Aerodynamics of rectangular slot-burners and combustion … · Aerodynamics of Rectangular Slot-Burners and Combustion in Tangentially-fired Furnace . A Thesis Submitted for the Degree

NOTE

This online version of the thesis may have different page formatting and pagination from the paper copy held in the Swinburne Library.

Aerodynamics of Rectangular Slot-Burners and Combustion in Tangentially-fired Furnace

A Thesis Submitted for the Degree of Doctor of Philosophy

By

Shakil Ahmed

Faculty of Engineering & Industrial Science Swinburne University of Technology

February 2005

Dedicated to my father

ii

Abstract

Abstract

The power generation industry in the state of Victoria, Australia stands to gain

significantly from process improvements and optimization which can potentially lead to

cleaner production of cost effective electricity. The efficient operation of lignite based

tangentially-fired combustion systems depends on critical issues such as ignition and

combustion of the fuel, which are largely controlled by burner aerodynamics. The

geometry of the burner and the ratio of velocities between the primary and secondary

jets play an important role in achieving stable combustion, high burnout of fuel, low

production of pollutants and control of fouling. Slot-burners are a vertically aligned

stack of rectangular nozzles delivering primary fuel and secondary air jets, and are

commonly used in tangentially-fired boilers. To obtain a better understanding of the

overall combustion process, it is important to understand the aerodynamics of jet

development from these burners.

The starting point of this research was a CFD investigation of aerodynamics in the near-

burner region of isolated rectangular slot-burners, using isothermal conditions, for

various secondary to primary jet velocity ratios (φ). Cross-flow was then added to

replicate a near-burner flow field similar to that found in a tangentially-fired furnace

and the effect of changing φ in the near-burner region of the developing jets was again

investigated. Experiments were carried out on an isothermal physical-burner model to

obtain mean velocity and turbulent statistics for different nozzle geometries and a range

of φ. A computational fluid dynamics investigation of these same jets was also

performed to gain further insights into the complexities of flow field with experimental

results used to validate CFD predictions. The primary jet substantially deviated from the

geometric axis of the burner towards the furnace wall and became very unstable for

higher φ. The causes of unfavourable aerodynamics were discussed and suggestions

were made on possible remedies for such behaviour. Conventional lignite combustion in

a full-scale tangentially-fired furnace was modelled. The model was used to assess the

possibility of utilizing a new type of mechanically thermally dewatered (MTE) coal in

existing furnaces.

iv

Acknowledgements

Acknowledgements

It is a pleasure to acknowledge a number of great persons who in different ways,

academically, professionally and psychologically, contributed to the successful

completion of this work.

First of all I gratefully acknowledge my supervisor Dr. Jamal Naser, who arranged

everything for my scholarship and gave me the opportunity to work in this project. In

particular I appreciate his invaluable and enterprising guidance throughout the execution

of this project and his emphasis in defining achievable time bound steps to reach the

overall goals.

I am grateful to Dr. James Hart for his constant encouragement, valuable suggestions

through this thesis and patience in reading and correcting the thesis within a short

period of time.

I would like to thank Dr. Chris Solnordal, Dr. William Yang and Dr. Jonian Nikolov of

CSIRO, Division of Minerals, Melbourne, Australia for their significant contribution to

design and develop the experimental rig and by giving their valuable time and effort to

complete the experiment in time.

I would also like to thank my wife, Rumana, for the family support by taking care of my

daughter and giving me the relief from cooking until the completion of the project.

Dr Peter Jackson, Chief Executive Officer, CRC for Clean Power from Lignite, I

appreciate your effort in keeping an eye on the progress of my work and thank you very

much for being available whenever I needed you in this research program. Dr. Malcolm

McIntosh of CRC for Clean Power from Lignite, I appreciate your innovative idea and

suggestions in time of my need.

Thanks to all of my friends, who don’t know they helped and inspired me, but they did.

I'm sure I've forgotten someone. I assure you that this is a shortcoming on my part and

not on yours. I beg you to forgive me for my oversight.

vi

Acknowledgements

Finally I gratefully acknowledge the financial and other support received for this

research from the Cooperative Research Centre (CRC) for Clean Power from Lignite,

which is established and supported under the Australian Government’s Cooperative

Research Centre’s program.

vii

Declaration of Originality This Thesis contains no material, which has been accepted for the award of any other

degree or diploma at any university and to the best of my knowledge and belief

contains no material previously published or written by another person or persons

except where due reference is made

………………………………….

Shakil Ahmed

viii

Tables of Contents

Tables of Contents

Page Number

ABSTRACT iii

ACKNOWLEDGEMENT v

DECLARATION viii

TABLE OF CONTENTS ix

LIST OF FIGURES xiv

LIST OF TABLES xxv

CHAPTER 1 1. Introduction 1

1.1. Primary research theme 2

1.2. Overview of the research program 2

1.3. Background 3

1.4. Objectives 6

1.5. Thesis structure 7

1.6. Perceived specific contributions of the research 8

CHAPTER 2 2. Literature review 9

2.1. Single-phase flow development 10

2.1.1. Slot-burners without cross-flow 10

2.1.2. Jet in cross-flow 13

2.1.2.1. Single round jet 13

2.1.2.2. Multiple round jets 17

2.1.2.3. Non-circular jets 18

2.2. Two-phase flow development 20

2.3. Modeling of coal combustion in tangentially-fired furnace 23

2.3.1. CFD as a modeling tool 23

2.3.2. Tangentially-fired furnace 24

2.3.3. Coal combustion scheme 25

x

Tables of Contents

2.3.3.1. Coal devolatilization 25

2.3.3.2. Gaseous combustion models 27

2.3.3.3. Char combustion models 29

2.3.3.4. Radiation models 31

2.3.3.5. NOx models 33

CHAPTER 3 3. Mathematical models, modeling techniques and methodologies 35

3.1. Mathematical model 37

3.2. Discretization method 37

3.3. Discretization scheme 38

3.4. Solution method 39

3.5. Difficulties in numerical simulation 40

3.6. Turbulence modeling 41

3.7. Dimensions of the burners and furnace 43

3.8. Model description grid and boundary conditions 49

3.8.1. Investigation without cross-flow 49

3.8.2. Investigation with cross-flow 51

3.8.3. Investigation of full-scale tangentially-fired furnace 54

3.9. Experimental set-up and methodology used in physical modeling 57

3.9.1. Experimental set-up 57

3.9.2. Methodology 58

3.10. Error analysis 63

3.11. Summary 64

CHAPTER 4 4. Numerical investigation of rectangular slot-burners without cross-flow 65

4.1. Grid independence test 66

4.2. Results and discussion 68

4.2.1. Comparison with experiment 68

4.2.2. Flow field prediction 81

4.3. Summary and conclusions 86

xi

Tables of Contents

CHAPTER 5 5. Experimental investigation of rectangular slot-burners with cross-flow 87


5.1.1. Geometry B 88

5.1.1.1. Flow field for φ=1.0 and 3.0 88

5.1.1.2. Comparison of mean and RMS velocity 91

5.1.2. Geometry D 100

5.1.2.1. Flow field for φ=1.0 and 3.0 101

5.1.2.2. Comparison of mean and RMS velocity 103


CHAPTER 6 6. Numerical investigation of rectangular slot-burners with cross-flow 113

6.1. Grid independence test 114

6.2. Geometry B 115

6.2.1. Secondary to primary jet velocity ratio of 1.0 115

6.2.1.1. Validation of numerical results 115

6.2.1.2. Flow field prediction 120




6.3 Geometry D 135







6.4 Summary and conclusions 151

xii

Tables of Contents

CHAPTER 7 7. Experimental investigation of two-phase flow with cross-flow 153


7.1.1. Geometry B 155

7.1.1.1. Comparison for φ=1.0 with cross-flow 155


7.1.2. Geometry D 159 7.1.2.1. Comparison for φ=1.0 with cross-flow 159



CHAPTER 8 8. Combustion in a tangentially-fired furnace 164

8.1. Numerical model verification 167

8.2. Results and discussions 169

8.3 Summary and conclusions 179

CHAPTER 9 9. Conclusions and recommendations 181

9.1. Conclusions 182

9.2. Recommendations for further work 184

CHAPTER 10 10. References 186

APPENDIX I 199

LIST OF PERSONAL PUBLICATIONS 203

xiii

List of Figures

List of figures

Page No. Figure 1.1(a) Tangentially-fired furnace showing the primary, secondary and

vapor burner.

3

Figure 1.1(b) Velocity vectors showing the swirling flow at the centre of the

furnace.

4

Figure 2.1 Flow configuration for low and high jet to cross-flow velocity

ratio.

14

Figure 2.2(a) Flow development for low (ψ=0.5) jet to cross-flow velocity

ratio.

14

Figure 2.2 (b) Flow development for high (ψ=2.0) jet to cross-flow velocity

ratio.

15

Figure 2.3 Structural features of jets in cross-flow (adapted from Johnston

& Khan, 1997).

16

Figure 2.4 Supply channel flow configurations. 18

Figure 2.5 Coordinate definitions for (a) perpendicular, (b) streamwise, (c)

span wise jets.

19

Figure 3.1 Geometry C-60ο to the furnace wall where the cavity has

parallel sidewalls.

44

Figure 3.2 Geometry D- Diverging inserts between furnace mouth and the

furnace wall.

45

Figure 3.3(a) Burner B (Dimensions are in mm). 46

Figure 3.3(b) Burner D (Dimensions are in mm). 46

Figure 3.3(c) Cross-flow nozzle (Dimensions are in mm). 47

Figure 3.4 Photographs showing burner B, burner D and cross-flow duct. 47

Figure 3.5 Schematic diagram of the furnace showing all the dimensions

in meter (m).

48

Figure 3.6 Orientation of the burners in tangentially-fired furnace. 49

Figure 3.7(a) Burner model showing geometry C, D and upstream ducts. 50

Figure 3.7(b) Solution domain showing the body fitted mesh. 50

Figure 3.8(a-b) Dimensioned view of flow containment box (a) and burner inlet

detail (b) (all units in m). 52

Figure 3.9(a) Mesh on geometry B and cross-flow duct. 53

xv

List of figures

Figure 3.9(b) Mesh on geometry D and cross-flow duct. 53

Figure 3.10(a-b) Inlet and outlet patches of Yallourn stage-2 furnace (a) and

primary and Secondary recessed nozzles (b).

55

Figure 3.11(a-b) Grid layout of the full furnace (a), primary and secondary

nozzles (b). 56

Figure 3.12 Schematic diagram of burner model and associated ducting. 57

Figure 3.13 Dual beam optical system and fringe pattern. 58

Figure 3.14 Schematic diagram of LDA apparatus. 59

Figure 3.15 Histogram display showing U and V component of velocity 62

Figure 4.1(a) Grid distribution in the xy plane. 67

Figure 4.1(b) Velocity profile in xy plane for grid independence test. 67

Figure 4.2(a) Centreline decay of peak axial velocity for primary and

secondary jet (geometry C, φ=1.0).

68

Figure 4.2(b) Centreline decay of peak axial velocity for primary and

secondary jet (geometry D, φ=1.0).

69

Figure 4.3 Comparison of centreline decay of the primary jet between

geometry C and D (φ=1.0).

70

Figure 4.4(a) The effect of jet velocity ratio on centreline velocity decay for

primary jet (geometry C).

71

Figure 4.4(b) The effect of jet velocity ratio on centreline velocity decay for

primary jet (geometry D).

71

Figure 4.5(a) Velocity contour at the centre plane of the primary jet for

geometry C at φ = 3.0.

71

Figure 4.5(b) Velocity contour at the centre plane of the primary jet for

geometry D at φ = 3.0.

72

Figure 4.5(c) Mixing of Secondary jets at the centre plane of the primary jet

for geometry D at φ=3.0.

72

Figure 4.6(a) Transverse velocity profiles at x/De=1.0 (geometry C, φ=1.0). 73

Figure 4.6(b) Transverse velocity profiles at x/De=5.0 (geometry C, φ=1.0). 73

xvi

List of figures

Figure 4.6(c) Transverse velocity profiles at x/De=9.0 (geometry C, φ=1.0). 74







Figure 4.9(a) Transverse velocity profiles at x/De=1.0 (geometry D, φ=1.0). 76

Figure 4.9(b) Transverse velocity profiles at x/De=5.0 (geometry D, φ=1.0). 77

Figure 4.9(c) Transverse velocity profiles at x/De=9.0 (geometry D, φ=1.0). 77

Figure 4.10(a) Transverse velocity profiles at x/De=1.0 (geometry D, φ=1.4). 78

Figure 4.10(b) Transverse velocity profiles at x/De=5.0 (geometry D, φ=1.4). 78

Figure 4.10(c) Transverse velocity profiles at x/De=9.0 (geometry D, φ=1.4). 78

Figure 4.11 Transverse velocity profiles at x/De=1.0 (geometry D, φ=3.0). 79

Figure 4.12(a) Deviation of jet centre with an increase in φ for Geometry C

(simulated jet).

79

Figure 4.12(b) Deviation of jet centre with an increase in φ for Geometry D

(simulated jet).

80

Figure 4.13(a) Validation of the simulated jet deviation with the experimental

values for geometry C (φ=3.0).

80

Figure 4.13(b) Validation of the simulated jet deviation with the experimental

values for geometry D (φ=3.0).

81

Figure 4.14

Pressure distribution (a) and velocity vector (b) in the xy plane

through the centre of the primary axis for geometry C (φ=1.0).

82

Figure 4.15 Pressure distribution (a) and velocity vector (b) in the xy plane

through the centre of the primary axis for geometry C (φ=1.4). 83

xvii

List of figures


through the centre of the primary axis for geometry C (φ=3.0).

83


through the centre of the primary axis for geometry D (φ=1.0).

84



85



85

Figure 5.1(a) Velocity vectors in the centre plane of the primary jet for

φ=1.0.

88

Figure 5.1(b) Velocity vectors in the centre plane of the lower base region for

φ=1.0.

89

Figure 5.1(c) Velocity vectors in the centre plane of the lower secondary jet

for φ=1.0.

89


φ=3.0. 90

Figure 5.2(b) Velocity vectors in the centre plane of the lower base region for

φ=3.0.

90

Figure 5.2(c) Velocity vectors in the centre plane of the lower secondary jet

for φ=3.0. 91

Figure 5.3 Schematic diagram showing the measurement positions. 91

Figure 5.4(a) Velocity profiles in the centre plane of the primary jet for

φ=1.0.

92

Figure 5.4(b) Velocity profiles in the centre plane of the primary jet for

φ=3.0. 92

Figure 5.5(a) Velocity profiles in the centre plane of lower base region for

φ=1.0. 93

Figure 5.5(b)

Velocity profiles in the centre plane of lower base region for

φ=3.0. 93

xviii

List of figures

Figure 5.6(a)

Velocity profiles in the centre plane of the lower secondary jet

for φ=1.0. 94

Figure 5.6(b) Velocity profiles in the centre plane of the lower secondary jet

for φ=3.0.

94

Figure 5.7(a) urms at the centre of the primary jet for φ=1.0. 95

Figure 5.7(b) vrms at the centre of the primary jet for φ=1.0. 95

Figure 5.7(c) uv stress at the centre of the primary jet for φ=1.0. 96

Figure 5.8(a) urms at the centre of the primary jet for φ=3.0. 96

Figure 5.8(b) vrms at the centre of the primary jet for φ=3.0. 97

Figure 5.8(c) uv stress at the centre of the primary jet for φ=3.0. 97

Figure 5.9(a) urms at the centre of the lower secondary jet for φ=1.0. 98

Figure 5.9(b) vrms at the centre of the lower secondary jet for φ=1.0. 98

Figure 5.9(c) uv stress at the centre of the lower secondary jet for φ=1.0. 99

Figure 5.10(a) urms at the centre of the lower secondary jet for φ=3.0. 99

Figure 5.10(b) vrms at the centre of the lower secondary jet for φ=3.0. 99

Figure5.10(c) uv stress at the centre of the lower secondary jet for φ=3.0. 100


φ=1.0.

101

Figure 5.11(b) Velocity vectors in the centre plane of the lower secondary jet

for φ=1.0. 101


φ=3.0.

102

Figure 5.12(b) Velocity vectors in the centre plane of the lower secondary jet

for φ=3.0.

102

Figure 5.13(a) Velocity profiles in the centre plane of the primary jet for

φ=1.0. 103

Figure 5.13(b) Velocity profiles in the centre plane of the primary jet for

φ=3.0.

104

Figure 5.14(a) Velocity profiles in the centre plane of the lower base region

for φ=1.0.

105

Figure 5.14(b) Velocity profiles in the centre plane of the lower base region

for φ=3.0.

105

xix

List of figures

Figure 5.15(a) Velocity profiles in the centre plane of the lower secondary jet

for φ=1.0.

106

Figure 5.15(b) Velocity profiles in the centre plane of the lower secondary jet

for φ=3.0.

106

Figure 5.16(a) urms in the centre plane of the primary jet for φ=1.0. 106

Figure 5.16(b) vrms in the centre plane of the primary jet for φ=1.0. 107

Figure 5.16(c) uv stress in the centre plane of the primary jet for φ=1.0. 107

Figure 5.17(a) urms in the centre plane of the primary jet for φ=3.0. 108

Figure 5.17(b) vrms in the centre plane of the primary jet for φ=3.0. 108

Figure 5.17(c) uv stress in the centre plane of the primary jet for φ=3.0. 109

Figure 5.18(a) urms in the centre plane of the lower secondary jet for φ=1.0. 109

Figure 5.18(b) vrms in the centre plane of the lower secondary jet for φ=1.0. 109

Figure5.18(c) uv stress in the centre plane of the lower secondary jet for

φ=1.0.

110

Figure 5.19(a) urms in the centre plane of the lower secondary jet for φ=3.0. 110

Figure 5.19(b) vrms in the centre plane of the lower secondary jet for φ=3.0. 110

Figure5.19(c) uv stress in the centre plane of the lower secondary jet for

φ=3.0.

110

Figure 6.1 Grid independence test for geometry B. 115

Figure 6.2 Comparison of resultant velocities at the centre plane of the

primary jet for φ=1.0.

117


lower base region for φ=1.0.

118


lower secondary jet for φ=1.0.

119

Figure 6.5(a-b) Pressure distribution in the xy plane through the centre of the

primary jet (a) and lower secondary jet (b) for φ=1.0. 120

Figure 6.6 Pressure distribution (a) and streamlines (b) in the yz plane at a

distance x/De=0.31 from the wall.

121

Figure 6.7(a) Centreline velocity decay of Primary jet for φ=1.0. 122

Figure 6.7(b) Centreline velocity decay of lower Secondary jet for φ=1.0. 123

xx

List of figures

Figure 6.8(a-b)

Velocity vectors at the centre of the primary jet (a) and the

lower secondary jet (b) for φ=1.0. 123

Figure 6.9(a-b)

Velocity vectors in the yz plane at x/De=0.078 and 0.156

showing the formation of twin vortex for φ=1.0.

124

Figure 6.9(c-d) Velocity vectors in the yz plane at x/De=0.234 and 0.312


124

Figure 6.9(e-f) Velocity vectors in the yz plane at x/De=0.390 and 0.468


125

Figure 6.10 Comparison of resultant velocities at the centre of the primary

jet for φ=3.0.

126

Figure 6.11 Comparison of resultant velocities at the centre of lower base

region for φ=3.0.

127

Figure 6.12 Comparison of resultant velocities at the centre of lower

secondary jet for φ=3.0.

128

Figure 6.13(a-b) Streamlines and pressure distribution in the xy plane through

the centre of the primary jet (a) and lower secondary jet (b).

130

Figure 6.14 Pressure distribution (a) and streamlines (b) in the yz plane at a


131

Figure 6.15(a-b) Velocity vectors at the centre plane of the primary jet (a) and

the lower secondary jet (b) for φ=3.0.

132

Figure 6.16(a-b) Pressure distribution (a) and the velocity vector (b) in the xz

plane at the centre of the geometric axis for φ=3.0.

132

Figure 6.17(a) Centreline velocity decay of primary jet for φ=3.0. 133

Figure 6.17(b) Centreline velocity decay of lower secondary jet for φ=3.0. 133

Figure 6.18(a-b) Velocity vectors in the yz plane at x/De=0.156 and 0.3125

showing the formation of twin vortex for φ=3.0. 134

Figure 6.18(c-d) Velocity vectors in the yz plane at x/De=0.468 and 0.625

showing the formation of twin vortex for φ=3.0. 134

Figure 6.18(e-f) Velocity vectors in the yz plane at x/De=0.781 and 0.938


135



136

xxi

List of figures


lower base region for φ=1.0.

137

Figure 6.21

Comparison of resultant velocities at the centre plane of the


138

Figure 6.22(a-b) Velocity vectors at the centre plane of the primary jet (a) and

lower secondary jet (b).

140

Figure 6.23(a-b) Streamlines and pressure distribution in the xy plane through

the centre of the primary jet (a) and lower secondary jet (b).

141

Figure 6.24(a-b) Pressure distribution (a) and streamlines (b) in the yz plane at a

distance x/De=0.31 from the wall. 142

Figure 6.25(a-d) Velocity vectors in the yz plane showing the formation of twin

vortex for φ=1.0.

143



144



145

Figure 6.28(a) Velocity vectors through the geometric centre of both primary

and secondary jets.

146

Figure 6.28(b) Pressure contour through the geometric centre of both primary

and secondary jets.

147

Figure 6.29(a) Velocity vectors at the centre plane of the primary jet for

φ=3.0.

147

Figure 6.29(b) Velocity vectors at the centre plane of the lower secondary jet

for φ=3.0.

148

Figure 6.29(c) Velocity vectors in the yz plane at x/De=0.31 from the wall for

φ=3.0.

148

Figure 6.30(a) Pressure contour at the centre plane of the primary jet for

φ=3.0.

149

Figure 6.30(b) Primary, upper and lower secondary streamlines inside the

recess.

149

Figure 6.30(c) Primary streamlines viewing from the top. 150

xxii

List of figures

Figure 6.31

Pressure contour and lower secondary streamlines at the centre

plane of the lower secondary jet. 150

Figure 7.1 Comparison of resultant velocity between the gas-phase and

particle-phase in the centre plane of the primary jet for φ=1.0.

156



158

Figure 7.3

Comparison of resultant velocity between the gas-phase and


160



161

Figure 8.1 Proposed schematic diagram showing all the processes before

the MTE lignite (velocity 20m/s) enters in to the furnace.

166

Figure 8.2(a-b) Measured data level. 167

Figure 8.3(a) Comparison of the predicted temperature contours with

experimental data’s at the center plane of the lower primary

nozzle.

168

Figure 8.3(b) Comparison of the predicted temperature contours with

experimental values in a vertical plane 1100 mm from the exit

of the burner towards the furnace.

169

Figure 8.3(c) Comparison of the predicted oxygen concentration with

experimental values in a vertical plane 1100 mm from the exit

of the burner towards the furnace.

169

Figure 8.4(a) Velocity vectors at the centre plane of the lower primary nozzle

for lignite combustion.

170

Figure 8.4(b) Velocity vectors at the centre plane of the lower primary nozzle

for MTE lignite combustion with velocity 7.47 m/s.

171

Figure 8.4(c) Velocity vectors at the centre plane of the lower primary nozzle

for MTE lignite combustion with velocity 20.0 m/s. 171

Figure 8.5(a) Velocity vectors at the centre plane of the upper primary nozzle

for lignite combustion.

172

Figure 8.5(b) Velocity vectors at the centre plane of the upper primary nozzle


173

xxiii

List of figures

Figure 8.5(c) Velocity vectors at the centre plane of the upper primary nozzle


173

Figure 8.6(a) Temperature contours at the centre plane of the lower primary

nozzle for conventional lignite combustion.

174

Figure 8.6(b) Temperature contours at the centre plane of the lower primary

nozzle for MTE lignite combustion with velocity 7.47 m/s.

175

Figure 8.6(c)

Temperature contours at the centre plane of the lower primary

nozzle for MTE lignite combustion with velocity 20.0 m/s.

175

Figure 8.7(a) Shaded temperature contours at the centre of the furnace for

conventional lignite combustion.

177

Figure 8.7(b) Shaded temperature contours at the centre of the furnace for

MTE lignite combustion with velocity 7.47 m/s.

177

Figure 8.7(c) Shaded temperature contours at the centre of the furnace for

MTE lignite combustion with velocity 20.0 m/s.

178

Figure 8.8(a) NOx concentration (vol ppm) for the conventional lignite

combustion in the XY plane at z=50m.

178

Figure 8.8(b) NOx concentration (vol ppm) for MTE lignite combustion

(velocity 20m/s) in the XY plane at z=50m.

179

Figure A 1 Interaction of stress components in a 2D plane jet 202

xxiv

List of Tables

List of tables

Page No. Table 3.1 Technical data of the LDA System. 60

Table 8.1 Coal characteristics before entering into the furnace. 165

Table 8.2 Comparison of extraction of hot flue gas and evaporaton of

water in mill.

180

xxvi

NOTE


Chapter-1 Introduction


1. Introduction 1.1. Primary research theme

This thesis reports an investigation into the aerodynamics of rectangular slot-burners of

the type used in tangentially-fired furnaces. The geometric design of the slot-burners

influences the near field development of the jets. Experiments were performed to

measure the mean flow characteristics of the near field region for different types of

burners. To get more details on aerodynamics, computational fluid dynamics (CFD), a

numerical technique used for calculating the flow field by solving the Navier Stokes

equations, was used and the numerical results were compared with the experiments.

Combustion of lignite coal (a low rank coal which contains 66-70% water) in a full-

scale tangentially-fired furnace was investigated by using CFD and the potential impact

of burning coal from a newly developed process called MTE, where coal is dried and

water is removed in the liquid state before going into the furnace by applying thermal

energy and mechanical force, was assessed.

