi, A. M., 1997, Heat Transfer Enhancement by Delta-Wing Vortex Generators on a Flat Plate: Vortex Interactions with the Boundary Layer,

Experimental Thermal and Fluid Science, Vol. 14, pp. 231242.

Ghaddar, N. K., Korzak, K. Z., Mikic, B. B. and Patera, A. T., 1986, Numerical Investigation of Incompressible Flow in Grooved Channels, Part 1, Stability and SelfSustained Oscillations, Journal of Fluid Mechanics, Vol. 163, pp. 99-127. Ghia, U.,Ghia, K. N., and Shin, C. T., 1982, High-Re Solutions for Incompressible Flow Using the Navier-Stokes Equations and a Multigrid Method, Journal of Computational

Physics, Vol. 48, pp. 387-411.

Gosh Roychowdhary, D., Das, S. K., and Sundararajan, T., 1999, An Efficient Solution Method for Incompressible Navier-Stokes Equations Using Non-Orthogonal Collocated Grid, International Journal for Numerical method in Engineering, Vol. 45, pp. 741-763. Greiner, M., Ghaddar, N. K., Mikic, B. B., and Patera, A. T., 1986, Resonant Convective Heat Transfer in Grooved Channels, Proceedings of the Eight International Heat

Transfer Conference, San Francisco, USA, Vol. 6, pp. 2867-2872.

Grosse-Gorgemann, A., Hahne, W., and Fiebig, M., 1993, Influence of Rib Height on Oscillations, Heat Transfer and Pressure Drop in Laminar Channel Flow, Proceedings of

Eurotherm 31, Vortices and Heat Transfer, Bochum, Germany, pp 36-41.

Grosse-Gorgemann, A., Weber, D. and Fiebig, M., 1995, Experimental and Numerical Investigation of Self-Sustained Oscillations in Channels with Periodic Structures,

Experimental Thermal and Fluid Science, Vol. 11, No. 3, pp. 226-233.
Harlow, F. H. and Welch, J.E., 1965, Numerical Calculation of Time-Dependent Viscous Incompressible Flow of Fluid with Free Surface, The Physics of Fluids, Vol. 8, pp. 2182-2188.

132

Harlow, F. H., and Amsden, A. A., 1970, The SMAC Method: A Numerical Technique for Calculating Incompressible Fluid Flows, Los Alamos Scientific Lab. Report, LA 4370. Heffington, S. N., 2001, Vibration-induced Droplet Atomization Heat Transfer Cell for Cooling of Microelectronic Components, Proceedings of the IPACK01, paper 15596. Hermann, C. V., and Mayinger, F., 1991, Experimental Investigation of the Heat Transfer in Laminar Forced Convection Flow in a Grooved Channel, Second World

Conference on Experimental Heat Transfer, Fluid Mechanics and Thermodynamics,

Dubrovnik, pp. 387-392. Hermann, C. V., Mayinger, F., Mikic, B. B. and Deculic, D. B., 1992, Numerical and Experimental Studies of Self-Sustained Oscillatory Flows in Communicating Channels,

International Journal of Heat and Mass Transfer, Vol. 35, pp. 3115-3129.
Hiravennavar, S. R., Tulapurkara, E. G., and Biswas, G., 2007, A Note on the Flow and Heat Transfer Enhancement in a Channel with Built-in Winglet Pair, International

Journal of Heat and Fluid Flow, Vol. 28, pp. 299-305.

Hirt, C. W. and Cook, J. L., 1972, Calculating Three-Dimensional Flows around Structures and over Rough Terrain, Journal of Computational Physics, Vol. 10, pp. 324-340. Hoffman, G., and Benocci, C., 1994, Numerical Simulation of Spatially Developing Planer Jets, Advisory Group for Aerospace Research and Development (AGARD), CP551, pp. 26.126.6. Incropera, F. P., and Schutt, J. A., 1985, Numerical Simulation of Laminar Mixed Convection in the Entrance Region of Horizontal Rectangular Ducts, Numerical Heat

Transfer, Vol. 8, pp. 707-729.