1.2. Overview of the research program

This research program was undertaken at Swinburne University of Technology and

CSIRO Division of Minerals in Melbourne, Australia between 2001 and 2004, under the

sponsorship of the Cooperative Research Centre (CRC) for Clean Power from Lignite.

In Australia the majority of large-scale electricity generation plants derive their heat

energy from the combustion of pulverized fuel. Because of the enormous reserves of

lignite available and its ease of extraction it is one of the cheapest fuels. Lignites

typically contain very low (less than 1% dry-ash-free basis) sulphur and nitrogen

contents and an ash content of less than 2% db (dry basis) but the high water contents

(66-70% wet basis) is a major drawback for the power generation (Durie, 1991). To

extract the same amount of energy more lignite has to be burnt compared with a black

coal boiler of similar capacity due to high water content. This will produce more

greenhouse gas emissions. At the same time for lignite fired power plants environmental

legislation have become more and more stringent. One of the main objectives of the

CRC is to reduce the greenhouse emissions associated with the lignite combustion. The

design of the burners directly influences the level of greenhouse gas emissions through

the stability of the combustion process, which can impact on the degree of wall fouling

and hence the heat transfer efficiency. The majority of the Victorian lignite fuelled

2


furnaces are tangentially-fired and use rectangular slot-burners. To obtain a better

understanding of the overall combustion process in lignite fired boilers it is important to

investigate the aerodynamics of the jet development from the slot-burners.

1.3. Background

In a tangentially-fired furnace, the burners engender a rotational flow in the furnace by

directing the jets tangent to an imaginary circle whose centre is located at the centre of

the furnace. The resultant swirling and combusting flow generates a fireball at the centre

of the furnace where the majority of combustion occurs. Each burner set is composed of

a series of vertically aligned slots. In the case of separation firing systems an inert or

vapor burner is used in addition to the main fuel burner, but it is located higher in the

furnace. Separation firing is a technique used to stage the combustion in the furnace,

giving greater control over heat release and pollutant formation. An individual burner in

a lignite fired furnace typically consists of a central or primary slot (fuel rich stream)

providing the pulverized fuel, primary air and hot flue gas and two outer slots (oxygen

rich stream) providing preheated secondary air. The main purpose of the jets in a lignite

fired burner is to heat the coal and air by mixing it with entrained hot furnace gases, and

to deliver the air and fuel to the correct location in the centre of the furnace. Figure

1.1(a) shows a typical tangentially-fired furnace. The fireball at the centre plane of the

lower primary nozzle is shown in figure 1.1(b).

To convection section

Flue gas offtakes

Vapor burner

Lower m bur

ain ner

Upper main burner

Upper top secondary nozzle

Lower primary nozzle

Lower bottom secondary nozzle

Figure 1.1(a): Tangentially-fired furnace showing the primary, secondary and vapor

burner.

3


4 5

6 3

7 2

1 8

Figure 1.1(b): Velocity vectors showing the swirling flow at the centre of the furnace.

The thermo-fluid interaction processes between neighboring burners and between the

burners and the furnace as a whole are complex and not well understood (Perry, 1982).

The entrainment of fluid into the jets near the nozzle and the resulting near field flow

development is extremely important. There needs to be a balance between entrainment

of hot furnaces gases into the fuel/air jets in order to aid devolatilization of coal and

some early combustion while ensuring sufficient momentum is conserved to maintain

the swirling fireball in the furnace. So it is of great importance to understand the

aerodynamics of near field region not only to ensure the jets reach the centre of the

furnace at the correct location but also to stabilize the flame in the centre.

While combustion in the furnace involves complex chemical reactions, radiative heat

transfer and significant changes in the gas density due to expansion of product of

combustion, isothermal investigation of slot-burners provide a useful starting point to

understand the complex flow behaviour in near field region of the jets. Since 1981,

Swinburne University of Technology has collaborated closely with the State Electricity

Commission of Victoria in a number of research projects designed to improve

understanding of mixing of air and fuel in near field region of jets. These projects

considered the influence of burner geometry and jet velocity ratio on the near field flow

development in a slot-burner in isolation. Jet velocity ratio (φ) was defined as the ratio

of secondary to primary jet velocity.

4


A part of those research programs was also to develop a numerical model using

computational fluid dynamics (CFD). Computing technology of the time restricted

modeling to coarse orthogonal grids. Relatively simplistic mathematical representation

of the turbulent physics reduced the ability of the computing models to usefully aid in

understanding the underlying mechanisms governing the burner behaviour. In the late

1990s and early 2000s computing power had increased to the point were a more realistic

attempt could be made to use CFD to model such flows.

In general, CFD has reduced the need to rely solely on expensive physical modeling. In

some cases physical modeling is not only expensive but also impractical. Once

confidence in the CFD model is acquired, many flows can be investigated at full-scale

and far more detailed and meaningful results may be obtained. However, some form of

physical modeling is usually desirable to validate the numerical model. Physical and

numerical modeling can work together in engineering design to provide more insight

than either technique by itself.

The development of the flow field in the near-burner region is influenced by burner

geometry, jet velocity ratio and complex rotational flow in the rest of the tangentially-

fired furnace. In a simple isolated burner study it is not possible to faithfully model all

of these influences, particularly the furnace flow field. In order to obtain the near flow

field similar to the tangentially-fired furnace a cross-flow was introduced in isothermal

burner model. Cross-flow has a significant effect in developing the near flow field of a

burner. By observing the jets coming from burner 6 and 7 (figure 1.1(b)) the effect of

cross-flow can be understood. For burner 6, after discharging into the furnace, the jets

are influenced by another side jet where the direction of the side jet is in the same

direction of the jet velocity component parallel to the furnace wall. On the other hand

for burner 7 (figure 1.1(b)), the side jet is in the opposite direction of the wall

component of jet velocity. So in a tangentially-fired furnace two types of cross-flow

exits; first, where the cross-flow favors the burner jets (burner 6) and the second one,

where the cross-flow opposes the burner jets (burner 7). In this thesis, only the cross-

flow similar to burner 6 has been investigated i.e. the direction of the cross-flow was

always in the same direction of the wall component of jet velocity.

5


Due to the high moisture content of the lignite (66-70%), thermal conversion efficiency

of current lignite-fired power stations is low, leading to higher levels of greenhouse gas

emissions than from plants of similar capacity fuelled with gas or even high-rank black

coal. Efficient dewatering or pre-drying is therefore the first and most important step in

improving the efficiency of generating power from this abundant fuel resource and

hence for reducing greenhouse gas emissions from both existing and new power plants.

Current technology for power generation from lignite removes most of the water by

evaporative drying prior to introduction of the coal into the boiler. This results in a

direct loss of the heat of evaporation from the recoverable energy of the system, making

power generation by this conventional approach inefficient. To overcome this

disadvantage new drying processes have been assessed for use in lignite fired power

stations. The technology currently of greatest interest is the mechanical thermal

expression (MTE) process, which removes coal moisture in a liquid state through the

application of thermal and mechanical energy. This is possible because four fifths of the

water in coal is liquid and is held either as free water or bound in small capillaries

(Wheeler and Lui, 2002). In the MTE process, the coal is heated and the water is

maintained in a liquid state throughout the system by application of a pressure above the

saturated steam pressure. A mechanical force is also applied which destroys the internal

pore structure and reduces the reabsorbing capability of the coal. The advantage of MTE

process is that a significant amount of water (around 70%) can be removed from the raw

coal (Favas and Chaffee, 2002) at relatively low temperatures (below 200oC). So MTE

process can play an important role in the power plant for dewatering the coal. It is of

great importance to see the effect of using the MTE lignite on combustion in the

existing furnaces.

1.4. Objectives

The objectives of this research program are to perform:

• A CFD investigation of the aerodynamics of recessed, isothermal burners with a

varying jet velocity ratio (single-phase) without cross-flow and validate the results

with the physical modeling of Perry et al. (1984)

• An Experimental investigation of aerodynamics of rectangular slot-burners by

varying jet velocity ratio in the presence of cross-flow for single-phase flow.

6


• CFD investigation of aerodynamics of rectangular slot-burners by varying jet

velocity ratio in the presence of cross-flow for single-phase flow to obtain further

insights into complexities of recess and near field flow.

• Experimental investigation of two-phase flow of rectangular slot-burners with cross-

flow for different jet velocity ratios

• CFD modeling of lignite in full-scale tangentially-fired furnace and combustion of

MTE lignite in the existing furnace.

1.5. Thesis structure

The current status of relevant aspects of physical and numerical modeling of round,

elliptic and rectangular jets is reviewed at the beginning of chapter 2. The review of

current state of knowledge on lignite and MTE lignite combustion in a full-scale

industrial furnace are described in the later section of chapter 2. Chapter 3 begins by

describing the procedure of solving the governing equations for fluid motion. Some

difficulties associated with the numerical methods are also presented. The detailed

description and dimensions of the burners and the furnace, methodologies and boundary

conditions used in this research program are discussed in the last portion of chapter 3.

The effects of jet velocity ratio on the aerodynamics of rectangular slot-burners without

cross-flow are described in chapter 4. Numerical investigation is performed and the

results are validated with the available experimental data. Cross-flow has a profound

effect in the near-burner region of developing jets. The next chapter (chapter 5)

describes the physical modeling on aerodynamics of rectangular slot-burners by varying

jet velocity ratio in the presence of cross-flow for single-phase flow. The experimental

program was conducted at CSIRO Division of Minerals, Melbourne, Australia and the

results for mean velocity and turbulent statistics are presented. Chapter 6 describes the

CFD investigation of the same rectangular slot-burners of chapter 5. First, the numerical

results are validated with the results obtained from chapter 5 and then more detail of the

aerodynamics of near-burner region of rectangular slot-burners are revealed with the

help of numerical simulation.

7


Chapter 7 presents the experimental investigation of two-phase flow of rectangular slot-

burners with cross-flow for different jet velocity ratios. With the increase in jet velocity

ratio, the jets significantly deviate from the geometric axis of the burner. As the mixing

of air and coal in the near-burner region is important, it is of great importance to locate

the path of the coal particles in the presence of cross-flow.

After successful investigation of single and two-phase flow development, the numerical

modeling of coal combustion in full-scale tangentially-fired furnace has been performed

in chapter 8. In the last section of chapter 8 the aerodynamics and combustion effects of

using MTE lignite in the existing furnace has been discussed. Conclusions and

recommendations are made in chapter 9.

1.6. Perceived specific contributions of the research

This research project contributed to an understanding how nozzle geometry affects the

development of the jets in the near field region of a burner, where the secondary to

primary jet velocity ratio was varied from 1.0. This will provide useful information for

design and modification of the rectangular slot-burners. Mixing of fuel with air in the

near flow field of the developing jet can be better understood which will lead to

efficient combustion in the furnace and ultimately reduce the greenhouse gas emissions.

Experiments were carried out on isothermal burner model to obtain mean velocity and

turbulent statistics for different nozzle geometry and jet velocity ratios. CFD

investigation was performed for isothermal burner model to get further insights into

complexities of recess and near field flow and experimental results were used to

validate CFD results. Conventional lignite combustion in full-scale tangentially-fired

furnace was modeled and compared with MTE lignite combustion. This knowledge can

be used to aid the design of new furnaces for MTE lignite combustion as well as to

make possible improvements in the existing furnaces.

8

Chapter-2

Literature review

Chapter-2 Literature review

2. Literature review

To minimize furnace fouling and pollutant production and to ensure optimum

combustion stability and efficiency when burning lignite, it is necessary to obtain an

improved understanding of the relevant combustion and heat transfer processes.

Combustion in a lignite fired boiler can be divided into two parts: ignition and early

combustion of the pulverized fuel which is controlled by the burner, and the overall

combustion and heat transfer which is controlled by the furnace environment. Although

a number of extensive studies of swirl burner design and operation have been

undertaken (Johnson et al., 1976 and Pleasance, 1980), a comprehensive study of the

operation and basis for optimization of slot-burners in tangentially-fired boilers is yet to

appear in the open literature.

This chapter starts with a literature survey on isothermal single-phase flow development

in slot-burners without cross-flow. Jets in cross-flow are discussed next as cross-flow

has a significant effect in a tangentially-fired furnace. Multi-phase phenomena are

significant in these burner jets, due to the interaction of the gas and solid coal particles,

and differences in particle size and loading between main and vapor burners will

significantly alter behaviour. Two-phase flow development in single, co-axial and

multiple rectangular jets has been reported and finally a review of the current state of

knowledge in modeling coal combustion in full-scale tangentially-fired furnace has been

presented.

2.1. Single-phase flow development

2.1.1. Slot-burners without cross-flow

In 1981 a research program was initiated by Perry to investigate the aerodynamics of a

slot-burner system. This was seen as the first step towards including details of fluid flow

in a general descriptive model of the ignition and early combustion processes in the

burner. It was found from the literature survey (Perry, 1982) that the understanding of

jet flows was not sufficiently advanced to enable the operating characteristics of the

slot-burner jets to be satisfactorily defined. Perry (1981) identified six geometric

characteristics and six flow parameters that may play an important role on the

development of the jet. The six geometric characteristics are:

10


• Thickness of the material separating the jets

• Angle between primary and secondary jet

• Degree of base venting present

• Angle between nozzle exit plane and boiler wall

• Upstream displacement of the nozzle exit plane with respect to the boiler wall

• Upstream duct geometry

The thickness of the base region separating the jets influences the static pressure

between adjacent jets, and therefore the entrainment characteristics. A finite jet velocity

ratio (φ) between jets and a thick base region induce low static pressure and

recirculation vortices (Miller and Comings, 1960) together with a degree of vortex

shedding (Ribero and Whitelaw, 1980). This low pressure tends to draw the two jets

together and increase the mixing between them. Matsumoto et al. (1973) suggested that

for multiple jets the self-preserving state is reached more rapidly than for single jets.

Ribero and Whitelaw (1980) also found that for moderate levels of φ the effect of base

thickness on jet mixing increased with increased φ.

Venting of the base region may occur due to the low base aspect ratio. Hot boiler gases

can be entrained into the base region and can be mixed with most of the circumference

of the primary jet before any mixing occurs between primary and secondary jets. Base

venting suppresses to some degree the influence of low static pressure and flow

recirculation between jets and thus delays the mixing between primary and secondary

jets. Smoot et al. (1975) advocated angles between the two adjacent jets of 30ο to 60ο

for maximum jet mixing. This trend was supported by Stambulernu (1976).

Unfortunately this conclusion from their work was clouded by the fact that they

changed the base thickness at the same time as the angle.

Angle between nozzle exit plane and the boiler wall may play an important role in the

development of the jet. Perry and Hausler (1982) investigated the flow coming from a

rectangular slot-burner perpendicular and at an angle of 60ο to the boiler wall and found

that for the later case the jets deviated from the geometric axis of the burner. Upstream

duct geometry may influence the near flow field of developing jet by altering the

velocity profile at the exit of the nozzle. The effect may be more complex with an

11


asymmetric velocity profile at nozzle exit, which may happen due to bends in the

upstream duct.

Two influences may result from moving the jet exit plane upstream of the boiler wall.

Firstly the base venting effect may be restricted and secondly the initial region of the jet

development may follow the walls of the recessed region. As a result limited early

mixing between the primary jet and the hot boiler gases may occur, together with early

mixing between primary and secondary streams (Perry and Hausler, 1982).

According to Perry (1981), six flow parameters that are important for the development

of jets are:

• Reynolds number effects

• Velocity ratio between secondary and primary jets (φ)

• Core and refractory cooling air

• Velocity profile at the nozzle exit plane

• Secondary to primary gas density ratio

• Temperature of ambient gas (furnace temperature)

For a single round jet the limit above which jet development is independent of the flow

Reynolds number is 2.5x104 (Ricou and Spalding, 1961 and Wall et al., 1980). For

double slots or rows of slots it appears that this limit may be somewhat lower (Marsters,

1977 and Tanaka, 1970). For concentric jets Wall et al. (1979) showed that for φ ≥0.6

the overall entrainment rate in the near field is comparatively independent of φ,

although a slight increase is apparent. Alternately Smoot et al. (1975) suggested a

marked increase in gas mixing rates between jets as a result of increasing φ from 1.0 to

2.0. The interaction between φ and the thickness of the base region would appear to be

most important. For example, as thickness increases in concentric jets (Champagne and

Wygnanski, 1971; Chigier and Beer, 1964; Ribero and Whitelaw, 1980) recirculation on

the centreline axis, in the near field, occurs at lower φ.

The review of the literature (Perry, 1982) revealed little information directly relevant to

the rectangular jets. As a result Perry and Hausler (1982, 1984) and Perry et al. (1986)

12


performed a series of experiments to investigate the influence on isothermal jet

development of secondary to primary jet velocity ratio, burner geometry and burner exit

velocity profiles. Four simple burner geometries were characterized including a single

near rectangular jet, three jets discharging at right angles to the furnace wall (geometry

A), the three-jet system discharging at an angle of 60° to the wall with the jet dividers

terminating at the furnace walls, geometry B (figure 3.3(a)), the same as geometry B but

terminating a short distance upstream of the furnace walls, geometry C (figure 3.1), and

the same as geometry C but with divergent recess walls, geometry D (figure 3.2 &

3.3(b)). The four basic geometries were evaluated with the secondary jets oriented to

discharge parallel with the primary stream and then modified to discharge at an angle of

30ο converging toward the primary. Experimental data covered a wide range of

secondary to primary jet velocity ratios and included flow visualization observations,

transverse velocity profile and static pressure measurements. From those experiments

they concluded that burner geometry can significantly influence the jet development and

that for a typical velocity ratio of 3:1 used in combustion systems, significant deviation

of the jets can occur from the geometric centreline. These experimental datum were

used to validate the numerical results of Hart (2001). He studied three types of burner

geometries (A, B and D) in detail with the aid of CFD and gave a detailed account of

the near field mixing and flow development mechanisms in jets from complex burner

geometries but only for a velocity ratio of 1.0. No such detailed CFD investigation has

been performed on these burners for φ other than 1.0.

2.1.2. Jets in cross-flow

2.1.2.1. Single round jet

The topic of jets issuing into deflecting streams has been the subject of numerous

studies because of their common occurrence in engineering problems; chimney plumes

for the dispersion of pollutants in the atmosphere, the cooling of turbine blades, lifting

jets for vertical stall take-off landing aircraft and combustion of coal in tangentially-

fired furnaces are just a few important examples. Many researchers have studied a

circular jet in cross-flow in detail. Foss (1980), Andreopoulos (1982, 1985),

Andreopoulos and Rodi (1984) reported on an extensive investigation of the near field

aerodynamics of a round jet issuing normal to the surface and to the cross-flow. The

ratio of jet to cross- flow velocity (ψ) was varied between 0.25 and 3.0. Andreopoulos

13


(1982) studied the upstream influence (figure 2.1) into the jet pipe of the interaction

between the jet and cross-flow. He found that for low ψ a non-uniformity in the pipe

extended some three diameters upstream and reduced with the increase in ψ.

(low ψ) (high ψ)

Figure 2.1: Flow configuration for low and high jet to cross-flow velocity ratio

Andreopoulos and Rodi (1984) noted that for high ψ, the near field of jets in a cross-

flow was controlled largely by complex inviscid dynamics and further downstream the

flow was always influenced by turbulence. For small ψ, the near field was controlled

both by complex inviscid dynamics and turbulence. From their investigation they

concluded that at low ψ (figure 2.2(a)), the jet was abruptly bent over by the cross-flow

and that with an increase in ψ, the jet penetrated further into the cross stream and the

bending became less abrupt (figure 2.2(b)).

Figure 2.2(a): Flow development for low (ψ=0.5) jet to cross-flow velocity ratio

14


Figure 2.2(b): Flow development for high (ψ=2.0) jet to cross-flow velocity ratio

Catalano et al. (1989) used experimental methods and performed numerical simulation

to investigate the development of a system for ψ equal to 2.0 and 4.0 where the cross-

flow was confined between two parallel surfaces. The two-equation turbulence model

used, when compared with experiment, predicted the downstream flow field

satisfactorily but showed poor agreement in the near field due to the turbulence being

highly anisotropic.

Sherif & Pletcher (1989), surveyed numerical and physical modeling studies of jets in

cross-flow and considered that these systems were, generally, more difficult to model

numerically than wall boundary-layer flows primarily because of the curvature of the

shear layer and the complex turbulent flow pattern in the jet wake region. They

undertook mean velocity and turbulent intensity measurements in this type of flow for

ψ=2.0, 4.0 and 6.0 and identified that the resultant mean velocity profiles were

characterised by two peaks. The larger peak occurred in the jet core, while the smaller

one corresponded to fluid entrained into the jet wake behind the vertical portion of the

jet. The location of both maxima was a strong function of the velocity ratio.

Sykes et al. (1986) developed a time marching solution of the incompressible Navier-

Stokes equations. The model was used to investigate the details of the flow within the

jet in cross-flow. They found that for high ψ, the source of the streamwise vorticity in

the vortex pair can be readily tracked back to the original streamwise vorticity in the

sides of the vertical jet. For lower ψ (ψ<4.0), the vertical component of vorticity at the

source is important.

15


Lester et al. (1999) reported on a series of large-eddy simulations of a round jet issuing

normally into a cross-flow. Simulations were performed for ψ=2.0 and 3.3, and for two

Reynolds numbers, 1050 and 2100, based on cross-flow velocity and jet diameter. Mean

and turbulent statistics were computed from the simulations and were compared with

experimental measurements. Large-scale coherent structures observed in experimental

flow were reproduced by the simulations, and the mechanisms by which these structures

form were described. The effects of coherent structures upon the evolution of mean

velocities, resolved Reynolds stress, and turbulent kinetic energy along the centre plane

were discussed.

Lim et al. (2001) looked at the vortical structures of jet in cross-flow in water by

releasing dye at locations around the jet exit. The results showed that there were no

evidence of ring vortices in jet in cross-flow, and the postulation that vortex loops were

formed from the folding of the vortex rings did not reflect the actual flow behaviour.

They concluded with the results that the vortex loops were formed directly from the

deformation of the cylindrical vortex sheet.

It is well established from all of these investigations that a circular jet in cross-flow

produces a multitude of vortical structures and the five most significant ones are the

leading edge vortices, lee-side vortices, counter-rotating vortex pairs, horseshoe vortices

and wake vortices (figure 2.3).

Figure 2.3: Structural features of jets in cross-flow (adapted from Johnston & Khan,

1997)

16


2.1.2.2. Multiple round jets

There are substantially fewer papers dealing with studies of multiple round jets in cross-

flow compared to those dealing with a single round jet in the same environment.

Examples of the papers dealing with multiple round jets include Isaac & Schetz (1982),

Makihata & Miyai (1983), Isaac & Jakubowski (1985), and Savory & Toy (1991).

Multi-jet configurations studied included two or three jets aligned in a row transverse to

the cross-flow direction, two jets in tandem and three jets each located at a corner of an

equilateral triangle. Velocity ratios between the jets (φ) equaled one, and between the jet

and cross-flow (ψ) range from 2.0 to 10.0. For the transverse jet orientations, jet spacing

ranged from 0.875 to 7.5.

Savory and Toy (1991) undertook a real time video analysis of two transversely aligned

jets for ψ=6 to 10 and jet spacing 1 to 5 nozzle diameters. They found that as with the

fully developed single jet in cross-flow, the combined jets showed the characteristic

kidney shape with the presence of only two outer lobes, which suggested that the inner

vortices of each pair either did not form or were short-lived. It was deduced that for

close spacing between the jets, the cross-flow no longer passed between the two jets. In

this case the inner vortices of the two counter rotating pairs did not appear to form, or

were extremely weak, with the initial jet vorticity being turned and concentrated in the

outer vortices only. These findings agree with the case of a row of holes of multiple

side-by-side jets, studied by Kamotani & Greber (1974), who found that at nozzle

spacing below about 3.0 the configuration behaved like a two-dimensional jet, with less

entrainment and a resultant increase in penetration.

The evolution of jets emanating from short holes into cross-flow at low ψ was presented

by Peterson and Plesniak (2004). The jet fluid issued into the cross-flow from the wind

tunnel floor via five span wise-collinear injection holes and the measurements were

performed on the central hole. The hole spacing of three diameters was similar to a

typical film-cooling configuration. Particle Image Velocimetry (PIV) was used to

determine structural features of the jet/cross-flow interaction through out its

development from within the jet supply channel, through the injection hole, and into the

cross-flow. The effect of supply channel feed orientations, i.e. counter to, or in the same

direction as the cross-flow was emphasized (figure 2.4). They found that feed

17


orientation profoundly affected jet characteristics such as trajectory and lateral

spreading, as well as its structural features. In the co flow supply channel geometry a

pair of vortices existed within the hole with the same sense of rotation as the primary jet

counter rotating vortex pair (CRVP). In contrast, the counter flow supply channel

configuration had in-hole vortices of opposite rotational sense to that of the CRVP.

Figure 2.4: Supply channel flow configurations.

2.1.2.3. Non-circular jets

In recent years, various attempts have been made to improve the mixing efficiency of a

jet in cross-flow by using non-circular nozzle geometries such as an ellipse, square and

rectangle. Haven & Kurosaka (1997) examined the effect of hole exit geometry on the

near field characteristics of cross-flow jets. Hole shapes investigated were round,

elliptic, square and rectangular, all having the same cross-sectional area. Laser-induced

Fluorescence (LIF) and particle image velocimetry were used. Their study was confined

to low ψ only (0.4 to 2) since they were concerned primarily with the effect of jet

geometry on film cooling. The main finding of their investigation was the existence of

“double-decked kidney and anti-kidney vortices. They found that the lower-deck

structures were kidney shaped and established to be the primary counter rotating vortex

pair inherent in all jets in cross-flow, while the upper deck structures were dependent on

the jet shape and could be either unsteady kidney shaped or unsteady anti kidney

shaped. A similar investigation was conducted by New et al. (2003) to study the flow

structures of an elliptic jet in cross-flow in a water tunnel using laser-induced

fluorescence and for a range of jet aspect ratios from 0.3 to 3.0, jet to cross-flow

18


velocity ratios (ψ) from 1 to 5, and jet Reynolds numbers from 900 to 5100. The results

showed that the effects of aspect ratio (or jet exit orientation) were significant only in

the near field, and diminished in the far field.

Findlay et al. (1999) examined the flow field characteristics of three different

geometries of multiple rectangular jets in a cross-flow at various ψ by wind tunnel

measurements. The geometries considered were: perpendicular, streamwise-inclined,

and spanwise-inclined jets as shown in figure 2.5. The inclined jets were at an angle of

30º to the wind tunnel floor. Mean velocity and turbulence measurements along with

film cooling effectiveness and scalar transport data were obtained. Jet to cross-flow

velocity ratios (ψ) of 1.5, 1.0 and 0.5 were used. They found that flow field at the jet

exit was strongly influenced by the cross-flow as well as by the inlet conditions at the

entrance to the jet orifice.

Figure 2.5: Coordinate definitions for (a) perpendicular, (b) streamwise, (c) span wise

jets.

Yan & Perry (1994) investigated multiple rectangular jets in the presence of cross-flow.

The cross-flow was the representation of the burner flow field similar to a tangentially-

19


fired furnace. Geometry B and D, which have been described earlier in this chapter,

were used for their investigation. The velocity ratio between the secondary to primary

jet (φ) was 1.0 and 3.0 and primary to cross-flow jet (ψ) was 1.0. The burners were

inclined at an angle of 60ο to the wall and the direction of the cross-flow was parallel to

the furnace wall, firstly in a direction counter to the direction of the wall component of

the burner exit velocity then in the same direction. They studied flow visualization and

took measurements for mean velocity using laser doppler anemometry (LDA) in the

near field region. They concluded that the presence of cross-flow in either direction had

a significant effect in the near field region of the developing jet. In the presence of

cross-flow the jets deviated significantly from their geometric centreline. They found

counter rotating vortex pair for both geometry B and D when the direction of the cross-

flow was opposite to the wall component of burner exit velocity.

2.2. Two-phase flow development

Many authors consider that when particles are small a particle-laden gas flow may be

considered as homogenous and jet behaviour will be similar to that of a single-phase jet.

This may apply for volumetric and mass fractions of particles to air of much less than

one (Melville and Bray, 1979). Many researchers (e.g. Hetsroni and Sokolov, 1971;

Laats, 1966; McComb and salih, 1978) have observed a reduction in axial velocity

decay and jet growth rates with the introduction of a liquid droplet or solid particle

phase into an air jet. Experimental studies have shown that the decay rate of particle

axial velocity and spreading rate is less than that of the fluid in the two-phase jet flow

(Ivanov et al., 1970; Laats and Frishman, 1970; McComb and Salih, 1978).

Laats and Frishman (1970) found that for a given particle mass loading, increasing

particle size increased the axial decay rate of jet velocity near the nozzle and for a given

particle size, an increased loading reduced the axial decay rate. They concluded that the

inertia of the additive had a suppressing effect on turbulent momentum transport

processes, which became greater with increased concentration and smaller with

increased particle size. Care needs to be taken in setting up the initial conditions when

studying two-phase jets. For example, Laats and Frishman (1970) noted that an initially

fully developed velocity profile for a jet tended to become uniform as the additive load

was increased. Also it was observed by Field, 1963 and Popper et al., 1974 that the

particle velocity could be less than the fluid velocity at the nozzle exit particularly for

20


denser particles. Non-equilibrium conditions at the nozzle exit plane can increase the

rate of spread of the two-phase jet (Elsharbagy et al., 1974).

Recent developments in Laser Doppler velocimetry (LDV) technology, particularly in

data recording and analysis electronics and software, have enabled researchers (e.g.

Modaress et al., 1984; Fleckhaus et al., 1987; Barlow, 1990) to simultaneously measure

fluid and particles velocities. Modaress et al., 1984, summarized the effects of the

dispersed phase on the development of the jet in comparison to a single-phase turbulent

jet as follows:

• The expansion rate of a two-phase jet is smaller to that of a single-phase jet.

• The centreline velocity decay of both phases in the two-phase jet is smaller than that

of a single-phase jet.

• The velocity fluctuations are reduced with increase in the initial mass loading.

• The Reynolds shear stress for a two-phase jet is smaller than that of a single-phase

jet.

They also used two-color LDV to investigate the far-field development of a co-flowing

jet where the jet was mounted downwards in the surround airflow. The results indicated

that the presence of the particles reduced the gas phase velocity fluctuations and the

Reynolds shear stress. A similar result was found in the work of Mostafa et al. (1990).

The experiment was performed on a particle-laden co-axial jet. They concluded that the

high velocity of the outer stream caused a rapid increase in the axial velocity of the

inner jet downstream of the exit plane because of the transfer of mass and momentum

from the external to the inner stream.

Fleckhaus et al., (1987) found that for a given mass loading, particle size had a

significant effect on decay rate, turbulent kinetic energy and shear stress for the gas

phase. Park and Chen (1989) measured the flow field development of confined, co-axial

jets for single-phase flow and for two-phase flow where particles were added to the

central jet. Experimental conditions considered included two particle loadings and

secondary to primary jet velocity ratios of 1.0 and 0.5. The results from the single-phase

flow study were compared with a numerical prediction employing the k-ε turbulence

21


model; the prediction overestimated the rate of flow development and underestimated

the turbulence intensities the recirculation length between the jet and the confining wall

For the two-phase flow measurements the presence of particles reduced the rate of fluid

velocity decay and increased the length of the recirculation region. The gas-phase mean

and fluctuating velocities were higher in the near field but were lower downstream.

Schefer et al. (1986) examined conditional sampling of velocity in turbulent non-

premixed, non-reacting propane co-flowing jet with LDV. Particles were independently

seeded in the fuel jet flow and co-flowing air so that the fluid originating from the jet

could be distinguished from the co-flowing air. They found that the velocity of particles

seeded in the jet was consistently higher than velocity of particles seeded in co-flowing

air at the same axial distance and near to the jet exit axial velocity fluctuations of the co-

flowing air were higher than those of the jet.

Yan and Perry (1994) first observed the two-phase flow development in rectangular

slot-burners with and without cross-flow, for geometry B and D, which have been

described earlier in this chapter. The velocity ratio between the secondary to primary jet

(φ) was 1.0 and 3.0 and in case of cross-flow, primary to cross-flow jet velocity (ψ) was

1.0 and kept constant throughout the experiments. The burners were inclined at an angle

of 60ο and the direction of the cross-flow was parallel to the furnace wall. Firstly in a

direction counter to the direction of the wall component of the burner exit velocity then

in the same direction. They observed flow by visualization and took the measurements

for mean velocity in near field region. Measurements of gas and solid phase velocities

were made using LDV. To stabilize the particle feed and to make the mean free paths

more representative, the mass loadings were reduced from typical furnace levels of the

order of 0.38 to model values ranging from 0.027 to 0.2, depending upon the particles

used. Three particle sizes were used in their experimental program as follows:

• Hollow glass spheres, 100µm diameter and 137kg/m3 density

• Solid glass spheres, 100µm diameter and 1450kg/m3 density

• Alumina oxides, 50µm diameter and 580kg/m3 density

22


For both types of cross-flow orientation they concluded that the solid phase appeared to

be moving slower than the gas phase close to the burner exit. Farther downstream the

spread of the solid phase appeared wider and turned slower than the gas phase under the

deflecting influence of the cross-flow. Due to time constraints, Yan and Perry (1994)

did not take any measurement for geometry D in the presence of cross-flow.

2.3. Modeling of coal combustion in tangentially-fired furnace

2.3.1. CFD as a modeling tool

CFD is increasingly accepted as an effective tool to model and study a broad range of

industrial problems involving fluid flows and heat transfer, from air conditioning and

ventilation to the power generation, minerals processing, aerospace and automobiles

industries. The advent of powerful computers and commercial CFD packages, with the

ability to generate body-fitted computational grids and multi-blocking facilities, has

opened up a new-era in CFD applications enabling modelling and optimisation of many

complex geometries and processes where, until recently, such modelling was dependent

on simple empirical relationships derived from traditional knowledge and experience

gained in a similar operating environment.

CFD as an investigative tool, when used carefully with a complete realization of its

limitations, can enhance the existing knowledge of most of these processes since it can

easily provide detailed information on various important parameters under different

operating conditions. It is particularly useful when the focus is on improving the current

technology by optimizing present operating conditions such as generating new

optimized designs of critical industrial components, or minimizing the level of

pollutants generated in a power plant operation. While it can supplement physical

modeling and pilot studies in the form of extending the range of parameters analyzed, it

also offers an attractive option in scaling-up of the geometry analyzed as this can be

executed with a minimum of additional cost of computational time. CFD is also a viable

design tool in the long term as it continues to benefit from rapid development in the

computing technology leading to much reduced processing times (often by many orders

of magnitudes) and advances made in the computational techniques improving the

capability and reliability of the current techniques increased emphasis on

comprehensive validation of the numerous physical and chemical models incorporated

and an extensive industrial experience gained over a long period of time.

23


In CFD, solving a particular problem generally involves first discretizing the physical

domain such as the interior of turbine engine or the radiator system of a car. On the

discretized mesh the Navier-Stokes equations take the form of a large system of

nonlinear equations. Going from the continuum to the discrete set of equations is a

problem because conservation of mass has to be maintained in the discrete equations. At

each node in the mesh, between 3.0 and 20.0 variables are associated such as pressure,

two or three velocity components, density, temperature, etc. Furthermore, capturing

physically important phenomena such as turbulence requires extremely fine meshes in

parts of the physical domain. The detailed discretization scheme and solution procedure

for the system of nonlinear equations is discussed in chapter 3.

In literature, there are two approaches in modelling two-phase flow. These are the

Eulerian and the Lagrangian approach. The Eulerian approach allows the particle flow

field to be treated as a continuum and is solved in a similar manner to the gaseous flow

field. Each particulate property, such as temperature and size is averaged over the entire

cell and the Eulerian partial differential equations are solved for each particle size group

for the whole computational field. Hence the main shortcoming of this method is the

large computational storage requirement.

The Lagrangian technique is based upon the tracking of individual particles by

considering the forces acting upon them. The resultant equation of motion (equation

2.1) is then solved.

gA)(C21

dt)(d

PPPPDPP mUUUUUm

+−−= ρ (2.1)

Lagrangian approach has many advantages over the Eulerian approach including the

ability to predict trajectories of single particle and the possibility of handling particle-

wall interactions.

2.3.2. Tangentially-fired furnace

In tangentially-fired furnaces, the burner sets are typically located either on the furnace

wall or at the corners and are angled to fire tangential to an imaginary circle whose

centre is located at the furnace centre. The momentum in the burner jets induces a

rotational flow in the furnace, which provides a long combustion path, ensuring

24


intensive mixing of combustion products and the fuel stream and improved combustion

stability. Investigations of full-scale power stations for pulverized coal combustion by

means of numerical simulation were presented by Abbas and Lockwood (1986), Boyd

and Kent (1986), Epple and Schnell (1992), Fiveland and Wessel (1988),

and Jacobsen (1992), Lendt (1991), Wirtz (1989), Epple et al. (1995), Mann

and Kent (1994) and others.

ldmannaKj &&

2.3.3. Coal combustion scheme

Coal combustion is a complex process and not all physical aspects are well understood.

The principle steps of the reaction progress are the thermal decomposition of the raw

coal and the subsequent burnout of the char and the volatile matter. The reaction scheme

needs to be simplified a great deal in order to perform simulations of coal combustion

within the current limitations of computing power.

2.3.3.1. Coal devolatilization

Devolatilization is the process where pulverized coal particles under elevated

temperature conditions release volatile matter in the form of tar, CO, and other

combustibles. These volatiles escape through the pore structure of the coal particles,

often destroying the original pore structure.

Devolatilization Rate: The devolatilization rate of a particle of coal depends upon the

rate of heating as well as on the instantaneous temperature. Initially Lockwood et al.

(1980) found that the combustion performance is not greatly sensitive to the

devolatilization rate and a complex description is not warranted at least with all the

other limitations of combustion modelling. Later, investigation revealed that coal

devolatilization plays an important role in the near burner region. Lockwood et al.

(1998) used a mathematical model to predict the pulverized coal combustion for two

swirl burners and found that the volatiles burning rate was essentially controlled by the

rate of mixing of pulverized fuel and air.

Volatile Content: Lockwood et al. (1980) performed a simulation for which the volatile

content was reduced to zero to model an anthracite coal. It was found that combustion

could not be sustained. Exceptionally, radiation transfer was not calculated for that run

25


and it is not known whether back radiation or radiation from the walls would be

sufficient to stabilize the flame or whether redesign of the burner to create local

recirculation would be necessary. Nonetheless, it is a fact well known by any pulverized

fuel furnace operator that the higher the volatile content of the coal the more easily

stabilised is the flame. (Lockwood et al., 1980).

Final Volatile Yield: The simple model to calculate the final volatile yield is the single

reaction model developed by Badzioch and Hawksley (1970). The coal is considered to

have fixed fractions of volatiles, char and ash. The rate of production of the volatile

gases is given by the first order reaction:

( VV−= fV Vk

dtd ) (2.2)

where V is the mass of volatiles, which have already evolved, from unit mass of raw

coal, and Vf is the final yield of the volatiles. The rate constant kV is expressed in

Arrhenius form as;

⎟⎟⎠

⎞⎜⎜⎝

⎛−=

P

VVV R

EexpAk T (2.3)

where TP is the temperature of coal particle (assumed uniform), AV and EV are

constants, determined experimentally for the particular coal. And R is the universal gas

constant.

An alternative to the single reaction model is the two competing reaction model of

Ubhayakar et al. (1976). This model considers the effect of temperature and heating rate

on volatile yields. Yan Ping Zhang et al. (1990) performed an experiment to estimate

the final volatile yield. They concluded that the final volatile of pulverized coal is

insensitive to heating rate whereas a number of studies have demonstrated that ultimate

volatile yield is sensitive to final temperature. Winter et al. (1997) found that coal with

high volatile content causes short burnout times because most of the carbon is released

rapidly during devolatilization and only a little char is left to burn relatively slowly.

26


2.3.3.2. Gaseous combustion models

In order to include the effects of combustion within the compressible flow equations it

is necessary to prescribe the density in terms of the fuel and oxidant mixing. In a fast

chemical reaction model, it is assumed that if fuel and oxidant are simultaneously

present at the same point then an instantaneous reaction occurs producing combustion

products. Fuel and oxidant are assumed to combine in a fixed stoichiometric ratio, i,

such that:

1 kg fuel + i kg oxidant -> (1 + i) kg products (2.4)

The mixture fraction f for the reaction can be defined by

OF

Oχχ

χχ−

−=f (2.5)

where

iO

Fmm −=χ (2.6)

and m is the mass fraction and subscripts F and O refer to fuel and oxidant respectively.

So the Oχ and Fχ used in equation 2.5 are Oχ = -i1 and Fχ =1.

The mean value of the mixture fraction, F, satisfies a conservative transport equation of

the form

)( UFρ•∇ 0))((LT

T =∇+•∇− Fσµ

σµ

(2.7)

Here ρ is the fluid density, U is the mean fluid velocity, µ and µT are molecular and

turbulent fluid viscosities and σ L and σ T are equivalent Prandtl numbers.

By definition, f is always positive and attains its stoichiometric value FST when 0=χ ;

thus FST i+

=1

1 (CFX User Guide, 1997).

Mixed-is-burnt Model

The mixed-is-burnt model assumes that fuel and oxidant cannot co-exit simultaneously.

The instantaneous mass fractions are given in terms of the instantaneous mixture

27


fraction by the following relationships. When f>FST, the mixture consists of fuel and

products such that:

FFm

ST

STf−−

=1F , 0O =m , mm FP 1−=

when f<FST , the mixture consists of oxidant and products with

mmF

mmST

fOPOF 1,1,0 −=−== (CFX User Guide, 1997)

Eddy-Break-Up (EBU) Model

A popular way to calculate turbulent gas combustion is the use of an Eddy Break-Up

Model. This model, first proposed by Spalding (1972) and modified by Magnussen

(1989), is based on the turbulence decay and assumes infinite-fast-chemistry. For the

EBU model an explicit equation is solved for the mass fraction of fuel:

( ) limARFLT

TF MCC

kεmUm ρ

σµ

σµρ −=⎟⎟

⎠

⎞⎜⎜⎝

⎛∇⎟⎟

⎠

⎞⎜⎜⎝

⎛+•∇−•∇ (2.8)

the terms and AR C,C limM are modelled as:

⎪⎪⎩

⎪⎪⎨

⎧

⎟⎟⎠

⎞⎜⎜⎝

⎛=

model mixingcollision 4.0,

model mixing viscous,6.23C41

2R kρµε

⎪⎩

⎪⎨

⎧≥=

chemistry rate finiteD 0.0,chemistry rate finite D ,0.1chemistry rate infinite ,0.1

C

ie

ieA

πDD

( )⎪⎪⎩

⎪⎪⎨

⎧

⎟⎟⎠

⎞⎜⎜⎝

⎛+

⎟⎠⎞

⎜⎝⎛

=rmproduct te a with

i1B,

i,min

rmproduct te aout withi

,minM

PROF

OF

lim mmm

mm (2.9)

The DamkÖhler number D is defined by

CH

eττ

≡D (2.10)

where ε

τ k≡e and CHτ is the chemical induction time:

bO

aF

T

CHCH )()(A2

A

mme T ρρτ = (2.11)

28


where ACH is a rate constant, TA is an activation temperature, a and b are exponents for

the fuel and oxygen density respectively and Die is the ignition/extinction value for D.

The mass fractions for oxidant and product are defined as follows:

PCST

FFO 1 m

FmFmm −

−−−= (2.12)

PCOFPR 1 mmmm −−−= (2.13)

where is the mass fraction of the char product. (User Guide, 1997) PCm

The inclusion of reaction kinetics through the EBU approach was found to significantly

improve the solution accuracy in modelling combustion of low calorific value gases

especially in an inert atmosphere.

2.3.3.3. Char combustion models

The most important coal property relevant to combustion is the chemical reactivity of

the char, quantified by the kinetic parameters for char oxidation (Jalaluddin, 1991).

When pulverized coal particles are injected into a hot gaseous environment, two

processes occur, often with considerable overlap, the first of which is the thermal

decomposition of the coal, producing combustible gases (volatiles), and the second

involves the slower reaction of the residual char with the oxidant. It is the comparatively

slower rate of char oxidation that determines the overall combustion efficiency.

Gas temperature and oxidant concentration influence the rate of combustion of char

particles. Early researchers (Walker, 1959 and Wicke, 1955) characterized the rate

controlling mechanism in this process into three zones depending upon temperature and

particle size. Zone I, the low temperature zone, is associated with chemically controlled

char combustion. In Zone II pore diffusion coupled with chemical reaction as a

controlling mechanism. Under Zone III conditions the temperatures are high enough to

make chemical kinetics very fast and the rate of char mass loss is limited by the rate of

which gas diffuses from the surrounding atmosphere to the particle surface. Several

researchers (Walker 1959, Charpenay 1992 and Mitchell 1982) have noted that

temperatures in a furnace correspond to Zone II and Zone III combustion. At these

temperatures low oxygen concentrations on the internal reacting surface of the char

limit combustion. Simulations of full-scale tangentially-fired furnaces (Chen et al,

1992) predict that particles pass through regions where oxygen concentrations are low.

29


A char burnout model based only on oxygen molecule concentrations predicts low rates

of char mass loss in these regions. Concentrations of other species, such as carbon

dioxide and water, are often relatively high in oxygen deficient regions. A char

combustion model that neglects oxidation reactions with these species may under

predict burnout. This was considered possible because the oxygen concentration falls,

while the concentration of H2O and CO2 rise during burnout. Mann and Kent (1994)

used a mathematical model of a corner-fired furnace to examine the burnout of an

Australian bituminous coal. The H2O and CO2 gasification reactions were included in

the reaction scheme for the char reacting in Regime II. The effect was especially

important in oxygen-deficient regions near the furnace walls. The findings were

contrary to the conventional assumption that there is no effect of water vapor on char

gasification reaction (Stanmore and Visona, 1998).

Char Burnout Model

Field et al. (1967) proposed a char burnout model based on a global first order reaction.

In Field’s model, a char particle is considered to be spherical and surrounded by a

stagnant boundary layer through which oxygen must diffuse before it reacts with the

char. The oxidation rate of the char is calculated on the assumption that the process is

limited by the diffusion of oxygen to the external surface of the char particle and the

effective char reactivity. An alternative is Gibb’s model (1985), which takes into

account the diffusion of oxygen within the pores of the char particle. The oxidation

mechanism of carbon can be characterised by molar ratio φm of carbon atoms/oxygen

molecules involved in the oxidation process so that oxides are produced according to

the equation

φmC+O2→2(φm-1)CO+(2-φm)CO2 (2.14)

The value of φm is assumed to depend on the particle temperature TP:

)T

exp(A2

)1(2

P

Ss Tm

m−=

−

−

φφ (2.15)

where the constants are given by Gibb as AS=2500 and TS=6240K.

By solving the oxygen diffusion equation analytically, the following equation is

obtained for the rate of decrease in the char mass mPC:

1132

11

chO

cPC ))kk(k(MM

13

dtd

2

−−−∞ ++−

−=ρρφ

em m

(2.16)

30


The far field oxygen concentration ∞ρ is taken to be the time-averaged value obtained

from the gas phase calculation and ρch is the density of the char. Physically, k1 is the

rate of external diffusion, k2 is the surface reaction rate and k3 represents the rate of

internal diffusion and surface reaction. These are defined as follows:

2EXT

1Dk

r= (2.17)

where DEXT is the external diffusion coefficient of oxygen in the surrounding gas;

2C

2k)e1(kr

−= (2.18)

where e is the void fraction of the char particle and kC is the carbon oxidation rate,

defined by the modified Arrhenius equation

)Texp(Ak CCC

PP T

T −= (2.19)

(2.20) aβ/)1βcothβ(kk 2C3 −=

where a is the particle volume/internal surface area ratio. The term β is defined as

5.0

INT

C )eaD

k(β r= (2.21)

Gibb recommended a value for the effective internal diffusion coefficient DINT an order

of magnitude less than DEXT.

2.3.3.4. Radiation models

An effective numerical representation of radiant heat transfer is essential for meaningful

mathematical modeling of furnace combustion. Radiation is the dominant mode of heat

transfer within most power station boilers, and is especially important here considering

that the proportion of total heat transfer due to radiation increases with increasing

combustor size, and that lignite fired furnaces are typically much larger than other

furnaces.

The governing radiant heat transfer integro differential equation for an absorbing and

scattering medium is:

( ) ( ) ⎥⎦⎤

⎢⎣⎡+−

⎥⎥⎦

⎤

⎢⎢⎣

⎡−+⎥

⎦

⎤⎢⎣

⎡−= ∫ ΩΩ

πππ ΩΩ dIpIIIdsdI

kkE

kE

k ssp

pg

g''

4

,41 (2.22)

31


where I is the intensity of radiation in the direction Ω, s is the distance in the x

direction, kg is the gas absorption coefficient of the gaseous medium, , kP and, ks are the

particle absorption and scattering coefficients, and p(Ω,Ω’) is the probability that

incident radiation in the direction Ω’ will be scattered into the increment of solid angle

dΩ about Ω.

The first widely used radiation modelling technique for combustion systems was the

zonal method, as described by Hottel and Sarofim (1967), where the furnace enclosure

is discretized into surface and gaseous volume elements of assumed uniform

temperature and radiative properties. Exchange factors are calculated between the

surface to volume, surface to surface and volume to volume elements from an equation

involving the relative separation and orientation of the elements and an exponential

attenuation involving the absorption properties of the gas. The zonal method for

radiative heat transfer calculation becomes impractical due to high computational

storage requirement. Additionally, this method is not well suited to complex or

generalized geometries.

Monte Carlo techniques (Howell and Perlmutter, 1964) provide a statistical solution

well suited to non-standard geometries, where rays of radiation are traced in randomly

generated directions, reflecting off any encountered boundaries until they become

extinct due to attenuation. The number of rays required to describe the process

accurately can be very large making this technique less attractive to the typically large

geometries encountered in coal combustion.

Flux methods separate the dependency of radiation intensity from the spatial

dependency by discretization of the intensity into vectors representing intervals of the

solid angle. The radiant transfer equation (2.22) is reduced from the integro differential

form to a series of coupled linear differential equations obtained by integration of

equation (2.22) over the solid angle. The system of equations can then be solved to

obtain the radiant heat transfer. Due to the spatial distribution of production intensity,

found in the common flux methods used in furnace combustion models using six free

parameters (Gosman, Lockwood & Salooja, 1978), insufficient accuracy was obtained.

Further discretization of the intensity is then required to improve the performance of the

32


radiation model, which is difficult to generalise and significantly reduces the

computational advantage of this technique.

The ‘Discrete Transfer Model’ (Lockwood and Shah, 1981) is essentially a hybrid

technique of the zonal, Monte Carlo and Flux methods. It uses a deterministic rather

than statistical method of tracing representative rays between two surfaces throughout

the furnace domain, where each surface element in the furnace enclosure is represented.

This technique efficiently ensures that a reasonable coverage of the enclosure is

obtained, and is readily adaptable to generalized geometries. The precision of the

technique is governed by the level of discretization of the intensity at each element and

by the number of iterations performed to establish consistent values of radiant flux. A

disadvantage of the technique is that the method, in common with the other discrete

ordinate techniques, has an approximation error due to ‘ray effects’. These effects yield

anomalies in the scalar flux distribution which may be of significance where there are

localized radiation sources in the gas and scattering is of less relative importance than

the absorption, which usually occurs in flames. However the discrete transfer method is

a relatively economical and flexible method for a general numerical code.