133

Incropera, F. P., Knox, A. L., and Maughan, J. R., 1987, Mixed-Convection Flow and Heat Transfer in the Entry Region of a Horizontal Rectangular Duct, Journal of Heat

Transfer, Vol. 109, pp. 434-439.

Issa, R. I., Gosman, A. D. and Watkins, A.P., 1986, The Computation of Compressible and Incompressible Recirculating Flows by a Non-Iterative Implicit Scheme, Journal of

Computational Physics, Vol. 62, pp. 66-82.

Jacobi. A. M., and Shah. R.K., 1995, Heat Transfer Surface Enhancement through the use of Longitudinal Vortices: A review of Recent Progress, Experimental Thermal and

Fluid Science, Vol. 11, pp. 295-309.

Jacobi. A. M., and Shah. R.K., 1998, Air-Side Flow and Heat Transfer in Compact Heat Exchangers: A Discussion of Enhancement Mechanisms, Heat Transfer Engineering, Vol. 19, No. 4, pp. 29-41. Jang, D.S., Jetli, R., and Acharya, S., 1986, Comparison of PISO, SIMPLER and SIMPLEC Algorithms for the Treatment of the Pressure Velocity Coupling in Steady Flow Problems, Numerical Heat Transfer, Vol.10, pp. 209-228. Joardar, A., and Jacobi, A. M., 2008, Heat Transfer Enhancement by Winglet type Vortex Generator Arrays in Compact Plain-Fin-and-Tube Heat Exchangers,

International Journal of Refrigeration, Vol. 31, pp. 87-97.

Karki, K. C., A Calculation Procedure for Viscous Flows at all Speeds in Complex Geometries, PhD Thesis, University of Minnesota, 1986. Kataoka, K., Doi, H., and Komai, T., 1977, Heat/Mass transfer in Taylor Vortex Flow with Constant Axial Flow Rates, International Journal of Heat and Mass Transfer, Vol. 20, pp. 57-63. Kays, W.M., and London, A.L, Compact Heat Exchangers 3rd edition McGraw Hill Book Company, 1984.

134

Kobayashi, M.H., and Pereira, C.F., 1991, Calculation of Incompressible Laminar Flows on a Non-Staggered, Non-Orthogonal Grid, Numerical Heat Transfer, Part-B, Vol. 19, pp. 243-262. Kost, A., Bai, L., Mitra, N. K. and Fiebig, M., 1991, Calculation Procedure for Unsteady Incompressible 3-D Flows in Arbitrary Shaped Domains, Proceedings of the Ninth

GAMM-Conference on Numerical Methods in Fluid Mechanics, pp. 269-278.

Kwak, K. M., Torii, K., and Nishino, K., 2003, Heat Transfer and Pressure Loss Penalty for the Number of Tube Rows of Staggered Finned-Tube Bundles with a Single Transverse Row of Winglets, International Journal of Heat and Mass Transfer, Vol. 46, pp. 175-180. Launder, B. E., and Spalding, D.B., 1974, The Numerical Computation of Turbulent Flows, Computer Methods in Applied Mechanics and Engineering, Vol. 3, pp. 296-289. Lawson, M. J., and Thole K. A., 2008, Heat Transfer Augmentation along the Tube Wall of a Louvered Fin Heat Exchanger using Practical Delta Winglets, International Journal

of Heat and Mass Transfer, Vol. 51, pp. 2346-2360.

Li, Q., and Xuan, Y., 2002, Convective Heat Transfer Performance of Fluids with Nanoparticles, Proceedings of the Twelfth International Heat Transfer Conference, Vol. 1, pp. 483-488. Liao, Q., Zhu, X., and Xin, M. D., 2000, Augmentation of Turbulent Convective Heat Transfer in Tubes with Three Dimensional Internal Extended Surfaces, Journal of

enhanced heat transfer, Vol. 7, No.3, pp. 139-151.

Majumdar, S., Rodi, W., and Zhu, J., 1992, Three-Dimensional Finite-Volume Method for Incompressible Flows with Complex Boundaries, Journal of Fluids Engineering

(ASME), Vol. 114, pp. 496-503.