2.3.3.5. NOx models

The NOx formation involves modeling of both thermal and fuel NOx. The fuel NOx

mechanism is generally accepted as being the major source of NOx emissions from a

pulverized coal flame. Epple and Schnell (1992) modeled the evolution of fuel–bound

nitrogen and homogeneous reactions of intermediate nitrogeneous gas-phase species by

using a probability density function approach for fluctuating temperature. The NOx

model involves solving two transport equations for the mass fraction of NO and HCN.

In the model the NOx formation mechanisms are reduced to a small number of global

reactions with rate determined by comparison with experimental data. The thermal NOx

mechanism is modeled by assuming that oxygen radicals are in chemical equilibrium

with CO, CO2 and O2. The fuel NOx mechanism is represented by: a reaction from the

volatiles and char products to HCN, a NOx production reaction from HCN to NO and a

NO depletion reaction with NO and HCN combining to give N2. The transport

equations for NO and HCN are

33


NONOLT

TNO R))(()( ρ

σµ

σµρ =∇+•∇−•∇ mUm (2.23)

HCNHCNHCN R))(()( ρσµ

σµρ =∇+•∇−•∇ mUm

LT

T (2.24)

where RNO and RHCN are the reaction rates (CFX User Guide, 1997) are evaluated at the

mean temperature and concentrations of the reacting small-scale (Kolmogorov micro-

scale) turbulent eddies.

To understand the mean and turbulent statistics in near and far field region of

developing jets more experiments and numerical simulations on rectangular slot-burners

are required, especially for geometry D as there is no data for geometry D in the

presence of cross-flow. The flow pattern in tangentially-fired furnaces is very complex

and needs detailed investigation. The aim of this research program is to produce more

data on mean velocity and turbulent fluctuations to reveal the near flow field

aerodynamics of rectangular slot-burners in tangentially-fired furnace.

34

Chapter-3 Mathematical models, modeling techniques and

methodologies

Chapter-3 Mathematical models, modeling techniques and methodologies

3. Mathematical models, modeling techniques and methodologies

CFD is a method of obtaining a solution to the Navier-Stokes equations using an

iterative calculation procedure. The Navier-Stokes equations are derived from the

principle of conservation of mass and momentum, and in their most general form are

capable of describing all fluid flows. This chapter describes the governing equations for

fluid motion and the numerical methods used to solve these equations. The description

covers the discretization methods, schemes, difficulties associated with the solution

procedure and methods of overcoming them. An overview of the turbulence models

currently available in computational fluid dynamics is also discussed. A detailed

description of the boundary conditions for the models is presented, along with issues

related to meshing and accurate model representation. The methodologies used in

physical modeling are also discussed.

Fluid flows and related phenomenon can be described by a set of highly non-linear

partial differential (or integro-differential) equations (PDE), which cannot be solved

analytically except in a few special cases. To obtain an approximate solution, a

discretization method is used, which approximates the differential equations by a system

of algebraic equations. These equations can then be solved on a computer, providing a

description of the flow field at discrete locations in space and time.

During the past few decades there has been a consistent increase in computing power,

which has allowed problems of ever greater size and complexity to be handled. This

growth rate in computing power appears to be sustainable over the foreseeable future.

This has proved to be a boon to the CFD modeler as it provides a means of solving ever

more realistic and significant problems. For example, it has been possible to extend

numerical models from simple one-dimensional analyses to two and three-dimensional

comprehensive computational solutions. The extension to 3D is an essential step when

addressing and solving real world problems. CFD is finding its way into process,

chemical, civil, mechanical and environmental engineering. Optimization in these areas

can produce large savings in equipment and energy costs and in reduction of

environment pollution.

36


3.1. Mathematical model

The basis of any numerical method is the mathematical model; in this case the set of

PDE’s in conjunction with appropriate boundary and initial conditions. A general form

of the transport equation used in CFD is shown in 3.1.

( ) Φ+⎟⎟⎠

⎞⎜⎜⎝

⎛∂Φ∂

Γ∂∂

=Φ∂∂

+Φ∂∂ S

xxU

xt iii

i

ρρ )( (3.1)

where ρ is the fluid density

xi is the distance in the ith direction

Ui is the velocity in the ith direction

Γ is diffusion coefficient of the variable Φ

SΦ is a source or sink term for the variable Φ

On the left hand side, the first term and the second term are the unsteady term and

convection term respectively. The first term on the right hand side is the diffusion term.

Depending on Φ, the above equation represents mass, momentum, species or energy

conservation. For example, if Φ=1 then the equation 3.1 becomes conservation of mass;

when Φ=Ui, it becomes momentum equation and so on.

3.2. Discretization method

After selecting the mathematical model, one has to choose a suitable discretization

method, a method of approximating the differential equations by a system of algebraic

equations for the variables at some set of discrete locations in space and time. There are

many approaches, but the most commonly used are: finite difference (FD), finite

volume (FV) and finite element (FE) methods.

Finite difference method

This is the oldest method for numerical solution of PDE’s, believed to have been

introduced by Euler in 18th century. It is also the easiest method to use for simple

geometries. The starting point is the conservation equation in differential form. The

solution domain is covered by a grid. At each grid point, the differential equation is

approximated by replacing the partial derivatives by approximations in terms of the

nodal values of the functions. The result is one algebraic equation per grid node, in

which the variable value at that and a certain number of neighbor nodes appear as

unknowns. In principle, the FD method can be applied to any grid type but is most

37


suitable for structured grids. The grid lines serve as local coordinate lines. Taylor series

expansion or polynomial fitting is used to obtain approximations to the first and second

derivatives of the variables with respect to the coordinates. When necessary, these

methods are also used to obtain variable values at locations other than grid nodes

(interpolation).

Finite Volume method

The FV method uses the integral form of the conservation equations as its starting point.

The solution domain is subdivided into a finite number of contiguous control volumes

(CVs), and the conservation equations are applied to each CV. At the centroid of each

CV lies a computational node at which the variable values are to be calculated.

Interpolation is used to express variable values at the CV surface in terms of the nodal

(CV-center) values. Surface and volume integrals are approximated using suitable

quadrature formulae. As a result, one obtains an algebraic equation for each CV, in

which number of neighbor nodal values appear. The FV method can accommodate any

type of grid, so it is suitable for complex geometries. The grid defines only the control

volume boundaries and need not be related to a coordinate system. The method is

conservative by construction, so long as surface integrals (which represent convective

and diffusive fluxes) are the same for the CVs sharing the boundary. The FV approach

is perhaps the simplest to understand and to program. All terms that need to be

approximated have physical meaning, which is why it is popular with engineers. In this

thesis FV method was used with structured grids due to simple shape of the geometry.

3.3. Discretization scheme

The algebraic equations involving the unknown values of Φ at chosen grid points are

derived from the differential equation governing Φ. Some assumption must have to be

employed about how Φ varies between the grid points. It is often more practical to use

piecewise profiles such that a given segment describes the variation of Φ over only a

small region in terms of the Φ values at the grid points within and around that region.

Thus, it is common to subdivide the calculation domain into a number of sub domains

or elements such that a separate profile assumption can be associated with each sub

domain and different profiles may even be used for different terms in the differential

equation (Patankar, 2001). The profile assumed between grid points is called the

38


discretization scheme. For a given differential equation, the possible discretization

equations are by no means unique. The different types arise from the differences in the

profile assumptions and in the methods of derivation. Most terms in the discretized

equations can be approximated using central differencing, which is simple linear

interpolation between grid points; however, the advection term, the second term on the

left hand side of equation 3.1, may become unstable when central differencing is

applied, and requires the use of other schemes. These vary from the simple upwind

scheme, where the value of Φ at the interface is taken as the value of Φ at the grid point

in the upwind side of that face, to more complex schemes such as QUICK (Ferziger and

Peric, 1997) which is a quadratic upwind scheme where a quadratic profile is fitted to

the two upwind and one downwind nodes.

3.4. Solution method

Discretization yields a large system of non-linear equations. The method of solution

depends on the problem. For unsteady flows, methods based on those used for initial

value problems for ordinary differential equations (marching in time) are used. At each

time step an elliptic problem has to be solved. Steady flow problems are usually solved

by pseudo-time marching or an equivalent iteration scheme. Since the equations are

non-linear, an iteration scheme is used to solve them. The Semi-Implicit Method for

Pressure-Linked equations (SIMPLE) is one basis for the iterative solution procedure

used in calculating a solution to the Navier-Stokes equations (Patankar, 1983). In

introducing the SIMPLE algorithm, Patankar pointed out the problem of generating

unrealistic solutions by determining velocity and pressure at the same mesh point.

Patanker suggested the use of a staggered grid (Harlow and Welch, 1965) through

which the velocity components are determined at control volume faces rather than at

their centres. The computational cost and difficulty in implementing such a grid for

complex shapes restricted much of the earlier numerical modeling to recti-linear grids.

In this context an important contribution was made by Rhie (1981) and Rhie and Chow

(1983) who judiciously included the pressure difference between cell centres in

determining the velocity values at the cell face and thus making the velocity values

determined at cell centres sensitive to variations in pressure. This interpolation scheme

obviated the necessity of staggered grids and thus reduced the computational burden of

modeling a complex geometry, using curvilinear coordinates.

39


Recent development of the commercial software (CFX-5) enables to save the computing

power by using coupled solver to solve the system of non-linear equations. In this

procedure u, v and w momentum of the Navier-Stokes equation are solved

simultaneously and thus the number of iterations required to obtain a converged solution

are much less than the SIMPLE algorithm.

3.5. Difficulties in numerical simulation

Numerical Diffusion

The discretised formulation of convection and diffusion terms can significantly

influence the accuracy and stability of the numerical solution developed. The central

differencing scheme, a simple scheme derived from the Taylor-series formulation, is

proved to produce unrealistic solutions when the strength of the convection exceeds that

of the diffusion by a factor of more than two. Hence the central differencing scheme has

been used in combination with the upwind scheme (Courant et al., 1952, and Runchal

and Wolfshtein, 1969), which assumes the value of a convected variable at an interface

to equal to the value at the grid point on the upwind side of the face. Such a combined

scheme, known as the hybrid scheme (Spalding, 1972), is highly dissipative in non-

orthogonal grids and predicts the decay rate of a jet developing at an angle to the

computational grid to be much greater than that which would occur physically (Perry

and Yan, 1992). This is commonly known as numerical or false diffusion and can be

estimated using the expressions developed by de Vahl Davis and Mallinson (1972) and

Leschziner (1980). These expressions indicate that the numerical diffusion attains a

maximum value for oblique flows(θ =45º) and disappear all together when θ =0º or 90º

for flows parallel to either of the grid axes. Orienting the grid lines with the flow

direction can reduce numerical diffusion. The other option is to make the mesh very

fine, which will result in unreasonable solution times or computer memory limitations,

and to use higher order schemes. In this research program numerical diffusion was

reduced by orienting the grid lines with the flow direction.

40


3.6. Turbulence modeling

The Reynolds average Navier-Stokes (RANS) equation for turbulent flow can be

written in the form

)''2()( jiijii

jij

i uusxx

pUUxt

Uρµρρ −

∂∂

+∂∂

−=∂∂

+∂

∂ (3.2)

which is almost the same as the equation for the laminar flow except the term ji uu ''ρ .

The quantity ji uu ''ρ is known as the Reynolds stress tensor and has six independent

components. Unfortunately there are only four equations (three momentum and one

continuity) for the ten unknown variables. Thus a method is required to close this

system of equations.

The turbulence model as defined by Rodi (1984) is a set of equations (algebraic or

differential), which determine the Reynolds stresses in the Navier-Stokes mean flow

equations and thus close the system of equations. It is based on a hypothesis about

turbulent processes and requires empirical input in the form of constants and functions.

It does not simulate the details of the turbulent motion but only the effects of turbulence

on the mean flow behaviour.

The first approach originates from the Boussinesq assumption that turbulent stresses can

be linked to the mean velocity gradients through a term called eddy or turbulent

viscosity. Based on the models approach to the definition/distribution of eddy viscosity

the models are classified as Zero, One and Two-equation models depending upon the

number of transport equations used.

Zero-equation models are based on the mixing length concept introduced by Prandtl

(1945) to represent a characteristic length of motion of eddies, by analogy to the mean

free path of the molecules in the kinetic theory of gases. These models define the eddy

viscosity based on mixing length and mean velocity gradients. The mixing length itself

is defined using empirical relations observed experimentally and thus limits the

application of these models.

In One-equation models, the eddy viscosity is defined using the more meaningful

turbulent kinetic energy, which represents the intensity of turbulence, in the place of

41


mean velocity gradients (Kolmogorov, 1942 and Prandtl, 1945). This, however,

necessitates the knowledge of the distribution of turbulent kinetic energy, which is

determined by solving its transport equation. One-equation models still need the

definition of the length scales, which in most models are defined through empirical

relations. Introduction of a transport equation for k has improved the accuracy of

predictions of the One-equation model compared to the Zero- equation model.

In Two equation models, the dependency on empirical relations is further reduced by

developing a transport equation for a length scale related variable such as dissipation

rate ε or turbulence frequency,σ. The standard k-ε two-equation turbulence model,

found in almost all CFD codes, is the basis for many turbulent flow calculations. This

model solves transport equations for k, the turbulent kinetic energy, and ε, the rate of

dissipation of turbulent energy to close the Reynolds-averaged Navier-Stokes equations.

The two transport equations are:

( ) ρεσµµρ −+=∇⎟⎟

⎠

⎞⎜⎜⎝

⎛+∇−∇ GPkUk

k

T.. (3.3)

( )( )k

CGCPk

CU T2

231 0,max.).( ερεεσµµερ

ε−+=⎟

⎟⎠

⎞⎜⎜⎝

⎛∇⎟⎟

⎠

⎞⎜⎜⎝

⎛+∇−∇ (3.4)

where P is shear production. The values of C1, C2 and C3 are 1.44, 1.92 and 0.0

respectively (CFX User Guide, 1997). It is the most robust of the models available in

commercial codes, in that it will generate a converged solution in almost all cases. The

drawback of this model is its overly diffusive characteristics and tendency to under-

predict velocity gradients. The RNG k-ε model, another two-equation model, uses

renormalization group theory to generate a different set of closure constants for the k

and ε transport equations, giving the C2 constant a dependence on the shear part of

turbulent kinetic energy production. Another Two-equation model is k-ω model where

an equation is still solved for k, but the equation for ε is replaced by an equation for the

turbulence frequency ω defined by ω=ε/k. Two layer models are also popular, which

use separate models near the wall and in the free stream. One such model is the Shear

Stress Transport (SST) turbulence model. This model is a combination of k-ε and k-ω

model. Very near to the wall it solves the k-ω model and as the solution moves towards

the free stream k-ω model switches to k-ε model.

42


The Reynolds stress model is the most complex of the RANS turbulence models, in

which transport equations are solved for each of the six Reynolds stresses and one for ε.

This model is more appropriate for modeling anisotropic turbulence typical of three-

dimensional flows, where the turbulent fluctuations are not equal in all three directions.

However solving six transport equations for the Reynolds stresses and one equation for

ε increase the computational cost and thus limit the application of the Reynolds stress

turbulence model in the industrial problems. In this thesis k-ε and SST turbulence model

were used for numerical simulation.

3.7. Dimensions of the burners and furnace

The dimensions of the burners and the furnace investigated in this thesis are described

here. Three types of burners were used throughout the whole study namely geometry B,

C, and D and the furnace investigated was the Yallourn stage-2 type, at Morwell,

Victoria, Australia. All the geometries (B, C, and D) were based on the physical models

of Perry and Hausler (1984) and were described briefly in chapter 2 but are repeated

here again for convenience. Geometry A was not used for investigation but was the

basis for all other geometries and so it is also included here.

Geometry A

A nearly square primary jet flanked over and below by rectangular secondary jets,

discharging orthogonally from a wall.

Geometry B

Same as geometry A but the jets making an angle of 60º to the wall. Geometrically

similar to Yallourn stage-1 furnaces

Geometry C

Same as geometry B but the jets divider terminating a short distance upstream of the

furnace wall, thus producing a straight walled recess.

Geometry D

Same as geometry C but with a diverging recess. This was geometrically similar to the

recessed burner used in Yallourn stage 2 furnaces.

43


Only the recessed burners (geometry C and D) are modeled in chapter 4 to investigate

the effect of jet velocity ratio without cross-flow. The dimensions of the burners are

shown in figure 3.1 and 3.2.

y

x

z

x

z

y Figure 3.1: Geometry C-60ο to the furnace wall where the cavity has parallel sidewalls.

44


x

y

x

z

z

y Figure 3.2: Geometry D- Diverging inserts between furnace mouth and the furnace wall.

The dimensions of the burners were one twenty-seventh (1:27) geometrically scale of

the Yallourn stage-2 furnace. Geometries B and D are used in chapters 5, 6 and 7 for

physical and numerical modeling of single-phase and two-phase flow development in

the presence of cross-flow. Dimensions of burner geometries B and D and the cross-

flow duct are shown in figure 3.3(a-c) and the photograph of the burners in figure 3.4. A

45


one-fifteenth (1:15) scale was chosen the same as Yan and Perry (1994), in order to get

detailed information on both mean velocity and turbulent fluctuating components in

near and far region of jets. Upstream duct lengths for the burner and the cross-flow jets

were greater than the 20 times the equivalent hydraulic diameter of the duct, to ensure

that the flow profile had become developed at the nozzle and to enable the particles to

accelerate to the flow speed by the nozzle.

Figure 3.3(a): Burner B (Dimensions are in mm). y z

x y

Figure 3.3(b): Burner D (Dimensions are in mm).

z y

x y

46


z

x

y

y Figure 3.3(c): Cross-flow nozzle (Dimensions are in mm).

Figure 3.4: Photographs showing burner B, burner D and cross-flow duct.

The numerical modeling of aerodynamics and combustion of conventional lignite and

MTE lignite in a full-scale industrial furnace is described in chapter 8. The dimensions

of the overall furnace are shown in figure 3.5. Some modification was made at the top

of the furnace for numerical stability. The top of the furnace was kept partially open for

moving the flue gas to the convection section. It was assumed that 42% of the total area

was open at the top and the rest was blocked with the boiler and super heater pipes. The

open area was represented by the reduced rectangular area as shown in figure 3.5. Each

burner was attached to the furnace wall at an angle of either 60ο or 66ο as shown in the

figure 3.6.

47


Figure3.5: Schematic diagram of the furnace showing all the dimensions in meter (m).

48


Figure 3.6: Orientation of the burners in tangentially-fired furnace

660 600

3.8. Model description, grid and boundary conditions

3.8.1. Investigation without cross-flow

Model description

The burner model and solution domain used in chapter 4 are shown in figure 3.7(a-b).

The dimensions of the cross section of the primary duct and the secondary duct in the

cold flow model were (37.5mm x 29mm) and (37.5mmx17mm) respectively. The

hydraulic diameter (De), which is the diameter of a round nozzle with the equivalent

cross-sectional area to the primary nozzle, was 0.0372m. Upstream of the nozzle, the

duct length was 1.95m, which was equivalent to 52De. In the physical model of Perry

and Hausler (1984) the upstream end of the duct was connected to a plenum chamber,

giving an almost uniform velocity profile at the duct entrance. This was repeated in the

numerical models, where a uniform velocity profile was set at the upstream end of the

duct, and the flow was allowed to develop along the duct before reaching the nozzle. In

the physical model the three jets discharged into a large room. In order to get the same

effect in the numerical modeling, the domain was made large enough to ensure that the

steep gradients at the boundary of the jets were contained within the model domain, and

appropriate pressure boundary conditions were applied to simulate an open atmosphere.

Due to geometric symmetry only half of primary burner and one full secondary burner

were modeled which halved the number of grids required, reducing the computational

cost.

49


Figure 3.7(a): Burner model showing geometry C, D and upstream ducts

Figure 3.7(b): Solution domain showing the body fitted mesh.

Downstream of the nozzle, the computational domain extended approximately 1.2m in

the stream wise direction, 0.4m from the nozzle in the cross-stream direction on the

short side and 0.8m on the long side. The secondary jet velocity was kept constant and

the primary jet velocity was varied to achieve the required jet velocity ratio. The total

number of cells used was approximately 1,000,000.

50


Boundary Conditions

Inlet

Dirichlet boundary conditions were set by specifying a flat velocity profile at the inlets.

The value of the primary jet velocity was 60 m/s, 42.86 m/s and 20 m/s for jet velocity

ratios of 1.0, 1.4 and 3.0 respectively. The secondary jet velocity was kept constant at

60m/s. Typical turbulence quantities at the inlet were calculated from inlet velocities by

considering the turbulence intensity as 0.05.

Opening

Modeling the open boundary conditions of the experimental rig required that a suitable

boundary condition be used to take into account the essentially infinite space into which

the jets discharged. The most suitable were Dirichlet pressure boundary conditions, for

which the pressure was assumed to be constant along the entire boundary. The constant

pressure approximation allows both inflow and out flow of the same boundary. In this

investigation the relative pressure was set to 0.0 Pa to all open boundaries. These

boundaries were placed far enough from the high velocity and pressure gradients near

the jet that the zero gradient assumption was reasonably valid.

Symmetry

A symmetry plane was included through the centre axis of the primary jet to reduce the

number of cells required to model the burner. Implementation of symmetry boundary

condition is such that no transport of any variable is allowed across the plane of

symmetry and gradient of all the variables is assumed to be zero across the symmetry

boundary.

3.8.2. Investigation with cross-flow

Model description

The burner model used in physical modeling (chapter 5 and 7) consisted of a large box

(1.85m x 1.5m x 1.6m) made from a frame of aluminium with perspex walls. The

dimensions of the cross section of the primary and the secondary ducts were 75mm x

58mm and 75mm x 34mm respectively. Dimensions of burner B, burner D and the

cross-flow nozzle are given in section 3.7. The hydraulic diameter (De) was 64 mm and

the spacing between the primary and secondary duct was 27mm. The duct length was

1.2m, to give a more developed velocity profile at the exit of the burner. The dimension

51


of the cross section of the cross-flow duct was 75mm wide x 252mm and the duct length

was 1.8m. Figure 3.8(a) shows the dimensions of the flow containment box and figure

3.8(b) shows the dimensions of the cross section of primary and secondary ducts.

60o

1.85

0.3

1.6

1.0

1.5

0.5

Burner inlet

Exitto fan

Cross-flow inlet0.075 x 0.252

(a)

(b)

Primaryduct

Lower secondary duct

Upper secondary duct

Figure 3.8(a-b): Dimensioned view of flow containment box (a) and burner inlet detail

(b) (all units in m)

The CFD model (chapter 6) was exactly the same as the physical model (figure 3.8(a)).

Figure 3.9(a) shows the mesh on geometry B and cross-flow illustrating how the grid

was refined in these regions. A uniform, high-density mesh was selected across the

nozzles at the wall. The same resolution was applied to the base region between the

nozzles. A geometric expansion factor was applied upstream of the nozzles to model the

ducts and along the wall away from the nozzles. The total number of cells used was

around 600,000.

52


Figure 3.9(a): Mesh on geometry B and cross-flow duct.

Figure 3.9(b): Mesh on geometry D and cross-flow duct.

Figure 3.9(b) shows the mesh on geometry D and cross-flow duct. Special care was

taken in modeling geometry D because of the diverging recess walls. One of the

requirements in creating the grid of geometry D was adequate resolution inside the

recess. A denser grid was selected to capture the small recirculations due to small step

change from the duct to the recess. Thus the number of cells used to model geometry D

was larger than that used for geometry B.

53


Boundary Conditions

Inlet

Dirichlet boundary conditions were set on the inlet for all jets by specifying a flat

velocity profile at the inlets. The velocity in the primary jet was 8m/s. The value of the

secondary jet velocity was 8m/s for φ=1.0 and 24m/s for φ=3.0 and the cross-flow jet

velocity was 8m/s. Typical turbulence quantities at the inlet were calculated from inlet

velocities by considering the turbulence intensity as 0.05.

Exit

A Dirichlet pressure boundary condition was selected at the exit on which the pressure

was assumed to be constant along the entire boundary. The constant pressure

approximation allows both inflow and out flow of the same boundary. In the present

simulation the relative pressure was set to 0.0 Pa at the exit.

Wall

The jet discharged into a containment box made of wall. Turbulent flow near a solid

boundary behaves differently to that in the free stream and for this reason it is important

to correctly account for the wall’s effect on the fluid motion. In the present simulation a

log layer profile was assumed at the wall.

3.8.3. Investigation of full-scale tangentially-fired furnace

Model description

The solution domain with burner inlets, flue gas offtakes and the top outlet are shown in

figure 3.10(a) together with the primary and secondary recessed nozzles in the figure

3.10(b). The furnace was tangentially-fired with eight groups of vertically arranged

burners (figure 3.10(a)). The overall dimension of the furnace was 15.9x15.6x50m. Six

of the eight groups of burners were in operation, which is the usual practice in the

power plant to keep two mills as standby or on maintenance. Each group of recessed

burners was divided into three parts. The lower main burner, upper main burner and

vapor burner (figure 3.10(a)). The lower and upper main burners were oriented parallel

to the horizontal plane. The vapor burners located on the top of the upper main burners

were declined at an angle of 15ο to the horizontal plane. Each vapor burner consisted of

four primary nozzles and three secondary nozzles stacked on top of one another. The

54


lower and upper main burners were again subdivided into primary and secondary

nozzles (figure 3.10(b)).

To convection section

Flue gas offtakes

Vapor burner

Lower m bur

ain ner

Upper main burner

Upper top secondary nozzle

Lower primary nozzle

Lower bottom secondary nozzle

(a) (b)

Figure 3.10(a-b): Inlet and outlet patches of Yallourn stage-2 furnace (a) and primary

and Secondary recessed nozzles (b).

Coal and flue gas was introduced into the furnace from the primary nozzles of the lower

main burners, upper main burners and vapor burners. Air at 320°C, was introduced

through the secondary nozzles. There were eight flue gas offtakes for the recycling of

the flue gas for coal drying. The flue gas was ultimately pumped into the furnace

through the primary nozzles along with the water vapor and the partially dry pulverized

coal at 140°C.

The computational grid used in this study consisted of 48 cells in the X, 47 in the Y and

153 in the Z direction (figure 3.11(a)). Increased grid resolution used in the burner

region enabled each primary and secondary nozzle to be resolved adequately. 24 cells in

the primary and 16 in the secondary were used (figure 3.11(b). Thus the total number of

cells used was 367632.

55


(a) (b)

Figure 3.11(a-b): Grid layout of the full furnace (a), primary and secondary nozzles (b).

Boundary Conditions

Inlets

Dirichlet boundary conditions were set by specifying a flat velocity profile at the

primary and secondary nozzle inlets. The velocity of the primary jet was set to 20m/s.

The value of the secondary jet velocity was 34m/s. Typical turbulence quantities at the

inlet were calculated from inlet velocities. Temperature at the primary and secondary

nozzles was set to 140οC and 320οC respectively.

Flue gas offtakes

Mass flow boundaries were selected at the flue gas offtakes. Implementation of mass

flow boundary condition is such that a fix amount of hot flue gas can be extracted from

the furnace.

56


Top outlet

Dirichlet pressure boundary condition was selected at the top outlet on which the

pressure was assumed to be constant along the entire boundary. In the present

simulation the relative pressure was set to 0.0 Pa at the top outlet.

Wall

No slip boundary condition was selected at the wall. The temperature of the wall was set

to 700οC.

Particle Tracking

Particles were tracked using Lagrangian equation of motion from the inlet ports until

particles burn out or leave the furnace.

3.9. Experimental set-up and methodology used in physical modeling

3.9.1. Experimental set-up

A schematic diagram of the experimental set-up used in the physical modeling (chapter

5 and 7) is shown in figure 3.12. Air was passed into the burner model via duct A

(Burner) and duct B (cross-flow), and exited through duct C. The air was driven into the

rig from a blower and extracted through the bag house using an exhaust fan. The flow

control baffles were adjusted to achieve the desired flow rate.

From blower To bag houseand fan

Flow controlbaffles

Flowseeding

ExitCross-flow

Burnerjet

Accessdoor

Ax

y

BC

Figure 3.12: Schematic diagram of burner model and associated ducting.

57


3.9.2. Methodology

Basic Principle of Laser Doppler Anemometry

Laser Doppler Anemometry (LDA) was used to measure the mean and fluctuating

component of velocity. Yeh and Cummings (1964) first introduced the concept of LDA

to measure the velocity of moving particles carried in a water flow. The principle of the

LDA technique is based on the fact that the light scattered by the particles in the flow is

Doppler shifted. The frequency of this Doppler shift is directly proportional to the

velocity of the particles. Combinations of optical set-ups are used to measure the

Doppler shift frequencies, and the most common optical system is the dual beam or

fringe mode system. In a dual beam arrangement, two Gaussian beams of equal

intensity cross each other in the flow field using a focusing lens to produce an ellipsoid

shaped measurement volume. A fringe pattern is generated at this point in the same

plane as the beams. Figure 3.13 shows the schematic diagram of the dual beam optical

system and the fringe description.

LASER

Beamsplitter Focusing Lens

U

θ/2

δf

Figure 3.13: Dual beam optical system and fringe pattern.

The fringe separation, δf, is defined as

2sin2 θλδ =f (3.5)

where λ is the laser wavelength and θ is the laser beam intersection angle. A particle

passing through the dark and light fringe patterns will scatter light the intensity of which

58


will vary. Assuming that the Doppler shift frequency is fD, then the particle velocity

component, U, normal to the fringe in the same plane, is given by

2sin 2

θ

λδ D

DfffU == (3.6)

This model provides a correct expression for the velocity of particles in the flow field,

and requires no calibration since the wavelength and intersection angle are constant

once the color of the laser beam is chosen and the optical arrangement is fixed. The

component of velocity must always be measured normal to the fringes, regardless of the

direction of the flow.

Description of LDA Facility Used in Current Study

A TSI-Aerometrics 2D LDA was used in this experiment. This system is capable of

simultaneous measurements of the mean and fluctuating component of velocity at any

point within the burner model. Figure 3.14 shows a schematic diagram of the LDA

apparatus used in the current study. The system consisted of a laser source, a transmitter

based optical arrangement system connected with a fiber optic probe and a signal

processor together with a data acquisition system to collect the measurements.

Signal Processor Computer Fiber Optic Probe

Photo Multiplier Burner Model

Transmitter

Ar-Ion Laser Manipulator

Figure 3.14: Schematic diagram of LDA apparatus

The laser source was a Argon-Ion laser with overall output power of 5W for all

continuous lines of wavelengths from 351.1 nm to 528.7 nm, with approximately 1.5W

for the blue beam at 488 nm wavelength and the green beam at 514.5 nm wavelength.

59


These blue and green beams were the two color beams chosen for the optical

arrangement of the two-dimensional LDA system.

The transmitter unit consisted of an integrated color separator, two frequency shifters

and two beam splitters. It function was to separate the incoming beam from the Argon-

Ion laser into the two individual color components and then divide each colored beam

into two beams with a 40 MHz frequency difference. The transmitter was aligned next

to the laser source to ensure the specified beam path from the input aperture to the

manipulator. There were four manipulators mounted on the transmitter for four output

beams. With adjustments on the manipulator, the output laser beam could be positioned

on the exact centre of the optic fiber core. The fiber distribution unit mounted at the end

of the transmitter was used to connect between the fiber-optic probe and manipulators.

The fibre optic probe had a lens of a 250 mm focal length and a 40 mm beam separation

which produced an ellipsoid shaped measuring volume with dimensions of 0.11 mm ×

0.11 mm × 1.5 mm. The photo multiplier was a type of photo detector that converted

changes in scattered light intensity into electrical signals, which were then analyzed in

the signal processor. Table 3.1 lists the general technical data of the LDA fiber optic

probe system used for the current study.

Table 3.1 Technical data of the LDA System

Blue beam Green beam Laser wavelength, λ, (nm) 488 514.5 Focal length (mm) 250 250 Beam separation (mm) 40 40 Diameter of Gaussian beam (mm) 1.4 1.4 Fringe separation, δf, (µm) 3.05974 3.2259 Number of fringes 36 36 Bragg cell frequency (MHz) 40 40

Particle Seeding and Signal Processing

The source of signals of LDA is the scattering light from the particles in the fluid flow.

Geometric and physical parameters of the particles influence the quality of signals

obtained from the photo detector of the LDA system. If particles are small enough to be

in dynamic equilibrium with the fluid flow then the particle velocity can be safely

assumed to be equal to the fluid velocity. Therefore, in choosing the correct seeding for

60


investigation, seed particles must be small enough to faithfully represent the fluid flow,

but at the same time be large enough to generate sufficient scattering of light for the

operation of the photo detector and the signal processor. Another important physical

parameter is the relative refractive index of the particle (the ratio of the refractive

indices of the particle and the medium), a measure of the ability of the particles to

scatter light. In these experiments, for single-phase flow, the airflow was seeded with a

fine mist of sugar particles introduced into the primary, secondary and cross-flow ducts.

The partially dried sugar particles with a mean diameter about 1 µm were generated by

a TSI six-jet atomizer from a 5% sugar solution. For two-phase flow, solid glass spheres

were used as the representative of the coal particles. The mean diameter of the solid

glass sphere was 66 µm and the density was 2450 kg/m3. The density of the coal in the

actual furnace is 1150 kg/m3, but for the density of the particles considered in the

model, the mean free path between particles had to be substantially reduced in order to

achieve the correct mass loading and this contributed to difficulties in feeding the

particles into the primary flow. To stabilize the particle feed and to make the mean free

path more representative, the mass loading was reduced from typical furnace levels of

order 0.2 to a model value of 0.01.

The particles were introduced only at the centre of the primary nozzle from a bubbling

fluidized bed at a rate of 20 gm/min. The location of the hole through which the

particles were seeded was near the beginning of the upstream duct to enable the

particles to accelerate to the flow speed by the time they reached the nozzle. The LDA

system used in this investigation was not able to simultaneously measure the velocity of

the relatively large second phase particles and very small partially dried sugar particles

used to mimic the gas velocity. As a result the gas and second phase particles were

measured separately by seeding the flow with each size of particles in separate test run.

The momentum transfer between phases was assumed to be negligible because of the

low mass loading of the particles and therefore the gas-phase velocity should not have

been influenced by the presence of the second phase.

Data Processing

The commercial software package Dataview was used for data acquisition. This

software could be configured with one two or three-dimensional LDA systems.

Furthermore, the software simplified the system setup by including hardware

61


diagnostics, processor setup, optics configuration, output documentation and flowfield

mapping in the same package. Figure 3.15 shows typical screen shot of the

simultaneous real-time histogram display of velocity, which helped to optimize the

system performance, and gave a quick display of statistics to ensure that valid data was

being collected. Dataview also output all measurement results in easily accessible

ASCII format.

Figure 3.15: Histogram display showing U and V component of velocity

Data was taken for 60 second at each position inside the burner model. The average data

rate was 400 Hz giving a total of around 24,000 particles counted at each position. As

discussed by Buchhave et al. (1979), George (1988) and Hussein et al. (1994), the

datum were processed using the residence time weighting technique in which all

measured values were weighted by the measured residence times of the individual

scattering particles. The residence time weighting technique is known to avoid most of

the sources of bias present in alternative processing techniques, so that additional errors

were not introduced by hardware deficiencies.

In the current study, mean component of velocities, U and V, RMS values of the

fluctuating components, 22 , vu and the cross moment, uv , which is the Reynolds

shear stress divided by the fluid density were calculated using following equations:

62


∑

∑

=

== N

ii

N

iii

t

tUU

1

1 ∆

∆ (3.7)

∑

∑

=

== N

ii

N

iii

t

tV

1

1V ∆

∆ (3.8)

2/1

1

1

2

2)(

⎥⎥⎥⎥

⎦

⎤

⎢⎢⎢⎢

⎣

⎡−

=∑

∑

=

=N

ii

N

iii

t

tUUu

∆

∆ (3.9)

2/1

1

1

2

2)(

⎥⎥⎥⎥

⎦

⎤

⎢⎢⎢⎢

⎣

⎡−

=∑

∑

=

=N

ii

N

iii

t

tVVv

∆

∆ (3.10)

∑

∑

=

=−−

= N

ii

N

iiii

t

tVVUUuv

1

1))((

∆

∆ (3.11)

where the subscript i denotes the ith particle, N is the number of particles (sample size)

and ∆ti is the transit time of ith particle for residence time weighting.

3.10. Error analysis

Although the LDA technique has many advantages compared to methods such as hot-

film probe and electroresistivity probe, there are some limitations and sources of errors

encountered with the technique. These are listed below.

• The LDA system can only accept the Doppler single of a "detectable" particle

passing through the measuring volume. So its measurement is discontinuous, unlike

the method of hot-film anemometry. This discontinued measurement could

introduce errors, especially in turbulent fluctuating velocity components.

• Due to the frequency and internal timer limitations of the signal processor, there

were errors of 1 % and 3.6 % for the mean velocity and turbulent fluctuating

63


component respectively. The accuracy of the beam spacing and the focal length on

the fibre optic probe can also affect the calculation of velocity using equation 3.6,

where the half intersection angle is calculated based on the ratio of the half beam

spacing and the focal length. In the existing LDA system, assuming ±0.2 mm

misalignment on the beam spacing and focal length, an error of 1 % may introduced

in the calculation of velocities.

3.11. Summary

The commercial CFD code, CFX-4 and CFX-5 were used in this thesis to solve the

Navier-Stokes equations in modeling tangentially-fired furnace and various multi-jet

burner geometries with and without cross-flow. The geometries were based on the

experimental modeling of Perry and Hausler (1984). SST and k-ε turbulence models

were used for turbulence closure. Experiments were conducted for geometries B and D

in the presence of cross-flow by using LDA technique.

64

NOTE


Chapter-4 Numerical investigation of rectangular

slot-burners without cross-flow

Chapter 4- Numerical investigation of rectangular slot- burners without cross-flow

4. Numerical investigation of rectangular slot-burners without cross-flow

This chapter presents the results of CFD simulations of rectangular slot-burners without

the presence of cross-flow. Burner geometries C and D have been considered for

investigation. Both burners consisted of three rectangular vertically aligned slots with

the centre known as primary nozzle and the top and bottom ones known as upper and

lower secondary nozzles. In both cases, the jets discharged into an open atmosphere at

an angle of 60ο to the furnace wall. Full descriptions of geometry C and D are given in

chapter 3. The velocity ratio, φ, was defined as the ratio of secondary to primary jet

velocity and the investigation was carried out for φ=1.0, 1.4 and 3.0. Hart (2001)

performed the simulation of geometry D for φ=1.0. This simulation is repeated here to

compare the effect of jet velocity ratio and to make the data set more complete. CFX-5

has been used in this study. CFX-5 uses a coupled solver for solution of the mass and

momentum equations. The time averaged transport equations for mass and momentum

were solved, without any heat transfer. k-ε model was used for turbulence closure. The

validation of the numerical results was carried out by comparison with the experimental

data of Perry and Hausler (1984).

4.1. Grid independence test

The purpose of grid independence test is to determine minimum grid resolution required

to generate a solution that is independent of the grid used. Starting with a coarse grid the

number of cells was increased in the region of interest until the solution from each grid

was unchanged for successive grid refinements. The grid independence test was

performed on geometry D and for jet velocity ratio of 1.0. Numerical diffusion was

reduced by aligning the grids parallel to the geometric axis of the burner as shown in

figure 4.1(a). The grid resolution was based on the number of cells used to model the

ducts. Three grid refinements were performed, the first setting 18x8 cells in the primary

duct cross section, 18x4 cells in the secondary duct cross section and 18x4 cells

covering the base between the jets. Due to the symmetry of the jet, the actual number of

grids used in modelling the ducts was 18x 4 cells in the primary jet, 18x4 cells in the

secondary jet and 18x4 cells covering the base region i.e. only half of the primary duct

was modelled. A geometric progression was used to expand the grid away from the jet

66


nozzle to reduce the grid resolution in regions away from the region of interest where

the velocity gradients were small.

Figure 4.1(a): Grid distribution in the xy plane

Successive grid refinements involved increasing the number of cells in the duct cross-

sections to 24x12 in the primary and 24x6 in the secondary and finally 36x18 and 36x9.

The same principles were applied to the grid expansion for each grid. The results of the

grid independence test are presented in figure 4.1(b) comparing the transverse velocity

profiles in the xy plane at x/De=1.0 downstream of the nozzle. This plane was through

the centre of the primary jet, and velocities were normalised to the centreline exit

velocity of the primary jet.

Velocity Profile (φ=1,x/De=1) for Geometry D

00.10.20.30.40.50.60.70.80.9

1

-0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8

y/De

U/U

ce

18x8 cells

24x12 cells

36x18 cells

Expt

Figure 4.1(b): Velocity profile in xy plane for grid independence test

67


Some change in the predicted profiles was found with each successive grid refinement.

The peak normalized velocity for the 18x8 grid was 0.90, compared with 0.93 for the

24x12 grid and 0.95 for the 36x18. For y/De>0.5, successive grid refinements also

produced a less diffused jet, the most diffuse being the 18x8 grid, followed by the

24x12 and 36x18. Both 24x12 and 36x18 grids gave a very close prediction compared

with the experiments on the short side. On the long side the prediction for 36x18 grid

was better than the 24x12 grid. The difference between the 24x12 and 36x18 predictions

was small enough to suggest that any further grid refinement would not yield a

substantially different profile in this plane. As the prediction with 36x18 grid was better

on the long side, this grid resolution was used for the simulation.

4.2. Results and discussion

4.2.1. Comparison with experiment

The following section compares the results of the simulation with the experimental data

from Perry and Hausler (1984). All velocities were normalized to the centreline velocity

at the exit of nozzle of geometry A (Hart, 2001) to compare the numerical results with

experimental values. Figure 4.2(a) shows the velocity decay along the geometric

centreline of the primary jet (PJ) and secondary jet (SJ) in geometry C for φ=1.0. The

numerical results under predicted the decay rate for both primary and secondary jets.

From the exit of the nozzle the decay rate for both the primary and secondary jets was

slow up to x/De =3.0. Beyond that the predicted secondary jet decayed faster than the

primary jet.

Centreline decay of peak axial velocity for φ=1.0

0

0.2

0.4

0.6

0.8

1

0 1 2 3 4 5 6 7 8 9 10x/De

U/U

ce

Geometry C (PJ) Expt.(PJ) Perry & Hausler

Geometry C (SJ)Expt.(SJ) Perry & Hausler

Figure 4.2(a): Centreline decay of peak axial velocity for primary and secondary jet

(geometry C, φ=1.0).

68


Figure 4.2(b) shows the velocity decay along the geometric centreline of the primary jet

and secondary jet in geometry D for φ=1.0. The behaviour was similar to geometry C at

the exit of the nozzle, where the decay rate for both the primary and secondary jet was

slow up to x/De =3.0 and then increased rapidly further downstream. Initially this decay

was due to momentum diffusion within and close to the cavity resulting from the

expansion of both the primary and secondary jets. Further downstream the decay

occurred due to entrainment into the primary jet from its surroundings consistent with

the observations of Perry and Hausler (1984) and Hart (2001).

Centreline decay of peak axial velocity for φ=1.0

0

0.2

0.4

0.6

0.8

1

0 1 2 3 4 5 6 7 8 9 10

x/De

U/U

ce

Geometry D (PJ) Expt.( PJ) Perry & HauslerGeometry D( SJ )Expt. (SJ) Perry & Hausler

Figure 4.2(b): Centreline decay of peak axial velocity for primary and secondary jet

(geometry D, φ=1.0).

The secondary jet decayed faster than the primary. The velocity for geometry D at the

nozzle was significantly higher due to low static pressure compared to geometry A.

That is why for geometry D the normalized velocities both for primary and secondary

jet at the exit of the burner were higher than 1.0. The predicted decay rate for the

primary jet matched reasonably with the experimental results. Although the simulated

secondary jet decay rate deviated slightly from the experimental results, the trend was

similar in that it decayed faster than the primary jet.

Figure 4.3 shows the comparison of the primary jet decay rate between geometries C

and D for φ= 1.0. For geometry D, the initial decay rate of the primary jet was much

slower than that observed for geometry C due to the reduced cavity diffusion rate.

Although the experimentally measured values at x/De =9.0 for geometries C and D were

69


the same, the numerical results under predicted the decay rate for geometry C. The

experimental values for geometry D matched reasonably well with the predictions.

Centreline decay (PJ) of peak axial velocity for φ=1.0

0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5 6 7 8 9 10x/De

U/U

ceGeometry CExpt. Geometry CGeometry DExpt. Geometry D

Figure 4.3: Comparison of centreline decay of the primary jet between geometry C and

D (φ=1.0).

Figures 4.4(a) and 4.4(b) show the primary jet decay rate for geometries C and D

respectively for φ=1.0, 1.4 and 3.0. For geometry C at φ=1.4 the decay rate increased

within and downstream of the cavity compared with φ=1.0. For φ=3.0 the peak

centreline value began to increase with increasing x/De beyond x/De=5. This was

attributed to the two secondary jets mixing across the whole of the primary field and

increasing fluid momentum in this region (Perry, 1984). In geometry D with increasing

φ, the primary jet decay rate increased both within and downstream of the cavity, and

for φ=3.0 the primary jet disappeared around x/De =4.5. As with trends observed for

geometry C, the secondary jets mixed across the primary field beyond x/De=5

increasing momentum in this region. This phenomenon can be understood more clearly

with the help of figures 4.5(a-c). Figures 4.5(a) and 4.5(b) show the shaded velocity

contours for the primary jet with φ=3.0 in the xy plane at z=0 for geometries C and D

respectively. For geometry D, the primary jet diverted a long way from the geometric

axis towards the long wall of the recess. The jet penetrated up to x/De=4.0, at which

point it disappeared, apparently reappearing after x/De=6.0. These CFD results confirm

the original observations of Perry & Hausler (1984) that it was not the primary jet,

which reappeared, rather it was the appearance of the secondary jets on the primary jet

axis, as shown in figure 4.5(c). For geometry C (figure 4.5(a)) the jet diverged

significantly but to a lesser extent than for geometry D.

70


Centreline decay (PJ) of peak axial velocity for Geometry C

0

0.2

0.4

0.6

0.8

1

1.2

1.4

0 1 2 3 4 5 6 7 8 9 10

x/De

U/U

ce

φ=1 .0φ=1 .4φ=3 .0

Figure 4.4(a): The effect of jet velocity ratio on centreline velocity decay for primary jet

(geometry C).

Centreline decay (PJ) of peak axial velocity for Geometry D

0

0.2

0.4

0.6

0.8

1

1.2

1.4

0 1 2 3 4 5 6 7 8 9 10

x/De

U/U

ce

φ=1 .0φ=1 .4φ=3 .0

Figure 4.4(b): The effect of jet velocity ratio on centreline velocity decay for primary jet

(geometry D).

Figure 4.5(a): Velocity contour at the centre plane of the primary jet for geometry C at φ

= 3.0.

71


Figure 4.5(b): Velocity contour at the centre plane of the primary jet for geometry D at

φ = 3.0.

Figure 4.5(c): Mixing of Secondary jets at the centre plane of the primary jet for

geometry D at φ=3.0

Figures 4.6(a-c) show the transverse velocity profiles for geometry C in the xy plane at

z=0 through the centre of the primary jet for φ=1.0. The comparison has been made at a

number of positions downstream of the jet, at x/De=1.0, 5.0 and 9.0. In figure 4.6(a), the

simulated jet and the experimental jet showed similar behaviour. The prediction of peak

velocity matched reasonably well while the boundaries were a little wider on both long

and short wall side than the measurements. The peak value in the centreline of the

72


geometric axis was over predicted at x/De=5.0 (figures 4.6(b)), where the boundary on

the long side matched very well but it was thinner on the short side. At this position

there was no indication of deviation of jet from the geometric axis, whereas for

Geometry D, both the measured and simulated jets deviated by a significant amount. At

x/De=9.0, figure 4.6(c), the simulated jet had a higher peak velocity than the

measurement. The boundary on the short side was thinner while it was wider on the

side. The simulated jet deviated from the geometric axis, while the measured jet did not.

With an increase in φ from 1.0 to 1.4, figure 4.7(a-c), the jet deviated towards the long

side, but the deviation was not as pronounced as it was in geometry D. For φ=3.0, figure

4.8(a-c), the deviation was more than φ=1.4. In figure 4.8(c) the peak velocity at

x/De=9.0 was more than at x/De=5.0 because of the increased momentum resulting

from the mixing of the secondary jets with the primary jet.

Transverse Velocity Profile (φ=1) for Geometry C

0

0.2

0.4

0.6

0.8

1

-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1y/De

U/U

ce

x/De=1

Experiment

Figure 4.6(a): Transverse velocity profiles at x/De=1.0 (geometry C, φ=1.0).


0

0.2

0.4

0.6

0.8

1

-2 -1.5 -1 -0.5 0 0.5 1 1.5 2y/De

U/U

ce

x/De=5

Experiment

Figure 4.6(b): Transverse velocity profiles at x/De=5.0 (geometry C, φ=1.0).

73



0

0.2

0.4

0.6

0.8

-2 -1.5 -1 -0.5 0 0.5 1 1.5 2

y/De

U/U

ce

x/De=9

Experiment

Figure 4.6(c): Transverse velocity profiles at x/De=9.0 (geometry C, φ=1.0).

Transverse Velocity Profile (φ=1.4) for Geometry C

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1y/De

U/U

ce

x/De=1

Experiment



0

0.1

0.2

0.3

0.4

0.5

0.6

-1.5 -1 -0.5 0 0.5 1 1.5y/De

U/U

ce

x/De=5

Experiment


74



0

0.1

0.2

0.3

0.4

0.5

-2 -1.5 -1 -0.5 0 0.5 1 1.5 2

y/De

U/U

ce

x/De=9

Experiment



0

0.05

0.1

0.15

0.2

0.25

0.3

-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1

y/De

U/U

ce

x/De=1

Experiment



0

0.05

0.1

0.15

0.2

0.25

-1 -0.5 0 0.5 1 1.5y/De

U/U

ce

x/De=5

Experiment


75



0

0.05

0.1

0.15

0.2

0.25

0.3

-2 -1.5 -1 -0.5 0 0.5 1 1.5 2

y/De

U/U

ce

x/De=9

Experiment


Transverse velocity profiles for geometry D in the xy plane at z=0 through the centre of

the primary jet for φ=1.0 are shown in figures 4.9(a-c). The locations for the comparison

were same as they were for geometry C i.e. x/De=1.0, 5.0 and 9.0. The figures show

that generally the behaviour of the simulated jet was same as observed in the physical

model. However, there were some discrepancies between the predicted results and the

experimental values. At x/De=1.0, figure 4.9(a), the peak value of the prediction

matched well matched with the experimental results but the boundaries were slightly

wider on both the long and short sides than the measurements indicated.

Transverse Velocity Profile (φ=1.0) for Geometry D

0

0.2

0.4

0.6

0.8

1

-0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1 1.2

y/De

U/U

ce

x/De=1.0

Experiment

Figure 4.9(a): Transverse velocity profiles at x/De=1.0 (geometry D, φ=1.0).

At x/De=5.0, figure 4.9(b), the predicted profile on the long side matched well but the

measured profile had a lower peak velocity in the centre of the jet and the boundary was

thinner on the short wall side. There is a clear indication of the deviation of the jet

76


towards the long wall side at this position, although the jet in physical model deviated

more from the geometric axis than the simulated jet.