135

Maughan, J. R., and Incropera, F. P., 1991, Use of Vortex Generators and Ribs for Heat Transfer Enhancement at the Top Surface of a Uniformly Heated Horizontal Channel with Mixed Convection Flow, Transactions of the ASME, Vol.113, pp. 504-507. Mukhopadhyay, A., Sundararajan, T., and Biswas, G., 1993, An Explicit Transient Algorithm for Predicting Incompressible Viscous Flows in Arbitrary Geometry,

International Journal of Numerical Methods in Fluids, Vol.17, pp. 975-993.

Nakamura, H., and Tanaka, M., 1973, Cross Rifled Vapour Generating Tube, U.S. Patent 3,734,140, May 22. Nichols, B. D., and Hirt, C. W., 1971, Improved Free Surface Boundary Conditions for Numerical Incompressible-Flow Calculations, Journal of Computational Physics, Vol. 8, pp. 434-448. Orlanski, I., 1976, A Simple Boundary Condition for Unbounded Flows, Journal of

Computational Physics, Vol. 21, pp. 251-269.

Pais, M. R., Chow, L.C. and Mahefkey, E.T., 1992, Surface Roughness and its Effects on the Heat Transfer Mechanism in Spray Cooling, Journal of Heat Transfer, Vol. 114, pp. 211-219 Patankar, S.V., and Spalding, D.B., 1972, A Calculation Procedure for Heat Mass and Momentum Transfer in Three-Dimensional Parabolic Flows, International Journal of

Heat and Mass Transfer, Vol.15, pp. 1787-1806.

Patankar, S.V., 1981, A Calculation Procedure for Two-Dimensional Elliptic Situations, Numerical Heat Transfer, Vol.4, pp. 409-425. Peric, M., A Finite Volume Method for the Prediction of Three-Dimensional Fluid Flow in Complex Ducts, Ph.D. Thesis, University of London, 1985.

136

Pesteei, S. M., Subbarao, P. M. V., and Agarwal, R. S., 2005, Experimental Study of the Effect of Winglet Location on Heat Transfer Enhancement and Pressure Drop in FinTube Heat Exchangers, Applied Thermal Engineering, Vol. 25, pp. 1684-1696. Prabhkar, V., Biswas, G., and Eswaran, V., 2003, Numerical Prediction of Heat Transfer in a Channel with a Built-in Oval Tube and Various Arrangements of the Vortex Generators, Numerical Heat Transfer- Part A, Vol. 44, pp. 315-333. Rhie, C. M., and Chow, W. L., 1983, Numerical Study of the Turbulent Flow Past an Airfoil with Trailing Edge Separation, The American Institute of Aeronautics and

Astronautics (AIAA), Vol. 21, pp. 1525-1532.

Robichaux, J., Tafti, D.K., and Vanka, S.P., 1992, Large-Eddy Simulations of Turbulence on the CM-2, Numerical Heat Transfer, Part-B, Vol.21, pp. 367-388. Russels, C. M. B., Jones, T.V., and Lee, G. H., 1982, Heat Transfer Enhancement using Vortex Generators, Proceedings of the Seventh International Heat Transfer Conference, Munich ,Vol.3, pp. 283-288. Saha, A.K., Biswas, G., and Murlidhar. K., 1999, Numerical Study of the Turbulent Unsteady Wake behind a Partially Enclosed Square Cylinder using RANS, Computer

Methods in Applied Mechanics and Engineering, Vol. 178, pp. 323-341.

Saha, A. K., Murlidhar, K., and Biswas, G., 2000, Vortex Structures and Kinetic Energy Budget in Two-Dimensional Flow Past a Square Cylinder, Computers and Fluids, Vol. 29, pp. 669-694. Saha, A.K., Biswas, G., and Murlidhar, K., 2003, Three-Dimensional Study of Flow past a Square Cylinder at Low Reynolds Numbers, International Journal of Heat and Fluid

Flow, Vol. 24, pp. 54-66.