0

0.2

0.4

0.6

0.8

1

-1 -0.5 0 0.5 1 1.5 2 2.5

y/De

U/U

ce

x/De=5.0

Experiment

Figure 4.9(b): Transverse velocity profiles at x/De=5.0 (geometry D, φ=1.0).

At x/De=9.0, figure 4.9(c), the measured jet had deviated further from the geometric

axis and the profile had become wider than at x/De=5.0. The simulated jet moved

further off from the geometric axis than the physical model and the peak velocity also

reduced but not enough to match the experiments.


0

0.2

0.4

0.6

0.8

1

-1 -0.5 0 0.5 1 1.5 2 2.5y/De

U/U

ce

x/De=9.0

Experiment

Figure 4.9(c): Transverse velocity profiles at x/De=9.0 (geometry D, φ=1.0).

With the increase of φ from 1.0 to 1.4 the overall behaviour of the jets remained

unchanged, only the extent of the deviation from the geometric axis changed. Figures

4.10(a-c) shows the transverse velocity profile for φ=1.4. At x/De=5.0, figure 4.10(b),

the deviation from the geometric axis was more than for φ=1.0. This can be seen more

clearly at x/De=9.0 where the deviation of the peak values of the simulated jet for φ=1.0

and φ=1.4 are 0.55m and 1.1m respectively. For φ=3.0, the jet started deviating shortly

77


after exiting the nozzle and the deviation was 0.2m at x/De=1.0 as shown in figure 4.11.

This trend continued with downstream and at x/De=4.5, the jet apparently disappeared

discussed earlier.

Transverse Velocity Profile(φ=1.4) for Geometry D

0

0.2

0.4

0.6

0.8

1

-0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8y/De

U/U

cex/De=1

Experiment

Figure 4.10(a): Transverse velocity profiles at x/De=1.0 (geometry D, φ=1.4).


0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

-1.2 -0.8 -0.4 0 0.4 0.8 1.2 1.6 2 2.4

y/De

U/U

ce

x/De=5

Experiment

Figure 4.10(b): Transverse velocity profiles at x/De=5.0 (geometry D, φ=1.4).


0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

-1.2 -0.8 -0.4 0 0.4 0.8 1.2 1.6 2 2.4

y/De

U/U

ce

x/De=9

Experiment

Figure 4.10(c): Transverse velocity profiles at x/De=9.0 (geometry D, φ=1.4).

78


Transverse Velocity Profile (φ=3) for Geometry D

0

0.1

0.2

0.3

0.4

0.5

-0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1y/De

U/U

ce

x/De=1

Experiment

Figure 4.11: Transverse velocity profiles at x/De=1.0 (geometry D, φ=3.0).

Figures 4.12(a) and 4.12(b) show the extent of deviation of the jet away from the

geometric axis for geometry C and D. With an increase in φ the deviation increased both

for geometry C and D. However for φ=3.0, the deviation at x/De=9.0 was less than

x/De=5.0 both for geometry C and D. The deviation for geometry D was more than

geometry C for the same jet velocity ratio.

Deviation of jet centre away from geometric axis

0

0.1

0.2

0.3

0.4

0.5

0 1 2 3 4 5 6 7 8 9 10

x/De

Dev

iatio

n/D

e

φ=1 .0φ=1 .4φ=3 .0

Figure 4.12(a): Deviation of jet centre with an increase in φ for Geometry C (simulated

jet).

79


Deviation of jet centre away from geometry axis

00.20.40.60.8

11.21.41.6

0 1 2 3 4 5 6 7 8 9 10

x/De

Dev

iatio

n/D

e

φ=1 .0φ=1 .4φ=3 .0

Figure 4.12(b): Deviation of jet centre with an increase in φ for Geometry D (simulated

jet).

The comparisons of deviation of the centreline between the simulated and measured

results for geometries C and D for φ=3.0 are shown in figures 4.13(a) and 4.13(b). In

figure 4.13(a) the values were well matched near to the wall up to x/De=1.5. With

increased downstream distance the numerical results under predicted the jet

displacement and the difference was largest at x/De=9.0. For geometry D, figure

4.14(b), the predictions for the deviation were very well matched up to x/De=4.0 and

after that the numerical results were under predicted like geometry C.

Deviation of jet centre away from geometry axis (φ=3.0)

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0 1 2 3 4 5 6 7 8 9 10x/De

Dev

iatio

n/D

e

Simulation

Experiment

Figure 4.13(a): Validation of the simulated jet deviation with the experimental values

for geometry C (φ=3.0).

80


Deviation of jet centre away from geometric axis (φ=3.0)

0

0.2

0.4

0.6

0.8

1

1.2

1.4

0 1 2 3 4 5 6 7 8 9 10

x/De

Dev

iatio

n/D

e

Simulation

Experiment

Figure 4.13(b): Validation of the simulated jet deviation with the experimental values

for geometry D (φ=3.0).

4.2.2. Flow field prediction

Geometry C

Figure 4.14(a) shows the static pressure distribution in the xy plane through the centre

of the primary jet for geometry C and at φ=1.0. Pressure inside the recess was below

atmospheric pressure. There was a low-pressure region on both sides of the nozzle. On

the long side, the low-pressure region was connected to the low-pressure zone inside the

recess but on the short side it was located entirely outside the recess. Figure 4.14(b)

shows the velocity vectors in the same plane. The creation of the low-pressure regions

on either side of the jet near the wall played a major role for the entrainment.

Entrainment of the surrounding fluid occurred as soon as the jet exited from the recess.

On the short side the vector field was normal to the geometric axis of the burner

whereas on the long side, near to the wall, entrainment vectors were inclined at an angle

of 30ο with the furnace wall and as the jet advanced farther downstream, the

entrainment vectors became parallel to the furnace wall.

81


(a) (b)

Figure 4.14: Pressure distribution (a) and velocity vector (b) in the xy plane through the

centre of the primary axis for geometry C (φ=1.0)

With the increase in jet velocity ratio from 1.0 to 1.4 there was no significant change in

the pressure field or the velocity field, except in the deviation of the jet from the

geometric axis. Figure 4.15(a) shows pressure contours at the same level for φ=1.4. The

low-pressure region on the short side was larger. As for φ=1.0, the pressure was sub-

atmospheric inside the recess and gradually increased after the jet entered into the

domain. The entrainment vector field was also similar to φ=1.0 on both sides.

Figure 4.16(a) shows the pressure distribution for φ=3.0 at the same level. The pressure

field was changed completely on the short side, with a large low-pressure region centred

at around x/De=5.0. Due to this low-pressure region, the surrounding fluid exerted an

external force to the primary jet, causing the jet to deviate from the geometric axis

towards the long side of the nozzle as shown in figure 4.5(a). The entrainment on the

short side was parallel to the furnace wall (figure 4.16(b)). On the long side, vector field

was similar to figure 4.15(b).

82


(a) (b)



(a) (b)



Geometry D

Figure 4.17(a) shows the pressure distribution for geometry D in the xy plane through

the centre of the primary jet for φ=1.0. On the short side, there was a large low-pressure

region connected to the low-pressure zone inside the recess. Entrainment of the

surrounding fluid into the recess occurred due to this low-pressure region. On the long

side, the low-pressure zone inside the recess was also extended into the domain but was

smaller than the short side. Figure 4.17(b) shows the velocity vectors in the same plane

83


for φ=1.0. The primary jet separated completely from the short wall whereas it neither

completely attached nor separated from the long wall, consistent with the observations

of Perry and Hausler (1982) and Hart (2001). The separation of the primary jet from the

short wall resulted in entrainment of fluid into the recess on that side. Because of the

partial attachment to the long wall and separation from the short wall the jet was pushed

off the geometric axis towards the long wall side.

(a) (b)


centre of the primary axis for geometry D (φ=1.0)

With the increase in jet velocity ratio from 1.0 to 1.4, figure 4.18(a), the entrainment of

the surrounding fluid inside the recess was increased on the short side i.e. the thickness

of separation from the short side was increased. On the long side, figure 4.18(b), the

primary jet completely separated from the long wall. As a result, the jet deviated more

than that of φ=1.0 towards the long side.

Figure 4.19(a) shows the pressure distribution for φ=3.0 at the same level. The thickness

of separation from the short side was further increased as shown in figure 4.19(b). A

large low-pressure zone was existed in front of the nozzle at around x/De=4.5. The

surrounding fluid moved towards the low-pressure region and exerted an external

impact to the primary jet. The momentum due to entrainment as well as internal

reorganization of pressure was responsible for the deviation of the primary jet from the

geometric axis and the jet apparently disappeared as mentioned earlier.

84


(a) (b)



(a) (b)



85


4.3. Summary and conclusions

A numerical investigation of two recessed burner geometries has shown that burner

geometry and jet velocity ratio have a significant influence on jet development.

Comparisons have been made between the numerical results with the experimental data

(Perry & Hausler, 1984) and reasonably good agreement was found. For a burner

geometry where the cavity had parallel sidewalls (geometry C), the primary jet

developed along the geometric axis of the burner when the secondary to primary jet

velocity was unity but tended to move away from this axis towards the long side wall of

the nozzle when the velocity ratio was increased. By replacing the parallel walls in the

cavity by diverging walls (geometry D) the centreline decay rate within the cavity was

reduced due to a reduced cavity diffusion rate but increased rapidly some distance

downstream as it discharged into the furnace. The primary jet deflected strongly

towards the long face of the cavity for all velocity ratios. The degree of deflection of the

primary jet increased with increasing φ. For a jet velocity ratio of 3.0, the primary jet

apparently disappeared at around four and a half diameters downstream of the jet exit

and became very unstable.

This numerical investigation described the effect of recessed burner geometry and jet

velocity ratio for the development of jet with out cross-flow. In a tangentially-fired

furnace, cross-flow has a profound effect on flow development in the near and far field

region of jets. The next chapter describes the effect of cross-flow on the development of

rectangular burner jets, for a variety of secondary to primary jet velocity ratios.

86

Chapter-5 Experimental investigation of rectangular

slot-burners with cross-flow

Chapter 5- Experimental investigation of rectangular slot- burners with cross-flow

5. Experimental investigation of rectangular slot-burners with cross-flow

This chapter presents the experimental investigation of rectangular slot-burners in the

presence of cross-flow for different jet velocity ratios. Burners B and D have been

studied for secondary to primary jet velocity ratios,φ, of 1.0 and 3.0. The primary to

cross-flow jet velocity ratio (ϕ) was 1.0 and was held constant throughout the

experiment. In both cases the jets discharged at an angle of 60° to the furnace wall. The

detailed description of geometry B, D, cross-flow duct and burner model are given in

chapter 3. One side of the cross-flow jet was attached to the wall and the jet induced a

circulation in the burner model. This was done to obtain a burner flow field similar to a

tangentially-fired furnace. Laser Doppler Anemometry (LDA) experiments were carried

out to measure the mean velocity component and turbulent fluctuations in the near field

region of the jets.


5.1.1. Geometry B

5.1.1.1. Flow field for φ= 1.0 and 3.0

Figure 5.1(a-c) shows the measured velocity vectors for secondary to primary jet

velocity ratio,φ, of 1.0. The planes are horizontal planes located through the centre of

the primary jet, the lower base region (region between primary jet and lower secondary

jet) and through the centre of the lower secondary jet respectively. The lower secondary

jet plane was chosen because it was farthest away from the influence of the roof surface.

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

y /D e

1

2

3

4

5

6

7

8

x/D

e

Figure 5.1(a): Velocity vectors in the centre plane of the primary jet for φ=1.0

88


-1 0 1 2 3 4 5 6 7 8 9y/De

1

2

3

4

5

6

7

8

x/D

e

Figure 5.1(b): Velocity vectors in the centre plane of the lower base region for φ=1.0

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

y /D e

1

2

3

4

5

6

7

8

x/D

e

Figure 5.1(c): Velocity vectors in the centre plane of the lower secondary jet for φ=1.0

The direction of the cross-flow jet was the same as the component of the jet velocity

parallel to the furnace wall. The cross-flow has a profound effect on the developing

flow field. For this burner geometry without cross-flow, (Perry & Hausler, 1984), the

three jets were aligned almost along the geometric axis of the burner. In the presence of

cross-flow the jets were pushed towards the wall and remained predominantly within

the cross-flow layer. The deflection of the lower secondary jet was slightly greater than

the primary jet as shown in figure 5.1(c).

Figure 5.2(a-c) shows the measured velocity vectors in the same planes but for φ=3.0.

The flow field changed significantly with the increase in secondary jet velocity. The

degree of deflection of primary jet towards the wall was reduced which is clear from

figure 5.2(a). The primary jet penetrated through the cross-flow layer whereas it was

almost entirely contained within the cross-flow for φ=1.0. Due to the mixing of the

89


secondary jets with the primary jet, the momentum of the primary jet was increased

allowing it to penetrate through the cross-flow. This phenomenon can be understood

more clearly in figure 5.2(c) where the deflection of the lower secondary jet from the

geometric axis was minor due to its high momentum showing that the jet pierced the

cross-flow layer.

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

y /D e

1

2

3

4

5

6

7

8

x/D

e


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

y/De

1

2

3

4

5

6

7

8

x/D

e

Figure 5.2(b): Velocity vectors in the centre plane of the lower base region for φ=3.0

90


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

y /D e

1

2

3

4

5

6

7

8

x/D

e

Figure 5.2(c): Velocity vectors in the centre plane of the lower secondary jet for φ=3.0

5.1.1.2. Comparison of mean and RMS velocity

Comparison of resultant velocity for φ=1.0 and 3.0 at the centre of the primary jet, the

lower base region and lower secondary jet are presented in figures 5.4(a-b), 5.5(a-b) and

5.6(a-b) respectively. Starting from y/De=0, comparisons have been made for a number

of locations downstream of the nozzles. The measurement locations y/De=0, 0.5, 1.0,

2.0, 3.0, 4.0, 5.0 and 9.0 are indicated in figure 5.3. The mean velocities were measured

in two perpendicular directions with U component normal and V component parallel to

the cross-flow and resultant velocity, Ures, has been plotted for comparison. All

velocities are normalized to the centreline velocity, Uce, at the exit of the primary nozzle

of φ=1.0 (x/De=0, y/De=0, z/De=0).

x y/De=0 y/De=9

y

x/De=0, y/De=0, z/De=0

Figure 5.3: Schematic diagram showing the measurement positions

91


In figure 5.4(a-b), at y/De=0, the peak velocity occurred at the exit of the nozzle

(x/De=0) for both jet velocity ratios. The velocities then reduced and reached a second

peak at x/De=1.1. There was little difference in the velocity distribution at y/De=0

between φ=1.0 and 3.0. The difference became noticeable from y/De=1.0, where the

peak values occurred at the same position (x/De=0.8) for both jet velocity ratios

although the spreading of the jet for φ=3.0 was greater than for φ=1.0. Further

downstream (y/De=2.0-9.0), the peak velocity for φ=3.0 occurred farther from the wall

than for φ=1.0 and the difference was clear at y/De=5.0 where the peak velocity for

φ=1.0 occurred at around x/De=2.2 and for φ=3.0 at x/De=3.8. At y/De=9.0 the

difference between the location of peak velocities was at a maximum. This clearly

indicates more spreading of the primary jet and less deviation from the geometric axis

for φ=3.0.

Velocity profiles for φ= 1.0

0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5x/De

Ure

s/U

ce

y/De=0y/De=0.5y/De=1y/De=2

Velocity profiles for φ=1.0

0

0.2

0.4

0.6

0.8

1

0 1 2 3 4 5 6x/De

Ure

s/U

ce

y/De=3y/De=4y/De=5y/De=9

Figure 5.4(a): Velocity profiles in the centre plane of the primary jet for φ=1.0


0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5 6x/De

Ure

s/U

ce



0

0.2

0.4

0.6

0.8

1

0 2 4 6 8 10 12 14x/De

Ure

s/U

ce


Figure 5.4(b): Velocity profiles in the centre plane of the primary jet for φ=3.0

Figures 5.5(a-b) show the velocity profiles in the centre plane of the lower base region

for φ=1.0 and 3.0. The primary jet and lower secondary jet mixed some distance away

from the wall and there was no flow near the wall. At y/De=0.5, there was a difference

92


in the profile between φ=1.0 and 3.0. For φ=1.0 there were two peaks, the first at

x/De=0.31 and the second at x/De=1.0. For φ=3.0 there was only one peak at x/De=1.0.

At y/De=1.0, the velocity was higher in the near wall region for φ=1.0 compared to

φ=3.0 and the more noticeable differences in the profiles started at this location. Further

downstream (y/De=2.0 to 5.0) the peak value for φ=3.0 shifted farther from the wall

than φ=1 similar to figure 5.4. The velocity profile at y/De=9.0 was totally different for

φ=1.0 and φ=3.0. For φ=1.0, it was uniform and had a constant value of 0.56 up to

x/De=2.5 and then gradually decreased. For φ=3.0, starting from a minimum value

(Ures/Uce=0.2) the profile showed a large peak at x/De=6.5. The difference between the

peak values was largest at this position.

Velocity profiles for f= 1.0

0

0.2

0.4

0.6

0.8

1

0 1 2 3 4 5x/De

Ure

s/U

ce


Velocity profiles for f= 1.0

0

0.2

0.4

0.6

0.8

1

0 1 2 3 4 5x/De

Ure

s/U

ce

6


Figure 5.5(a): Velocity profiles in the centre plane of lower base region for φ=1.0


0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5 6


0

0.2

0.4

0.6

0.8

1

1.2

0 2 4 6 8 10 12 14x/De

Ure

s/U

ce


x/De

Ure

s/U

ce


Figure 5.5(b): Velocity profiles in the centre plane of lower base region for φ=3.0

Figures 5.6(a-b) show a comparison of the velocity profiles in the centre plane of the

lower secondary jet for φ=1.0 and φ=3.0. For φ=1.0, at y/De=0, the magnitude of the

peak value was 1.0 and for φ=3.0 it was at around 2.5. At y/De=1.0, the difference in

magnitude of the peak values was almost double. Similar to figure 5.5, this was the

93


starting point of deviation of peak values between φ=1.0 and φ=3.0. Further downstream

(y/De=3.0-5.0), the peak values for φ=1.0 occurred almost in the same position but for

φ=3.0, they shifted farther from the wall and again the difference was maximum at

y/De=9.0. The velocity profiles at y/De=9.0 clearly indicate that the jets with φ=3.0

moved far away from the wall that is less deviation from the geometric axis.


0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5


0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5x/De

Ure

s/U

ce6


x/De

Ure

s/U

ce


Figure 5.6(a): Velocity profiles in the centre plane of the lower secondary jet for φ=1.0


0

0.5

1

1.5

2

2.5

3

0 1 2 3 4 5 6


0

0.2

0.4

0.6

0.8

1

1.2

0 2 4 6 8 10 12 14x/De

Ure

s/U

ce


x/De

Ure

s/U

ce


Figure 5.6(b): Velocity profiles in the centre plane of the lower secondary jet for φ=3.0

The stress distribution can be explained with the help of Reynolds stress equations. Full

development and description of the equations required for explanation are included in

Appendix I. Figures 5.7(a-c) show the rms values of the fluctuating component, urms,

vrms and the component of the shear stress, uv, at the centre of the primary jet for φ=1.0.

urms and vrms were normalized to the centreline exit velocity and uv was normalized to

the square of the centreline exit velocity. In figure 5.7(a), at y/De=0, very near to the

wall (x/De=0.31), there was a non-zero value (0.06) of urms. This non-zero value might

occur due to diffusion transport in the cross-stream directions from regions of peak

generation. After that there was a sudden peak of urms at x/De=0.625. As the jet exited

94


from the duct earlier on the short side, it expanded. Due to sudden change in velocity,

there was a steeper velocity gradient, which might cause an increase in urms at this

position. The magnitude then fell, gradually increased and reached the second peak at

x/De=2.2. This second peak occurred in the shear layer between the primary jet and the

surrounding fluid where the velocity gradient was again high. At y/De=1.0, peak value

occurred at x/De=0.31 because of the generation of turbulence due to high velocity

gradient. The generation of turbulence can be understood more clearly from figures

5.7(b) and 5.7(c).

u rms for Velocity Ratio 1.0

0

0.05

0.1

0.15

0.2

0.25

0.3

0 1 2 3 4 5 6 7 8x/De

u rm

s/U

ce



0

0.05

0.1

0.15

0.2

0.25

0.3

0 1 2 3 4 5 6 7 8x/De

u rm

s/U

ce


Figure 5.7(a): urms at the centre of the primary jet for φ=1.0

v rms for Velocity Ratio 1.0

0

0.050.1

0.150.2

0.250.3

0.350.4

0.45

0 1 2 3 4 5 6 7 8x/De

v rm

s/U

ce



0

0.05

0.1

0.15

0.2

0.25

0.3

0 1 2 3 4 5 6 7 8x/De

v rm

s/U

ce


Figure 5.7(b): vrms at the centre of the primary jet for φ=1.0

95


uv stress for Velocity Ratio 1.0

-0.025

-0.02

-0.015

-0.01

-0.005

0

0.005

0.01

0.015

0 1 2 3 4 5 6 7 8x/De

uv/U

ce2



-0.025

-0.02

-0.015

-0.01

-0.005

0

0.005

0.01

0.015

0 1 2 3 4 5 6 7 8x/De

uv/U

ce2


Figure 5.7(c): uv stress at the centre of the primary jet for φ=1.0

In figure 5.7(b), at y/De=2.0 and for x/De=0.625, vrms as well as velocity gradient was

high. The creation of shear stress is strongly dependent with these two terms. As a result

shear stress uv was high at this location (figure 5.7(c)) which is again one of the main

components for the production of turbulence in the urms equation as given in Appendix I.

Further downstream (y/De=3.0-9.0), the peak values of urms shifted and occurred at the

centreline of the jet. The peak values in this region occurred because of diffusive

redistribution of the normal stresses from the cross-stream directions.

Figures 5.8(a) shows the urms values at the centre of the primary jet for φ=3.0. At

y/De=0 stress distribution was similar to figure 5.7(a). At y/De=1.0 urms was high (0.22)

near to the wall because of the generation of turbulence due to high velocity gradient

(figure 5.4(b).

u rms for velocity Ratio 3.0

0

0.05

0.1

0.15

0.2

0.25

0 1 2 3 4 5 6 7 8x/De

u rm

s/U

ce



0

0.05

0.1

0.15

0.2

0.25

0.3

0 2 4 6 8 10 12 14x/De

u rm

s/U

ce


Figure 5.8(a): urms at the centre of the primary jet for φ=3.0

96



0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0 1 2 3 4 5 6 7 8x/De

v rm

s/U

cey/De=0y/De=0.5y/De=1y/De=2


0

0.05

0.1

0.15

0.2

0.25

0.3

0 2 4 6 8 10 12 14x/De

v rm

s/U

ce


Figure 5.8(b): vrms at the centre of the primary jet for φ=3.0


-0.02-0.015

-0.01-0.005

00.005

0.010.015

0.02

0 1 2 3 4 5 6 7 8x/De

uv/U

ce2



-0.015

-0.01

-0.005

0

0.005

0.01

0.015

0 2 4 6 8 10 12 14x/De

uv/U

ce2


Figure 5.8(c): uv stress at the centre of the primary jet for φ=3.0

The distribution of urms at y/De= 3.0, 4.0, 5.0, and 9.0 were quite different than from

φ=1.0. With increasing distance downstream of the nozzle, the peak values of urms

gradually increased up to y/De=5.0 and shifted farther away from the wall. This may be

due to generation of turbulence arising from sharp velocity gradient that existed

between the primary and secondary jets. However this high turbulence generated at the

shear layer between the primary and secondary jets took some time to diffuse to the

centre of the primary jet and hence did not show up near the nozzle exit. The magnitude

of the urms values were high and at y/De=5.0 it reached at a maximum (0.25). These

high values persisted over a much wider region as shown in figure 5.8(a).

Figures 5.8(b) show vrms values at the same plane for φ=3.0. Unlike urms, there was only

one peak at y/De=0. At y/De=1.0, the peak value occurred near to the wall (x/De=0.31)

because of high velocity gradient. Further downstream (y/De=2.0-9.0) the peak values

for the vrms shifted farther from the wall and occurred at the centreline of the jet axis.

This might happen due to the diffusive redistribution of the stresses. The difference in

97


magnitudes and distribution of urms and vrms, both for φ=1.0 and 3.0, clearly indicates the

anisotropic nature of the turbulence.

Figure 5.8(c) shows the uv stress distribution at the same level for φ=3.0. The Reynolds

stress arises from the correlation of two components of the velocity fluctuation at the

same point. A non-zero value of this correlation implies that the two components were

not independent of one another. In figure 5.8(c), at y/De=2.0, the uv stress was high for

x/De=0.93 because of the high velocity gradient at this location.

Comparisons of the turbulent velocity fluctuations and uv shear stress in the centre

plane of lower secondary jet for φ=1.0 and 3.0 are shown in figures 5.9 and 5.10. Due to

the increase in secondary jet velocity the distribution for φ=3.0 was quite different than

for φ=1.0.