137

Sanchez, M., Mitra, N. K., and Fiebig, M., 1989, Numerical Investigation of ThreeDimensional Laminar Flows in a Channel with Built-in Circular Cylinder Wing-Type Vortex Generators, Proceedings Eighth GAMM Conference on Numerical methods in

Fluid Mechanics, pp. 484-492.

Shah, R. K., and Sekuli, P. Duan., Fundamentals of Heat Exchanger Design, John Wiley & Sons Incorporation, 2003. Sohankar, A., and Davidson, L., 2001, Effect of Inclined Vortex Generators on Heat Transfer Enhancement in a Three Dimensional Channel, Numerical Heat Transfer, Part-

A, Vol. 39, pp. 433-448.

Sohankar, A., 2004, The LES and DNS Simulations of Heat Transfer and Fluid Flow in a Plate-Fin Heat Exchanger with Vortex Generators, Iranian Journal of Science and

Technology: Transaction B: Technology, Vol. 28, pp. 443-452.

Sohankar, A., 2007, Heat Transfer Augmentation in a Rectangular Channel with a V-Shaped Vortex Generator, International Journal of Heat and Fluid Flow, Vol. 28, pp. 306-317. Thomas, W. C. and Sunderland, J.E., 1970, Heat Transfer between a Plane Surface and Air Containing Water Droplets, Industrial and Engineering Chemistry Fundamentals, Vol. 9, pp. 368-374. Thompson, J. F., Warsi, Z. U. A., and Mastin, C. W., 1982, Boundary Fitted Coordinate System for Numerical Solution of Partial Differential Equations- A Review, Journal of

Computational Physics, Vol. 47, pp. 1-108.

Tiggelbeck, S., Mitra, N. K. and Fiebig, M., 1992, Flow Structure and Heat Transfer in a Channel with Multiple Longitudinal Vortex-Generators, Experimental Thermal and

Fluid Science, Vol. 5, pp. 425-436.

138

Tiggelbeck, S., Mitra, N. K. and Fiebig, M., 1994, Comparison of Wing type Vortex Generators for Heat Transfer Enhancement in Channel Flows, Transactions of the

ASME, Vol. 116, pp. 880-885.

Tiwari, S., Maurya, D., Biswas, G., and Eswaran, V., 2003, Heat Transfer Enhancement in Cross Flow Heat Exchangers using Oval Tubes and Multiple Delta Winglets,

International Journal of Heat and Mass Transfer, Vol. 46, pp. 2841-2856.
Tiwari. S., Chakraborty, D., Biswas, G., and Panigrahi, P. K., 2005, Numerical Prediction of Flow and Heat Transfer in a Channel in the Presence of a Built-in Circular Tube with and without an Integral Wake Splitter, International Journal of Heat and

Mass Transfer, Vol. 48, pp. 439-453.

Torii, K., Yanagihara, J. I., and Nagai, Y., 1991, Heat Transfer Enhancement by Vortex Generators, Proceedings of the ASME/JSME Thermal Engineering Joint Conference, J. R. Lloyd, and Y. Kurosaki, eds., ASME Book No. I0309C, ASME, New York, pp. 77 83. Torii, K., Nishino, K. and Nakayama, K., 1994, Heat Transfer Augmentation by Longitudinal Vortices in a Flat Plate Boundary Layer, Proceedings of the Tenth

International Heat Transfer Conference, Brighton, Vol. 6, pp. 123-128.

Torii, K., Kwak, K. M. and Nishino, K., 2002, Heat Transfer Enhancement Accompanying Pressure-Loss Reduction with Winglet-type Vortex Generators for FinTube Heat Exchangers, International Journal of Heat and Mass Transfer, Vol. 45, pp. 3795-3801. Turk, A.J., and Junklan, G.H., 1986, Heat Transfer Enhancement Downstream of Vortex generators on a Flat Plate, Proceedings of the Eighth International Heat Transfer

Conference, San Francisco, Vol. 6, pp. 2903-2908.

Valencia, A., Fiebig, M., and Mitra, N. K., 1996, Heat Transfer Enhancement by Longitudinal Vortices in Fin-tube Heat Exchanger Element with Flat tubes, Journal of

Heat Transfer (ASME), Vol. 118, pp. 209-211.

139

Van-Doormaal, J.P., and Raithby, G.D., 1984, Enhancements of the SIMPLE Method for Predicting Incompressible Fluid Flows, Numerical Heat Transfer, Vol. 7, pp. 147-163. Vasudevan, R., Eswaran, V., and Biswas, G., 2000, Winglet type Vortex Generators for Plate-Fin Heat Exchangers using Triangular Fins, Numerical Heat Transfer, Part A, Vol. 58, pp. 533-555. Viecelli, A. J., 1971, A Computing Method for Incompressible Flows Bounded by Moving Walls, Journal of Computational Physics, Vol. 8, pp. 119-143. Webb, R. L., 1987, Enhancement of Single-Phase Convective Heat Transfer, Hand Book of Single-Phase Convective Heat Transfer, (Ed Kakac, S., Shah, R. K. and Aung, W.) John Wiley & Sons, New York. Wen, M. Y., and Ho, C. Y., 2009, Heat Transfer Enhancement in Fin-and-Tube Heat Exchanger with Improved Fin Design, Applied Thermal Engineering, Vol. 29, pp. 10501057. Wu, J. M., and Tao, W. Q., 2008, Numerical Study on Laminar Convection Heat Transfer in a Channel with Longitudinal Vortex Generator Part B: Parametric Study of Major Influence Factors, International Journal of Heat and Mass Transfer, Vol.51, pp. 3683-3692. Xia, C., 2002, Spray/Jet Cooling for Heat Flux High to 1 kW/cm2, Eighteenth IEEE

Semiconductor Thermal Measurement and Management Symposium, pp. 159-163.

Yabe, A., 1991, Active Heat Transfer Enhancement by Applying Electric Fields,

Proceedings of the Third ASME/JSME Thermal Engineering Conference, Vol. 3,

pp.15-23 Yanagihara, J. L. and Torii, K., 1990, Heat Transfer Characteristics of Laminar Boundary Layer in the Presence of Vortex Generators, Proceedings of the Ninth

International Heat Transfer Conference, Vol. 6, pp. 323-328.

140

Yanagihara, J. I., and Torii, K., 1993, Heat Transfer Augmentation by Longitudinal Vortices Rows, Proceedings of the Third World Conference on Experimental Heat

Transfer, Fluid Mechanics, and Thermodynamics, M. D.Kelleher et al., eds., 1,

pp. 560567. Yang, J.S., Seo, J.K., and Lee, K.B., 2001, A Numerical Analysis on Flow Field and Heat Transfer by Interaction between a Pair of Vortices in Rectangular Channel Flow,

Current Applied Physics, Vol. 1, pp. 393-405.

Yang, J. S., Lee, D.W., and Choi, G.M., 2008, Numerical Investigation of Fluid Flow and Heat Transfer Characteristics by Common Flow-up, International Journal of Heat

and Mass Transfer, Vol. 51, pp. 6332-6336.

Zhang, L. Z., 2007, Laminar Flow and Heat Transfer in Plate-Fin Triangular Ducts in Thermally Developing Entry Region, International Journal of Heat and Mass Transfer, Vol. 50, pp. 1637-1640. Zhang, Y.H., Wu, X., Wang, L.B., Song, K. W., Dong, Y. X., and Liu, S., 2008, Comparison of Heat Transfer Performance of Tube Bank Fin with Mounted Vortex Generators to Tube Bank Fin with Punched Vortex Generators, Experimental Thermal

and Fluid Science, Vol. 33, pp. 58-66.

Zhu, X.J., Fiebig, M., and Mitra, N. K., 1993(a), Comparison of Numerical and Experimental Results for a Turbulent Field with a Longitudinal Vortex Pair, Journal of

Fluids Engineering, Vol. 115, pp. 270-274.

Zhu, X. J., Mitra, N. K., and Fiebig, M., 1993(b), Effects of Longitudinal Vortex Generators on Heat Transfer and Flow Loss in Turbulent Channel Flows, International

Journal of Heat and Mass Transfer, Vol. 36, pp. 2339-2347.