0

0.05

0.1

0.15

0.2

0.25

0.3

0 1 2 3 4 5 6 7 8

x/De

u rm

s/U

ce



0

0.05

0.1

0.15

0.2

0.25

0.3

0 1 2 3 4 5 6 7 8x/De

u rm

s/U

ce


Figure 5.9(a): urms at the centre of the lower secondary jet for φ=1.0


00.05

0.1

0.150.2

0.250.3

0.350.4

0.45

0 1 2 3 4 5 6 7 8x/De

v rm

s/U

ce



0

0.05

0.1

0.15

0.2

0.25

0 1 2 3 4 5 6 7 8x/De

v rm

s/U

ce


Figure 5.9(b): vrms at the centre of the lower secondary jet for φ=1.0

98



-0.02

-0.015

-0.01

-0.005

0

0.005

0.01

0.015

0 1 2 3 4 5 6 7 8

x/De

uv/U

ce2



-0.02

-0.015

-0.01

-0.005

0

0.005

0.01

0.015

0 1 2 3 4 5 6 7 8x/De

uv/U

ce2


Figure 5.9(c): uv stress at the centre of the lower secondary jet for φ=1.0


00.05

0.10.150.2

0.250.3

0.350.4

0.450.5

0 1 2 3 4 5 6 7 8x/De

u rm

s/U

ce



0

0.050.1

0.150.2

0.25

0.30.35

0.40.45

0 2 4 6 8 10 12 14x/De

u rm

s/U

cey/De=3y/De=4y/De=5y/De=9

Figure 5.10(a): urms at the centre of the lower secondary jet for φ=3.0


0

0.1

0.2

0.3

0.4

0.5

0.6

0 1 2 3 4 5 6 7 8x/De

v rm

s/U

ce



0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

0 2 4 6 8 10 12 14x/De

v rm

s/U

ce


Figure 5.10(b): vrms at the centre of the lower secondary jet for φ=3.0

99



-0.08

-0.06

-0.04

-0.02

0

0.02

0 1 2 3 4 5 6 7 8x/De

uv/U

ce2



-0.045

-0.035

-0.025

-0.015

-0.005

0.005

0.015

0 2 4 6 8 10 12 14x/De

uv/U

ce2


Figure 5.10(c): uv stress at the centre of the lower secondary jet for φ=3.0

For φ=1.0, figure 5.9(a), the highest peak value of urms occurred near to the wall

(x/De=0.31) at y/De=1.0, after which the magnitudes of the peak values decreased and

occurred farther from the wall. For φ=3.0, figure 5.10(a), the magnitudes of the peak

values were high and gradually increased up to y/De=2.0. At y/De=2.0, the peak value

of urms (0.43) occurred at the centre of the jet. This can be understood more clearly by

referring to figure 5.6(b) where at y/De=2.0 the position of the peak velocity was at

around x/De=2.6. Farther downstream (y/De=3.0-9.0) the magnitude of the peak values

decreased gradually. The peak values at these positions occurred because of the

interaction of the shear layer between the lower secondary jet and the surrounding fluid.

At these positions the uv stresses were also high which is shown in figure 5.10(c). 5.1.2. Geometry D

The jet nozzles in geometry D were recessed into the wall. This burner configuration

was similar the type used in the Yallourn stage-2 power stations. As discussed in

chapter 4, without cross-flow and for φ=1.0, flow patterns in the recess were complex,

caused by a combination of the internal reorganization of energy in the jet, adverse

pressure gradients in the recess and entrainment of fluid from the open atmosphere into

the recess. For φ=3.0, the primary jet disappeared at x/De=4.5 and its apparent

reappearance was in fact the two secondary jets meeting at the primary jet plane. It is of

great interest to determine the effect of cross-flow on geometry D. Next section

discusses the flow patterns in near field region of geometry D in the presence of cross-

flow.

100


5.1.2.1. Flow field for φ= 1.0 and 3.0

Figure 5.11(a-b) shows the measured velocity vectors for φ=1.0. The planes are through

the centre of the primary jet and lower secondary jet. As observed for geometry B, the

presence of cross-flow has a profound effect on the development of the burner flow

field for geometry D. Perry et al. (1986) investigated the flow field for geometry D

without cross-flow and found that for φ=1.0 the three jets deflected from the geometric

axis of the burner towards the long side of the cavity. The angle of deviation was 9°. In

the presence of cross-flow the three jets diverged completely from the geometric axis of

the burner and attached to the wall. The degree of deflection of the lower secondary jet

was slightly more than the primary jet (figure 5.11(b)).

- 1 . 5 - 1 - 0 . 5 0 0 . 5 1 1 . 5 2 2 . 5 3 3 . 5

y /D e

0 . 5

1

1 . 5

2

2 . 5

3

3 . 5

4

x/D

e


- 1 . 5 - 1 - 0 . 5 0 0 . 5 1 1 . 5 2 2 . 5 3 3 . 5

y /D e

0 . 5

1

1 . 5

2

2 . 5

3

3 . 5

4

x/D

e

Figure 5.11(b): Velocity vectors in the centre plane of the lower secondary jet for φ=1.0

101


For primary jet, there was a recirculation zone very near to the wall which started at

y/De=2.0 and extended up to y/De=3.0. For the lower secondary jet the recirculation

zone started at y/De=2.0 but the region of recirculation zone was smaller and finished

before y/De=3.0.

Figure 5.12(a-b) shows the measured velocity vectors for φ=3.0 in the centre plane of

the primary and lower secondary jet. The flow field of the primary jet (figure 5.12(a))

was markedly different than that of geometry B (figure 5.2(a)). After exiting the nozzle

the primary jet in geometry B pierced the cross-flow layer whereas for geometry D it

appeared to be pushed against the furnace wall consistent with the observation of Yan

and Perry (1994).

- 1 . 5 - 1 - 0 . 5 0 0 . 5 1 1 . 5 2 2 . 5 3 3 . 5

y /D e

0 . 5

1

1 . 5

2

2 . 5

3

3 . 5

4

x/D

e


- 1 . 5 - 1 - 0 . 5 0 0 . 5 1 1 . 5 2 2 . 5 3 3 . 5

y /D e

0 . 5

1

1 . 5

2

2 . 5

3

3 . 5

4

x/D

e

Figure 5.12(b): Velocity vectors in the centre plane of the lower secondary jet for φ=3.0

102


The flow field inside the recess burner was complex, greatly affecting the unconfined

part of the jet. Upon exiting the ducts and entering the recess the velocity of the primary

jet was reduced. The lower momentum primary jet diffused outwards in an attempt to

fill the recess and mixed with high momentum secondary jets. A major portion of the

volume of fluid coming out of the primary duct were sucked into the secondary jets (this

phenomenon will be explained in detail in the next chapter) leading to a situation where

cross-flow jet can easily penetrate through the primary jet. So when the primary jet

exited from the recess it was not the primary jet but the cross-flow jet, which flowed

parallel to the furnace wall. With the increase in secondary jet velocity the lower

secondary jet penetrated through the cross-flow layer (figure 5.12(b)) due to the higher

momentum of the jet with only a small deflection from the geometric axis of the burner.

5.1.2.2. Comparison of mean and RMS velocity

Comparisons of velocity for φ=1.0 and 3.0 in the centre plane of the primary jet, lower

base region and lower secondary jet are presented in figures 5.13(a-b), 5.14(a-b) and

5.15(a-b) respectively. The measurements lines were same as for geometry B in figure

5.3 and the resultant velocity, Ures, was plotted for comparison. All velocities were

normalized to the centreline exit velocity of the primary jet, Uce, located at (x/De=0,

y/De=0, z/De=0) the exit of the recess for φ=1.0.

Figure 5.13(a-b) shows the comparison between the velocity profiles for φ=1.0 and 3.0

in plane at the centre of the primary jet. At y/De=0, for φ=1.0, there were two peaks in

the velocity profile. The first peak was just after the exit from the recess at x/De=0 and

the second peak was at x/De=0.96. For φ=3.0 there was only one peak occurred at

x/De=0.6. The boundary of the velocity profile for φ=3.0 was wider than for φ=1.0.


0

0.2

0.4

0.6

0.8

1

1.2

1.4

0 1 2 3 4 5x/De

Ure

s/U

ce



0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5 6x/De

Ure

s/U

ce


Figure 5.13(a): Velocity profiles in the centre plane of the primary jet for φ=1.0

103



0

0.2

0.4

0.6

0.8

1

1.2

1.4

0 1 2 3 4 5 6


0

0.2

0.4

0.6

0.8

1

0 2 4 6 8 10 12 14x/De

Ure

s/U

ce


x/De

Ure

s/U

cey/De=0y/De=1y/De=2

Figure 5.13(b): Velocity profiles in the centre plane of the primary jet for φ=3.0

At y/De=1.0, although the peak values occurred at the same position (x/De=0.55) for

both velocity ratios, the spreading of the jet for φ=3.0 was greater than for φ=1.0. The

existence of a vortex near to the wall for both φ=1.0 and 3.0 can be understood by

observing the velocity profile at y/De=2.0. For φ=1.0, the area of the recirculation zone

away from the wall extended up to x/De=0.25 whereas for φ=3.0 it extended up to

x/De=0.2. A vortex also occurred at y/De=3.0 for φ=1.0 indicating the extension of the

recirculation zone beyond y/De=3.0. For φ=3.0 the vortex was just finished at y/De=3.0.

This clearly indicates that the area of the recirculation zone for φ=3.0 was smaller than

for φ=1.0. At y/De=5.0, the peak velocity for φ=3.0 occurred farther from the wall than

for φ=1.0 and at y/De=9.0, where the difference between the peak values was at a

maximum.

Figure 5.14(a-b) shows the velocity profile in the plane through the centre at the base

region for φ=1.0 and 3.0. For φ=1.0 the peak velocities were at the same position

(x/De=0.95) up to y/De=2.0 and then shifted farther from the furnace wall. For φ=3.0,

the peaks in the profiles started shifting from the furnace wall at the beginning

(y/De=0.5) and continued with the increase in distance downstream of the nozzle. For

φ=3.0, the magnitude of the peak velocities were higher and the boundaries of the

velocity profiles were wider.

104



0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5x/De

Ure

s/U

ce



0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5x/De

Ure

s/U

ce


Figure 5.14(a): Velocity profiles in the centre plane of the lower base region for φ=1.0


00.20.40.60.8

11.21.41.61.8

0 1 2 3 4 5


0

0.2

0.4

0.6

0.8

1

1.2

1.4

0 2 4 6 8 10 12 14x/De

Ure

s/U

ce


x/De

Ure

s/U

ce


Figure 5.14(b): Velocity profiles in the centre plane of the lower base region for φ=3.0

The recirculation zone existed at y/De=2.0 both for φ=1.0 and 3.0 but finished before

y/De=3.0 indicating that in this plane the area of the recirculation zone was smaller than

in the centre plane of the primary jet. The velocity profile at y/De=9.0 was completely

different for φ=1.0 and 3.0. For φ=1.0 it was uniform in the region 0.4≤x/De≤2.3 and

then gradually decreased. For φ=3.0, starting from a minimum value of 0.42 the velocity

profile showed a large peak at x/De=5.0. The difference between the peak values was

largest at this position.

The comparison of the velocity profiles in the centre plane of the lower secondary jet is

shown in figure 5.15(a-b) for φ=1.0 and 3.0. Due to change in secondary jet velocity

(from 8m/s to 24m/s) the profiles were markedly different between φ=1.0 and 3.0. The

penetration of the lower secondary jet through the cross-flow layer for φ=3.0 can be

understood by observing the magnitude of the peak velocity. The shifting of the peak

velocity farther from the wall clearly indicates the spreading of the lower secondary jet

and less deviation from the geometric axis of the burner. For φ=1.0, at y/De=2.0, the

105


recirculation zone extended up to x/De=0.25 whereas for φ=3.0 it was not as prominent

as for φ=1.0.


0

0.2

0.4

0.6

0.8

1

1.2

1.4

0 1 2 3 4 5x/De

Ure

s/U

ce



0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5x/De

Ure

s/U

ce

6


Figure 5.15(a): Velocity profiles in the centre plane of the lower secondary jet for φ=1.0


0

0.5

1

1.5

2

2.5

3

0 1 2 3 4 5x/De

Ure

s/U

ce



0

0.2

0.4

0.6

0.8

1

1.2

1.4

0 2 4 6 8 10 12 14x/De

Ure

s/U

cey/De=3y/De=4y/De=5y/De=9

Figure 5.15(b): Velocity profiles in the centre plane of the lower secondary jet for φ=3.0

The distribution of turbulent stresses for φ=1.0 in the xy plane through the centre of the

primary jet are shown in figures 5.16(a-c), plotted as the urms and vrms velocities

normalized to Uce and the uv stress normalized to the square of Uce for consistency with

previous results.

u rms for φ= 1.0

0

0.05

0.1

0.15

0.2

0.25

0.3

0 1 2 3 4 5

u rms for φ= 1.0

0

0.05

0.1

0.15

0.2

0.25

0.3

0 1 2 3 4 5 6 7 8x/De

u rm

s/U

ce


x/De

u rm

s/U

ce


Figure 5.16(a): urms in the centre plane of the primary jet for φ=1.0

106


v rms for φ= 1.0

0

0.050.1

0.150.2

0.250.3

0.350.4

0 1 2 3 4 5x/De

v rm

s/U

cey/De=0y/De=0.5y/De=1y/De=2

v rms for φ= 1.0

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0 1 2 3 4 5 6 7 8x/De

v rm

s/U

ce


Figure 5.16(b): vrms in the centre plane of the primary jet for φ=1.0

uv stress for φ= 1.0

-0.025

-0.02

-0.015

-0.01

-0.0050

0.005

0.01

0.015

0.02

0 1 2 3 4 5x/De

uv/U

ce2



-0.03

-0.02

-0.01

0

0.01

0.02

0.03

0 1 2 3 4 5 6 7 8x/De

uv/U

ce2


Figure 5.16(c): uv stress in the centre plane of the primary jet for φ=1.0

In figure 5.16(a), at y/De=0, the peak value of urms (0.27) occurred at x/De=0.3. The

position of the peak values shifted farther from the wall at y/De=0.5 and 1.0 and

occurred at x/De=0.6 and 0.95 respectively. The peak values at these positions (y/De=0,

0.5 and 1.0) occurred because of the generation of turbulence. This might happen due to

the interaction between the primary jet and the cross-flow jet. At y/De=2.0, near to the

wall (x/De=0.3) urms was high because of the high velocity gradient (figure 5.13(a)). In

this region there was a reverse flow. Further downstream (y/De=3.0, 4.0 and 5.0) the

distribution of urms was similar and two peaks occurred in the profile. The first peak was

at x/De=0.4 and the second peak was at around x/De=3.0. High velocity gradient was

responsible (figure 5.13(a)) for the first peak. The second peak occurred due to

interaction of shear layer between the primary jet and the surrounding fluid.

The generation of turbulence near to the wall (x/De=0.3) at y/De=2.0 can be well

understood with the help of figures 5.16(b) and 5.16(c). In figure 5.16(b), at y/De=2.0,

vrms was very high (0.36) near to the wall. As a result uv stress was relatively high at

107


this position (figure 5.16(c)). This high uv stress as well as high velocity gradient was

responsible for the generation of turbulence.

Figure 5.17(a) shows the urms in the centre plane of the primary jet for φ=3.0. The

distribution of urms was completely different than for φ=1.0.

u rms for φ= 3.0

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0 1 2 3 4 5 6 7 8x/De

u rm

s/U

ce

y/De=0y/De=2y/De=3

u rms for φ= 3.0

0

0.050.1

0.15

0.20.25

0.3

0.350.4

0 2 4 6 8 10 12 14x/De

u rm

s/U

ce

y/De=4y/De=5y/De=9

Figure 5.17(a): urms in the centre plane of the primary jet for φ=3.0

At y/De=0, the peak value occurred at the exit of the recess (x/De=0). At y/De=2.0 and

3.0, urms were high near to the wall because of the generation of turbulence due to high

velocity gradient. Further downstream (y/De=4.0, 5.0 and 9.0), the peak values of urms

shifted farther from the wall. The region of highest peak values was the shear layer

between the primary jet and the surrounding fluid. This can be understood more clearly

by observing figure 5.17(c) where the uv stresses were highest at these positions.

v rms for φ= 3.0

0

0.1

0.2

0.3

0.4

0.5

0 1 2 3 4 5 6 7 8x/De

v rm

s/U

ce

y/De=0y/De=2y/De=3

v rms for φ= 3.0

0

0.050.1

0.15

0.2

0.25

0.30.35

0.4

0 2 4 6 8 10 12 14x/De

v rm

s/U

ce

y/De=4y/De=5y/De=9

Figure 5.17(b): vrms in the centre plane of the primary jet for φ=3.0

108



-0.03

-0.02-0.01

0

0.010.02

0.03

0.040.05

0 1 2 3 4 5 6 7 8x/De

uv/U

ce2

y/De=0y/De=2y/De=3


-0.03

-0.02

-0.01

0

0.01

0.02

0.03

0.04

0 2 4 6 8 10 12 14x/De

uv/U

ce2

y/De=4y/De=5y/De=9

Figure 5.17(c): uv stress in the centre plane of the primary jet for φ=3.0

Comparison of the turbulent velocity fluctuations and uv shear stress in the centre plane

of lower secondary jet for φ=1.0 and 3.0 are shown in figures 5.18 and 5.19. Due to the

increase in secondary jet velocity the distribution of turbulent velocity fluctuations and

uv stresses were quite different and the peak values of urms, vrms and uv stresses were

higher for φ=3.0. For example in figure 5.18(a), φ=1.0, the magnitude of the highest

peak value of urms was 0.285 and occurred near the wall (x/De=0.31) at y/De=0 while

for φ=3.0, figure 5.19(a), it was 0.56 and occurred away from the wall at y/De=2.0.

u rms for φ= 1.0

0

0.05

0.1

0.15

0.2

0.25

0.3

0 1 2 3 4 5x/De

u rm

s/U

ce


u rms for φ= 1.0

0

0.05

0.1

0.15

0.2

0.25

0.3

0 1 2 3 4 5 6 7 8x/De

u rm

s/U

ce


Figure 5.18(a): urms in the centre plane of the lower secondary jet for φ=1.0

v rms for φ= 1.0

00.05

0.10.150.2

0.250.3

0.350.4

0.45

0 1 2 3 4 5x/De

v rm

s/U

ce


v rms for φ= 1.0

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0 1 2 3 4 5 6 7 8x/De

v rm

s/U

ce


Figure 5.18(b): vrms in the centre plane of the lower secondary jet for φ=1.0

109



-0.03

-0.02

-0.01

0

0.01

0.02

0.03

0 1 2 3 4 5x/De

uv/U

ce2



-0.03

-0.02

-0.01

0

0.01

0.02

0.03

0 1 2 3 4 5 6 7 8x/De

uv/U

ce2


Figure 5.18(c): uv stress in the centre plane of the lower secondary jet for φ=1.0

u rms for φ=3.0

0

0.1

0.2

0.3

0.4

0.5

0.6

0 1 2 3 4 5 6x/De

u rm

s/U

ce


u rms for φ=3.0

0

0.1

0.2

0.3

0.4

0.5

0.6

0 2 4 6 8 10 12 14x/De

u rm

s/U

ce


Figure 5.19(a): urms in the centre plane of the lower secondary jet for φ=3.0

v rms for φ=3.0

0

0.1

0.2

0.3

0.4

0.5

0.6

0 1 2 3 4 5 6

v rms for φ=3.0

0

0.1

0.2

0.3

0.4

0.5

0.6

0 2 4 6 8 10 12 14x/De

v rm

s/U

ce


x/De

v rm

s/U

ce


Figure 5.19(b): vrms in the centre plane of the lower secondary jet for φ=3.0

uv stress for φ=3.0

-0.06

-0.04

-0.02

0

0.02

0.04

0.06

0 1 2 3 4 5

uv stress for φ=3.0

-0.06

-0.04

-0.02

0

0.02

0.04

0 2 4 6 8 10 12 14x/De

uv/U

ce2


x/De

uv/U

ce2


Figure 5.19(c): uv stress in the centre plane of the lower secondary jet for φ=3.0

110



An experimental investigation of the effect of jet velocity ratio for rectangular slot-

burners (geometry B and D) in the presence of cross-flow has been presented in this

chapter. To obtain a rotational flow field similar to the tangentially-fired furnace one

side of the cross-flow jet was attached to the wall. The experiment was conducted for jet

velocity ratios of 1.0 and 3.0. The LDA technique was used to measure the mean

velocities and rms.

From the experimental results it was concluded that cross-flow had a significant effect

in developing the near field region for both geometry B and D. In the presence of cross-

flow both the primary jet and the lower secondary jet deviated from their geometric axes

towards the wall. The degree of deviation was dependent on jet velocity ratio.

Geometry B consisted of three nozzles (one primary and two secondary) at an angle of

60ο to the wall. For φ=1.0, the primary jet deviated completely from the geometric axis

of the burner and remained within the cross-flow. The deflection of the lower secondary

jet was slightly greater than the primary jet. For φ=3.0, the primary jet penetrated

through the cross-flow layer. The penetration was more in the centre plane of the lower

secondary jet due to higher momentum of the secondary jets. There were two peaks for

urms at y/De=0 both for jet velocity ratios of 1.0 and 3.0. At y/De=1, near to the wall

(x/De=0.3), urms was high because of the generation of turbulence due to high velocity

gradient.

In case of geometry D, the nozzles were recessed into the wall. The flow pattern inside

the recess was very complex and greatly influenced the flow outside the recess. For

φ=1.0, like geometry B, the jets diverged completely from the geometric axis of the

burner and lie against the wall. The degree of deflection of the lower secondary jet was

slightly more than the primary jet. There was a recirculation zone very near to the wall

started at y/De=2.0 for both primary and lower secondary jet but the region of

recirculation in the centre plane of the lower secondary jet was smaller and finished

before y/De=3.0. For φ=3.0, the flow pattern in the centre plane of the primary jet was

different than geometry B. Instead of penetrating through the cross-flow layer primary

jet was almost parallel with the cross-flow jet and lie close against the wall. After

111


exiting from the nozzle, the primary jet diffused outwards and mixed with the higher

momentum secondary jets in the recess and came out with the secondary jets. It was

mainly the cross-flow jet that appeared on the centre plane of the primary jet. The lower

secondary jet penetrated through the cross-flow layer due to higher momentum of the

secondary jet.

It was difficult to explain the finer details of the jet/cross-flow interaction with these

limited experimental results. For this reason numerical simulations were performed for

geometries B and D in the presence of cross-flow. Experimental results of this chapter

were used for validation purposes and then more detailed descriptions of the flow were

revealed with the help of numerical results. The next chapter will discuss the numerical

results for the same rectangular slot-burners in the presence of cross-flow.

112

NOTE


Chapter-6 Numerical investigation of rectangular

slot-burners with cross-flow

Chapter 6-Numerical investigation of rectangular slot- burners with cross-flow

6. Numerical investigation of rectangular slot -burners with cross-flow

This chapter presents the computational fluid dynamics (CFD) investigations of fluid

flow in the presence of cross-flow for geometries B and D. With the experimental

results presented in the previous chapter, it was difficult to elucidate all mechanisms by

which the jets deviated from the geometric axis. For this reason, numerical investigation

of geometries B and D were performed, using the commercial CFD software CFX-5.

Steady state Navier-Stokes equations were solved with SST turbulence model. The

numerical results were first validated against the experimental results in the preceding

chapter and then visualization of the developing flow field was used to reveal the finer

details of the cross-flow/burner jet interaction.

6.1. Grid independence test

A grid independence test was performed on geometry B and for φ=3.0. The grid

resolution was based on the number of cells used to model the nozzles. Three grid

refinements were performed, the first setting 4x4 cells in the primary nozzle cross-

section and 4x2 cells in the secondary nozzle cross-section. A geometric progression

was used to expand the grid away from the jet nozzle to reduce the grid resolution in

regions of small velocity gradients. Successive grid refinements involved increasing the

number of cells in the nozzle cross-sections to 8x8 in the primary and 8x4 in the

secondary and finally 16x16 and 16x8. The same principles were applied to the grid

expansion for each grid. The results of the grid independence test are presented in figure

6.1 comparing the velocity profiles in the xy plane one hydraulic diameter (y/De=1)

downstream of the jet. The plane has been considered through the centre of the primary

jet. Velocities were normalized to the centreline exit velocity for velocity ratio of 1.0 of

the primary jet.

The peak velocity for the 4x4 cells was 1.13 and occurred at x/De=0.94. The peak

velocities for 8x8 cells and 16x16 cells were 1.11 and 1.13 respectively and occurred

(x/De=0.63) before the 4x4 cells. For x/De>2, successive grid refinements also

produced a less diffusive jet, the most diffuse being the 4x4 cells, followed by the 8x8

cells and 16x16. Both 8x8 and 16x16 cells gave a very close prediction having the peak

values at the same position. Although the jet with 16x16 cells was less diffusive in the

114


region 2≤x/De≤3.5, the difference between 8x8 and 16x16 cells was small. So to reduce

CPU time, 8x8 cells was used for the numerical simulation. The total number of cells

used in the simulation was 600,000.

Velociry Distribution for velocity ratio of 3 at y/De=1

0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4x/De

Umea

n/Uc

e

5

4x4 cells

8x8 cells

16x16 cells

Figure 6.1: Grid independence test for geometry B

6.2. Geometry B

6.2.1. Secondary to primary jet velocity ratio of 1.0

6.2.1.1. Validation of numerical results

This section presents the validation of the CFD simulations against the experimental

data. Comparison of the resultant velocities for φ=1.0 on the centre plane of the primary

jet, lower base region and lower secondary jet are shown in figures 6.2, 6.3 and 6.4

respectively. The SST model was used for turbulence modelling, which has been

discussed in chapter 3.

Figure 6.2 shows a comparison of the resultant velocity in a horizontal plane. The plane

was at the centre of the primary jet. The figure shows that generally the velocities of the

simulated jet are qualitatively similar to that observed in the measured jet, however,

there were some discrepancies. The peak velocities were predicted at the correct

locations and the magnitudes of these peaks also agreed well, although there was a

tendency to slightly over predict them. The most noticeable differences occurred near

the wall where the calculated velocities where generally much higher than the

experiments at downstream locations between y/De=3.0 to5.0. At y/De=9.0 the shape of

the profile was reasonably well predicted. The experimental profile was quite flat

indicating the self-similarity region, whereas the simulation still showed a significant

115


central peak, although it was quite broad, indicating the jet in the simulation was also

approaching self-similarity.