Zhu, X.J., Fiebig, M., and Mitra, N. K., 1995, Numerical Investigation of Turbulent Flows and Heat Transfer in a Rib-Roughened Channel with Longitudinal Vortex Generators, International Journal of Heat and Mass Transfer, Vol. 38, No 3, pp. 495501. 141

Appendix A Non-Dimensional Formulation of Governing Equations Navier-Stokes Equations

The complete Navier-Stokes equations in conservative form are derived in many of the text books. The x-momentum equation in dimensional form is as below.

( u ) u2 + t x + x

)+

( u v ) ( u w) P + = y z x
A-1

u u v V + 2 + x + y x y +

w u + f x + z z x

The incompressible Navier-Stokes equation can be obtained from compressible form simply by setting density equal to constant and with to be constant, continuity equation becomes
V = 0

With the further assumption that is constant throughout, Equation A-1 is written as (u ) u2 + t x

( )+

(u v ) ( u w) P 2u + = + 2 y z x x2 v u x + y + z u w z + x + fx

+ y

A-2

Now from the continuity equation V = u w v + + = 0 x y z

A-3

Rearranging Equation A-3, we have u v w = x y z Differentiating Equation A-3a with respect to x, we obtain
2u x2 = 2v 2w x y x z

A-3a

A-4

Adding

2u x2

to both sides and then multiplying the terms of Equation A-4 by .

142

We obtain 2 2u x2 = 2u x2 2v 2w x y x z
A-5

Substituting equation A-5 for the second term on the right side of Equation A-2
(u ) + t u2 x

( )+
2

(u v ) + y

( u w) P 2u 2v = + z x x y x2 + z u w z + x + fx

w + x z y

v u x + y

A-6

Canceling terms in Equation A-6 and neglecting the body force, we obtain a convenient form of x-momentum equation for a viscous, incompressible flow as
(u ) + t u2 x

( )+

(u v ) + y +

P ( u w) = x z 2u x2 + 2u y2 + 2u z2

A-7

2u Du P 2u 2u + = + 2 + Dt x y2 z2 x
Du P = + 2 u Dt x

where 2 u is the laplacian of the x component of velocity, u. Compressible NavierStokes equations in y and z directions are treated in a similar fashion and the continuity equation and the momentum equations for the incompressible, viscous flow are as under Continuity : V = 0
x momentum:
Du P = + 2 u Dt x Dv P = + 2 v Dt x Dw P = + 2 w Dt x

A-8 A-9

y momentum: z momentum:

A-10

A-11

The momentum equation in x direction is non-dimensionlized as follows. The non dimensional time () is
= t H / U av

143

The non dimensional velocity is

U = u U av

V =

v U av

W =

w U av

First term on the left hand side of Equation A-7 in non-dimensional variables is as under
2 U av u U = t H

A-12

The lengths are non-dimensionlized by the spacing between the neighboring plates H i.e.
x y z Y = Z = H H H Second term on the left hand side of x-momentum equation is as below X =
2 U av U 2 u2 = x H X

A-13

Third term on the left hand side of x-momentum equation is as below

2 U av u v U V = y H Y

A-14

Similarly the forth term can be written as

2 U av u w U W = z H Z

A-15

Non-Dimensional form of the pressure is

P = p
2 U av

Now the pressure gradient term on the right hand side of Equation A-7
2 U av p P = H x X

A-16

The second order derivatives of U are set to non-dimensional form as under.

2u x2 = U av H2 2U X 2

A-17

Similarly, it can be written

2u y2 2u z2 = U av H2 U av H2 2U Y 2 2U Z 2
A-18

A-19

144

Substituting the Equations A-12 to A-19 into the Equation A-7

2 U av H

U U 2 U V U W + + + X Y Z +

2 = U av H

P X

U av 2U 2U 2U + + Y 2 Z 2 H 2 X 2

U U 2 U V U W P + + + = + 2U X Y Z X U av H

The non dimensional Reynolds number (Re) is

Re =

(U av H )

Hence the non-dimensional momentum equation in X-direction is as under

U U 2 U V U W P 1 + + + = + 2U X Y Z X Re

Similarly the momentum equation in Y and Z direction is

V V 2 VU V W P 1 + + + = + 2V Y X Z Y Re

( )

W W 2 WU WV P 1 + + + = + 2W Z X Y Z Re

Energy Equation
The conservation dimensional form of the energy equation in terms of internal energy is written as below
( e ) + ( eV ) = q + t x u v w P x + y + z
2

T k x + y

T k y + z
2

T k z
2

u v w + x + y + z
2

u v + 2 x + 2 y
2

E-1

w u u v v w w + 2 z + y + x + z + x + z + y

Flow considered is incompressible, so with to be constant, continuity equation becomes

V = u v w + + = 0 x y z

145

Internal heat generation is neglected and the thermal conductivity k is also assumed to be constant, the Equation E-1 can be written as
( e ) + ( eV ) = k 2 T + t
E-2

where

u = 2 x

v + 2 y

w + 2 z

u v + y + x
2

u w + z + x

v w + z + y

The * is a dissipative term which is usually neglected for the incompressible flows. Using the relation e = c p T , the left side of the Equation E-2 can be written as
cp (uT ) T (vT ) (wT ) 2 + cp + + = k T t y z x

cp

T u T T T v w + c p u +v + w + T x + y + z t y z x

= k 2 T

Using continuity equation, the dimensional form of energy equation is as follows

cp T T T T 2 + c p u +v + w = k T t y z x

E-3

The non dimensional temperature ( ) is

= T ) Tw T

(T

The non dimensional time () is

t H / U av

So the non-dimensional form of T t is

T (T T )U av = w t H

E-4

The lengths are non-dimensionlized by the spacing between the neighboring plates H.

146

We obtain

x y Y = H H The non-dimensional form of T x is X = T T T = w x H X

Similarly T y and T z can be written as

Z =

z H

E-5

T T T = w y H Y T T T = w z H Z
The non-dimensional form of 2T x 2 is

E-6

E-7

2 T x2

T T 2 = w 2 H X2

E-8

In the same way, the other spatial second order terms are
2 T y2 2 T z2 T T 2 = w 2 H Y 2 T T 2 = w 2 H Z2

E-9

E-10

The non dimensional velocity is

U = u U av

V =

v U av

W =

w U av

E-11

Substituting the Equations E-4 to E-11 into E-3

c p U av (Tw T ) + U X + V Y + W Z = H k (Tw T ) 2 2 2 + + 2 H2 Y 2 Z 2 X

2 k 2 2 +V +W = + + +U c p HU av X 2 Z X Y Y 2 Z 2
The non dimensional Reynolds number (Re) is
Re =

E-12

(U av H )

147

The non dimensional Prandtl number (Pr) is Pr =

k c p

Substituting these numbers in Equation E-12, we get the non-dimensional energy equation.

1 1 2 2 2 +U +V +W = + + Pr Re X 2 X Y Z Y 2 Z 2

148

Appendix-B The Program Substructure

In order to solve the non-dimensional Navier-Stokes and energy equation, a computer program in Visual FORTRAN is developed based on the MAC algorithm as described in Chapter-3. The code consists of several subroutines each of which has a set of specific tasks to carry out. The main program (RWING) and the subroutines are written in the form of modules. The RWING code determines the velocities U, V, W and the pressure distributions in all over the domain. The flow chart provided in Figure B-1 shows the operational sequence of the various subroutines and their communication links with the main program. The velocity distribution provided by the RWING program is used to solve the energy equation by successive over-relaxation technique and it is done in another program (ENERGY). The code (ENERGY) determines the temperature distribution in the domain. After getting the velocity, pressure and temperature distributions by these codes, another code (POSTPRO) is run which determines the bulk temperature, combined spanwise average Nusselt number, vorticity, and the pressure drop. A brief description of the different indices and the functions of the subroutines of RWING program are given below. INITILIZE This subroutine reads the variables such as Reynolds number, Prandtl number, epsi, length of the channel, location of the wing etc from an input data file and initialize the calculations. Dimensions and total number of cells in X, Y and Z directions are calculated according to the input angle of attack of the wing. START CONTI This subroutine specifies the guessed distribution of the velocities and pressures all over the computational domain. This subroutine compares the maximum divergence of the continuity equation and controls the pressure-velocity iteration loop. CONBC This subroutine deals with the boundary conditions required for continuity equation. At the inlet, exit and inclined no-slip plane, the boundary conditions are specified. 149

BCW CONEQ

This subroutine specifies the boundary conditions for the plane of the wing. In this subroutine, the maximum divergence of the velocity vector in the complete domain is calculated. The pressure is corrected and velocity field is modified. The CONTI subroutine iterates this subroutine along with the subroutines specifying boundary conditions i.e. CONBC, BCW until the maximum divergence is less than the specified value.

NSBC

This subroutine specifies the boundary conditions required for the solution of Navier-Stokes equation. Only at the inlet plane, these conditions are specified. BCW subroutine is also called in this subroutine.

VALT TICORR OUTPUT

This subroutine stores the current (nth time step) velocity field in order to start the flow field at the n+1th time step. In this subroutine the value of the time increment and the cell coefficient () are calculated. This subroutine checks the maximum time increment of the field variables and if it is less than the specified value STAT along with the iteration counter ITA > 1, it stops the program. This subroutine can also stop the program if the iteration exceeds the specified number of iterations ITAMAX.

EQNS

This subroutine explicitly determines the flow field for the next time step from the discretized Navier-Stokes equations. This subroutine itself applies the inclined no-slip and the plane of symmetry boundary conditions at the concerned cells while calculating the velocity field.

TIGRAD BCOW

This subroutine determines the time-gradients of the velocity components. This subroutine is called by the EQNS subroutine while calculating the flow field for each cell. In this subroutine the numerical boundary conditions for the wing vortex generator are defined. The subroutine BCW and CONBC together define the vortex generator.

EPSI

It is a value prescribed as the upper bound for the velocity divergence in each cell. 150

MDIV STAT ITAMAX ITI ITA

It is the maximum value of the divergence in the entire flow field computed in the subroutine CONEQ. It is a small value which determines the conditions for steady state solution. It is the maximum number of time steps allowed to achieve the required numerical steady state solution. It is an iteration counter for the CONTI subroutine It is the number of time steps in which the solution of the problem is achieved. A brief description of the different indices and the functions of the subroutines of

ENERGY program are given below. INIT This subroutine reads the output file of the main RWING program and also defines the set of variables such as cell dimensions, no of cells, Reynolds number etc. Guessed values of the temperature field are also specified in this subroutine. THBC BCOT EQTE Thermal boundary condition i.e. temperature is defined at the inlet and exit plane. This subroutine specifies the temperature boundary conditions for the wing vortex generator. This subroutine determines the temperature field by for the next iteration from the discretized energy equation. This subroutine itself applies the constant wall temperature boundary condition on the inclined no-slip and the plane of symmetry. OUTPUT TALT DT TSTAT This subroutine writes the temperature distribution in the assigned output file. This subroutine stores the current (nth iteration) temperature field in order to start the temperature computation at the n+1th iteration. This denotes the value of the maximum difference of the temperature in the entire field between two iterative steps. It is a small value that determines the convergence for steady solution of the energy equation.

151

START

RWING

INITILIZE

START

CONTI

BCW

CONBC

CONEQ

ITI+1

MDIV < EPSI

NO

YES NSBC BCW

VALT

TICORR

Figure B-1 Flow chart for the velocity simulation

152

OUTPUT

MDIF<STAT AND ITA>1 NO EQNS

YES

STOP

BCOW

NSBC

BCW

TISTEP

Figure B-1 Flow chart for the velocity simulation

153

START

ENERGY

INIT

THBC

TALT

EQTE

BCOT

ITT+1

NO

DT< TSTAT

YES OUTPUT

STOP

Figure B-2 Flow chart for the temperature simulation

154