Figure 6.3 shows the comparison of resultant velocities at the centre plane of the lower

base region between the primary and lower secondary jets for φ=1.0. The numerical

results matched well with the measurements up to y/De=1.0. The velocities were then

over predicted in the near wall region at y/De=2.0. The simulated jet had a higher peak

value at y/De=3.0 and this trend continued up to y/De=9.0. The comparison at the centre

plane of the lower secondary jet is shown in figure 6.4. The predictions were excellent

for φ=1.0, only the near wall region of the profiles showed any real discrepancy, most

noticeable at y/De=1.0 and 9.0. Therefore, the numerical results are considered to be

reliable enough to explain the physics of near field region and mechanism of jets

development in the next section, where the reasons for the variation near to the wall are

also discussed.

116


Velocity Distribution at y/De=0

0

0.2

0.4

0.6

0.8

1

0 1 2 3 4 5x/De

Ure

s/U

ceUres (Expt)

Ures (SST M odel)

Velocity Distribution at y/De=0.5

0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

0 1 2 3 4 5x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

0 1 2 3 4 5x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

0 1 2 3 4 5 6x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

0 1 2 3 4 5 6x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

0 1 2 3 4 5x/De

Ure

s/U

ce

6

Ures (Expt)

Ures (SST M odel)

Figure 6.2: Comparison of resultant velocities at the centre plane of the primary jet for

φ=1.0

117



0

0.2

0.4

0.6

0.8

1

0 1 2 3 4 5x/De

Ure

s/U

ceUres (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

0 1 2 3 4 5x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

0 1 2 3 4 5x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

0 1 2 3 4 5 6x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

0 1 2 3 4 5 6x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

0 1 2 3 4 5 6x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

0 1 2 3 4 5 6


0

0.2

0.4

0.6

0.8

1

0 1 2 3 4 5x/De

Ure

s/U

ce

6

Ures (Expt)

Ures (SST M odel)

x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)

Figure 6.3: Comparison of resultant velocities at the centre plane of the lower base

region for φ=1.0

118



0

0.2

0.4

0.6

0.8

1

0 1 2 3 4 5x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

0 1 2 3 4 5x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

0 1 2 3 4 5x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

0 1 2 3 4 5x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

0 1 2 3 4 5 6x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

0 1 2 3 4 5x/De

Ure

s/U

ce

6


0

0.2

0.4

0.6

0.8

1

0 1 2 3 4 5x/De

Ure

s/U

ce

6

Ures (Expt)

Ures (SST M odel)

Ures (Expt)

Ures (SST M odel)

Figure 6.4: Comparison of resultant velocities at the centre plane of the lower secondary

jet for φ=1.0

119


6.2.1.2. Flow field prediction

The pressure distribution together with streamlines in the xy plane through the centre of

the primary jet are shown in figure 6.5(a) for φ=1.0. The major contribution of bending

the primary jet came from the cross-flow jet. Just after exiting from the primary duct,

the jets mixed with the cross-flow jet and became one jet. Upon impact with the primary

jet, the cross-flow jet exerted an external force and as a result the primary jet deviated

from the geometric axis towards the wall. Although it was the cross-flow, which was

mainly responsible for the bending of the primary jet, the contribution came from the

cross-stream pressure difference was also significant which is described in the next

section.

(a) (b)

Short-side

Figure 6.5(a-b): Pressure distribution in the xy plane through the centre of the primary

jet (a) and lower secondary jet (b) for φ=1.0.

As the jet exited the duct, it did so earlier on the short side. Due to the expansion of the

jet kinetic energy was converted to pressure. As a result there was a high-pressure zone

on the short side. This high-pressure zone extended into the domain up to a certain

distance and then gradually dropped just above the atmospheric pressure. On the long

side, stream wise pressure drop occurred inside the duct before the jet entered into the

domain. This low-pressure zone on the long side of the nozzle was connected to a large

low-pressure zone just outside the duct along the wall of the domain. As a result a cross-

stream pressure gradient occurred across the nozzle exit. This cross-stream pressure

difference can cause bending of the streamlines towards the solid boundaries (Massey,

1998); in this case it bent the primary streamlines from the centreline towards the long-

side wall. Figure 6.5(b) shows the pressure distribution in the xy plane through the

120


centre of the lower secondary jet. The pressure gradient was slightly higher as compared

to the primary jet causing the secondary jet to bent more.

Figure 6.6 shows the pressure distribution and streamlines in the yz plane at a distance

x/De=0.31 from the wall. After exiting from the nozzles, primary and secondary jets

experienced a low-pressure region in front of them (figure 6.6(a)). The primary

streamlines passed over and under the low-pressure region and mixed with the cross-

flow jet. The upper and lower secondary streamlines also passed underneath and over

the low pressure in front of them and ultimately remained with in the cross-flow.

(a) (b)

Figure 6.6: Pressure distribution (a) and streamlines (b) in the yz plane at a distance

x/De=0.31 from the wall

The phenomenon for the cross-flow jet was different. There was a high-pressure zone

(figure 6.6(a)) just before the cross-flow jet mixed with the primary and secondary jets

and then a sudden pressure drop in front of the ducts. As a result a large-pressure

gradient formed and the cross-flow jet exerted a force on primary and secondary jets

causing the primary and secondary jets bent towards the wall. The cross-flow at the

centre plane of the primary jet passed over and under the low-pressure zone, mixed with

the primary jet and advanced further downstream. Most of the cross-flow streamlines at

the centre plane of the upper secondary jet passed over the upper secondary jet and

moved in the downward direction towards the centre plane of the primary jet. On the

other hand streamlines at the centre plane of the lower secondary jet passed underneath

the lower secondary jet, advanced in the upward direction, and penetrated the centre

plane of the primary jet. At this near wall location in the experiment the cross-flow

121


streamlines did not penetrate the primary plane. As a result the velocity for numerical

investigation near to the wall was higher than expected, as there was no evidence of this

in the measurements. This was the reason for the velocities near the wall being higher in

the numerical results. The difference in velocities near the wall extended upto y/De=5.0

at the centre plane of the primary jet but reduced to y/De=1.0 at the centre plane of the

lower secondary jet. This is clear from figure 6.6(b) where the portion of the cross-flow

streamlines responsible for the discrepancy extended only upto y/De=1.0. So it was

neither the primary jet nor the secondary jets but a portion of the cross-flow jet that

came underneath the lower secondary jet and penetrated at the centre plane of the

primary jet was responsible for the difference in velocity near to the wall.

Figures 6.7(a-b) compares the centreline velocity decay at the centre plane of the

simulated primary and lower secondary jet with the measurements for φ=1.0. The

centreline was taken as the jet centreline. The velocities have been normalized to the

centreline velocity at the nozzle exit of the primary jet. The centreline decay of the

simulated jet matched reasonably well with the experiments. For the primary jet, the

agreement was good up to y/De=2.0, the simulated jet then decayed slower than the

experiments. The measured velocity at y/De=9.0 was 0.56 while the predicted value was

0.65.

Centreline Velocity Decay of Primary Jet

0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5 6 7 8 9 10

y/De

Um

ean/

Uce

Experiment

SST Model

Figure 6.7(a): Centreline velocity decay of Primary jet for φ=1.0

122


Centreline Velocity Decay of lower Secondary Jet

0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5 6 7 8 9 10

y/De

Um

ean/

Uce

Experiment

SST Model

Figure 6.7(b): Centreline velocity decay of lower Secondary jet for φ=1.0

Centreline decay pattern for lower secondary jet was similar to the primary jet upto

y/De=5.0 in that it under predicted the decay rate. After that the simulated jet decayed

rapidly and matched with the measurement at y/De=9.0. For both primary jet and lower

secondary jet, the velocity increased at y/De=0.5 because of the mixing with the cross-

flow jet.

Figures 6.8(a-b) shows the velocity vectors at the centre plane of the primary jet and

lower secondary jet for φ=1.0. Entrainment of the surrounding fluid into the jet was

evident at the interface between the cross-flow and the surrounding fluid.

(a) (b)

Figure 6.8 (a-b): Velocity vectors at the centre of the primary jet (a) and the lower

secondary jet (b) for φ=1.0

Figures 6.9 shows the velocity vectors in the yz plane at x/De=0.078 for φ=1.0. The

plane was very near to the wall. An interesting feature was revealed at this plane, which

was the classical formation of the counter rotating vortex pair. This twin vortex system

123


formed behind the initial region of the primary jet core. It started at this plane and

extended upto x/De=0.468 as shown in figure 6.9. After that the double vortex

disappeared due to presence of the cross-flow jet. The primary jet mixed with the cross-

flow and advanced downstream as a single jet.

(a) (b)

x/De=0.078 x/De=0.156 Figure 6.9(a-b): Velocity vectors in the yz plane at x/De=0.078 and 0.156 showing the

formation of twin vortex for φ=1.0.

(c) (d)

x/De=0.234 x/De=0.312 Figure 6.9(c-d): Velocity vectors in the yz plane at x/De=0.234 and 0.312 showing the


124


(e) (f)

x/De=0.390 x/De=0.468 Figure 6.9(e-f): Velocity vectors in the yz plane at x/De=0.390 and 0.468 showing the




Comparison of the resultant velocities at the centre plane of the primary jet, lower base

region and lower secondary jet are presented in figures 6.10, 6.11 and 6.12 respectively

for φ=3.0. All velocities have been normalized to the centreline velocity at the exit of

the primary for φ=1.0. The measurement locations were the same as they were for

φ=1.0.

In figures 6.10, the general shape of the velocity profiles was well predicted by the

numerical model. The peaks velocities were predicted at the correct locations up to

y/De=5.0 but the magnitudes of the peak velocities were slightly over predicted and the

most noticeable differences occurred at y/De=0 and 0.5. The predicted velocities near

the wall were higher than the experiments between y/De=1.0 to 3.0 and lower at

downstream locations for y/De=5.0 and 9.0. At y/De=9.0 both the magnitude and the

location of the predicted peak velocity were shifted from the experiments and occurred

farther from the wall.

125



0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5

x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5x/De

Ure

s/U

ce

6

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5 6x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5 6

x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

0 1 2 3 4 5 6 7 8

x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

0 1 2 3 4 5 6 7 8

x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

0 2 4 6 8 10x/De

Ure

s/U

ce

12

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

0 2 4 6 8 10 12 14x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)

Figure 6.10: Comparison of resultant velocities at the centre of the primary jet for

φ=3.0.

126



0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5x/De

Ure

s/U

ceUres (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5 6x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

0 1 2 3 4 5 6 7 8x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

0 1 2 3 4 5 6 7 8 9x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

0 2 4 6 8 10 12x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

0 2 4 6 8 10 12 14x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)

Figure 6.11: Comparison of resultant velocities at the centre of lower base region for

φ=3.0.

127



0

0.5

1

1.5

2

2.5

3

0 1 2 3 4 5x/De

Ure

s/U

ceUres (Expt)

Ures (SST M odel)


0

0.5

1

1.5

2

2.5

3

0 1 2 3 4 5x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.5

1

1.5

2

2.5

3

0 1 2 3 4 5x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.5

1

1.5

2

2.5

0 1 2 3 4 5 6x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.5

1

1.5

2

2.5

0 1 2 3 4 5 6 7 8x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.5

1

1.5

2

2.5

0 1 2 3 4 5 6 7 8

x/De

Ure

s/U

ce

Ures (Expt)Ures (SST M odel)


0

0.5

1

1.5

0 2 4 6 8 10 12x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

1.2

0 2 4 6 8 10 12 14

x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)

Figure 6.12: Comparison of resultant velocities at the centre of lower secondary jet for

φ=3.0.

128


Figures 6.11 shows the comparison of resultant velocities at the centre plane of the

lower base region for φ=3.0. At y/De=0 and 0.5, the peak values of the predicted jet

matched well with the experiment but did not occur at x/De=0 and the boundaries were

thinner than the experiments. At y/De=1.0 and 2.0, the velocity was over predicted near

the wall. The predicted velocity profile was fuller between 0.31≤x/De≤2.5 at y/De=2.0.

Further downstream (y/De=3.0 to 5.0), the velocities near the wall were under predicted

but the peak values were higher and occurred slightly before the experimental values. At

y/De=9.0, the numerical simulation failed to predict the velocity profile in the region

between 8≤ x/De≤12.

Figures 6.12 show the comparison at the centre plane of the lower secondary jet for

φ=3.0. Generally the results were over predicted at this plane. At y/De=0, 0.5 and 1.0,

both the peak values and the values near wall region were over predicted. The values

matched well near to the wall at y/De=2.0 and then the results were under predicted

upto y/De=5.0. At this location (y/De=2.0) the peak values both for numerical results

and measurements occurred at the same position but magnitude of the peak value of the

predicted jet was higher than the experiments. This trend continued upto y/De=5.0

indicating that the predicted jet penetrated more than the experimental jet at this plane.

At y/De=3.0, the peak value of the predicted jet occurred after the experimental peak

value and the distance increased further downstream.


The streamlines and pressure distribution on the centre plane of the primary jet and

lower secondary jet for φ=3.0 are shown in figures 6.13(a) and 6.13(b) respectively.

Figure 6.13(a) shows a high-pressure zone on the short side, which extended into the

domain before reducing to near atmospheric pressure. On the long side, stream wise

pressure drop occurred inside the duct and just before the jet entered into the domain it

became negative. This low-pressure zone was connected to a low-pressure region

outside the duct along the wall of the domain. As a result cross-stream pressure

difference occurred across the nozzle exit causing the jet bent from the geometric axis

towards the long side.

129


(a) (b)

Short side

Figure 6.13(a-b): Streamlines and pressure distribution in the xy plane through the

centre of the primary jet (a) and lower secondary jet (b).

The deviation of the primary jet from the geometric axis for φ=3.0 was less than that

observed for φ=1.0. After exiting from the duct, primary jet came into contact with the

secondary jets, mixed with them, and the momentum of the primary jet was increased

which was enough to penetrate the cross-flow layer. The penetration and less deviation

from the geometric axis can be understood more clearly by observing the lower

secondary jet (figure 6.5(b) and 6.13(b)). For φ=3.0, at the centre plane of the lower

secondary jet, figure 6.13(b), the shape of the cross-stream pressure difference was

different than the primary jet (figure 6.13(a)). As a result, for primary jet the streamlines

started bending from the middle of the nozzle whereas for lower secondary jet they bent

after a short distance from the short side of the nozzle.

Figure 6.14 shows the pressure distribution and the streamlines in the yz plane at a

distance x/De=0.31 from the wall for φ=3.0. There were low-pressure zones in front of

the upper and lower secondary nozzles and a high-pressure zone upstream of the

primary and secondary nozzles. The cross-flow jet coming from the lower portion of the

cross-flow nozzle passed underneath the low-pressure zone in front of the lower

secondary nozzle and being sucked by the higher momentum lower secondary jet. On

the other hand most of the cross-flow jet that came from the upper portion of the cross-

flow nozzle passed over the low-pressure zone in front of the upper secondary nozzle

and being sucked by the upper secondary jet. A portion of the cross-flow came over the

130


low-pressure zone of the upper secondary nozzle, moved in the downward direction,

and penetrated the plane located at the centre of the primary jet (figure 6.14 (b)).

(a) (b)

Figure 6.14: Pressure distribution (a) and streamlines (b) in the yz plane at a distance

x/De=0.31 from the wall.

This portion of the cross-flow jet was responsible for the higher velocity near to the

wall. For φ=1.0, a portion of the cross-flow jet was also responsible for the difference

between the numerical and predicted results near the wall but that came from the lower

portion of the cross-flow nozzle whereas for φ=3.0, they came from the upper portion of

the cross-flow nozzle.

Figures 6.15(a-b) shows the velocity vectors at the centre plane of the primary jet and

lower secondary jet for φ=3.0. As for φ=1.0, entrainment of the surrounding fluid into

the jet was evident at the interface of the cross-flow jet and the surrounding fluid but the

vectors of the entrainment field were more complex than that for φ=1.0. Upstream of the

burner the flow was parallel to the geometric axis whereas downstream of the burner the

vector field was normal to the cross-flow jet.

131


(a) (b)

Figure 6.15(a-b): Velocity vectors at the centre plane of the primary jet (a) and the

lower secondary jet (b) for φ=3.0.

Another contribution to entrainment was the creation of low-pressure regions at the

bottom of the lower secondary nozzle and the top of the upper secondary nozzle as

shown in figures 6.16. Figures 6.16(a-b) show the pressure distribution and the velocity

vectors in the xz plane at the centre of the geometric axis. Fluid entrained into the jet

must be replaced by fluid from the surroundings. Because of the low-pressure region,

fluid from the surroundings entered easily into the jet through these regions.

Figure 6.16(a-b): Pressure distribution (a) and the velocity vector (b) in the xz plane at

the centre of the geometric axis for φ=3.0.

(a) (b)

Figure 6.17 (a-b) compares the jet centreline velocity decay of the simulated jet with the

experiments for φ=3.0. Both for primary and secondary jet, the numerical results under

predicted the decay rate. In figure 6.17(a), the velocity of the measured jet increased

132


upto y/De=1.0 whereas for predicted jet the peak value occurred at y/De=0.5. The

measured velocity at y/De=9 was 0.57 while the predicted value was 0.68.

Centreline Velocity Decay for Primary Jet

00.2

0.40.60.8

1

1.21.4

0 1 2 3 4 5 6 7 8 9 10y/De

Um

ean/

Uce

Experiment

SST Model

Figure 6.17(a): Centreline velocity decay of primary jet for φ=3.0

Figure 6.17(b) shows the comparison of the centreline velocity decay of lower

secondary jet for φ=3.0. The experimental jet decayed more rapidly than the simulated

jet between 1.0≤ y/De≤ 2.0. Beyond that the decay rate reduced while the simulated jet

showed a constant rate of decay. After y/De=5.0, the simulated jet decayed more rapidly

than the measurement but not enough to match the measured velocity at y/De=9.0.

Centreline Velocity Decay of lower Secondary Jet

0

0.5

1

1.5

2

2.5

3

0 1 2 3 4 5 6 7 8 9 10y/De

Um

ean/

Uce

Experiment

SST Model

Figure 6.17(b): Centreline velocity decay of lower secondary jet for φ=3.0

Figure 6.18 shows the velocity vectors in the yz plane at x/De=0.156, near to the wall.

The classical formation of twin vortices was more prominent for this case than for

φ=1.0. The twin vortices formed at the same position as for φ=1.0 but for φ=1.0 this

vortex pair extended up to x/De=0.468 while in this case extended up to x/De=0.938.

Yan and Perry (1994) investigated the flow field by visualization of this rectangular

slot-burner for the same cross-flow configuration by taking the first plane at x/De=0.781

but did not observe any twin vortex system. For φ=1.0, the twin vortex disappeared

133


before x/De=0.781 and for φ=3.0, the magnitude of the vortex pair was small. Hence it

is not surprising that they did not observe the vortex pair.

(a) (b)

x/De=0.156 x/De=0.3125

Figure 6.18(a-b): Velocity vectors in the yz plane at x/De=0.156 and 0.3125 showing

the formation of twin vortex for φ=3.0.

(c) (d)

x/De=0.468 x/De=0.625

Figure 6.18(c-d): Velocity vectors in the yz plane at x/De=0.468 and 0.625 showing the


134


(e) (f)

x/De=0.781 x/De=0.938

Figure 6.18(e-f): Velocity vectors in the yz plane at x/De=0.781 and 0.938 showing the


6.3. Geometry D



Comparison of the resultant velocities at the centre plane of the primary jet, lower base

region and lower secondary jet are presented in figures 6.19, 6.20 and 6.21. All

velocities have been normalized with the centreline velocity at the exit of the recess

(x/De=0, y/De=0, z/De=0) for φ=1.0. The measurement locations were the same as they

were for geometry B.

Overall the predicted velocity profiles matched well with the measurements at the centre

plane of the primary jet. At y/De=0, 0.5 and 1.0, peak velocity as well as velocities near

to the wall was very well predicted. However, the boundaries of the simulated jet were

thinner than the measurement and this trend continued up to y/De=9.0. At y/De=2.0,

near to the wall, the recirculation zone was very well captured by the simulated jet and

the peak velocity was slightly under predicted but occurred at the same position. At

y/De=3.0, velocities were over predicted near to the wall and this trend continued up to

y/De=5.0. At y/De=5.0 and 9.0 although the velocity profiles in the region between

0.8≤x/De≤1.8 were under predicted the peak velocity matched well with the

measurements.

135



0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5x/De

Ure

s/U

ceUres (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5 6x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5 6x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5 6 7 8x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


φ=1.0

136



0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

1.2

1.4

0 1 2 3 4 5x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5 6x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5 6x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5 6 7 8x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)

Figure 6.20: Comparison of resultant velocities at the centre plane of the lower base

region for φ=1.0

137



0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5


0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)

x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

1.2

1.4

0 1 2 3 4 5


0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)

x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5 6x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5 6x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5 6 7 8x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)

Figure 6.21: Comparison of resultant velocities at the centre plane of the lower

secondary jet for φ=1.0

138


Figure 6.20 shows the comparison of resultant velocities at the centre plane of the lower

base region for φ=1.0. At y/De=0 and 0.5, the velocities were under predicted near to

the wall. The peaks were predicted well but as in figure 6.19 the boundaries of the

predicted profiles were thinner than the measured profiles. The predicted jet had a

higher velocities near to the wall at y/De=2.0 and this trend continued up to y/De=5.0.

At y/De=9.0, the peak velocity of the predicted jet was higher than the measurement

indicated.

The comparison of the resultant velocities at the centre plane of the lower secondary jet

is shown in figure 6.21. As at the centre of the lower base region the velocity were over

predicted near to the wall for y/De=2.0-4.0. At y/De=5.0 the peak value of the

prediction occurred earlier than the measured profile and at y/De=9.0 the agreement was

also good. Despite of some discrepancy near to the wall at the centre plane of the lower

base region and lower secondary jet the numerical model predicted the jets reasonably

well. Generally the simulations appear to match the physical model, however, there are

significant discrepancies near the wall. The reasons for this can be understood from

flow visualization.


The recess and the multiple-jet system is three-dimensional, however, when viewed on

the xy plane of the primary jet (figure 6.22(a)) the diverging recess is reminiscent of a

two dimensional diffuser, and description of the flow in these terms aids an

understanding of the three dimensional flow field in general.

The idea of an efficient diffuser, according to Massey (1998), is to recover kinetic

energy, from the mean flow in the form of a rise in pressure, by expanding the duct

smoothly to prevent separation of the flow from the walls. An inefficient diffuser is one

in which the flow separates as it moves into diverging section, as this prevents adequate

expansion and deceleration of the flow to give the required pressure rise.

Figure 6.22(a) shows the velocity vectors at the centre plane of the primary jet. The

flow experienced a step-change at the side-walls as it moved from the duct to the recess.

On the long side there was a reverse flow at the beginning of the recess indicating the

existence of a separation bubble there. After the separation bubble the primary jet was

139


attached to the long wall. On the short side the primary jet was completely separated

from the wall. After exiting from the recess, the primary jet mixed with the cross-flow

jet and advanced further downstream as a single jet. Upon mixing, the cross-flow

exerted an external force on the primary jet resulting the primary jet deviated from the

geometric axis and lie against the wall. At y/De=2.0 there was a reverse flow near the

wall and this region extended up to y/De=3.0 consistent with the measurement (figure

5.22(a)). Entrainment of the surrounding fluid into the jet occurred at the interface

between the cross-flow and the surrounding fluid. As shown in figure 6.22(a), the

vectors of the entrainment field were normal to the geometric axis as for geometry B.

(a) (b)

Figure 6.22(a-b): Velocity vectors at the centre plane of the primary jet (a) and lower

secondary jet (b)

Figure 6.22(b) shows the velocity vectors at the centre plane of the lower secondary jet.

On the long side there was no separation and associated reverse flow at the beginning of

the recess but on the short side the reverse flow was more prominent than the centre

plane of the primary jet. No recirculation was observed near to the wall downstream of

the lower secondary jet, although the experiments showed the existence of a small

separation bubble at this near wall downstream location. The entrainment field was

similar to the centre plane of the primary jet i.e. normal to the geometric axis of the

burner.

The pressure distribution as well as streamlines in the xy plane through the centre of the

primary jet are shown in figure 6.23 (a) for φ=1.0. The pressure distribution inside the

recess was very complex. On the short side, the pressure gradually increased and at the

exit of the recess there was a high-pressure region. On the long side the pressure

140


increased up to the half way of the recess and then a stream wise pressure drop occurred

and the pressure became sub-atmospheric before the jet exit from the recess.

(a) (b)

Figure 6.23(a-b): Streamlines and pressure distribution in the xy plane through the

centre of the primary jet (a) and lower secondary jet (b).

The low-pressure zone extended into the domain along the near wall region. Two

factors were responsible for the deviation of the primary jet. The major reason was the

external force from the cross-flow, which added to the cross-stream pressure drop

across the recess. Figure 6.23(b) shows the pressure distribution and streamlines on the

centreline of the lower secondary jet. The lower secondary jet bent slightly more than

the primary jet.

Figure 6.24 (a) shows the pressure distribution in the yz plane at a distance x/De=0.31

from the wall for φ=1.0. There was a high-pressure zone just before the cross-flow jet

mixed with the primary and secondary followed by a sudden pressure drop in front of

the recess. As a result the cross-flow jet exerted a force on primary and secondary jet

causing them to deviate from the geometric axis and lie against the wall.

141


(a) (b)

Figure 6.24 (a-b): Pressure distribution (a) and streamlines (b) in the yz plane at a


Figure 6.24(b) shows that a portion of the lower secondary jet was responsible for

higher velocity prediction near to the wall at the centre plane of the lower base region

and lower secondary jet. The cross-flow at the centre plane of the lower base region

passed the low-pressure zone, mixed with the primary jet, then advanced further

downstream. A portion of the lower secondary jet passed underneath the low-pressure

zone in front of it and penetrated the centre plane at the lower base region. In the

measurement this portion of the secondary jet might not have penetrated the centre

plane of the lower base region. As a result the predicted mass flow near to the wall was

higher than the experiment, which ultimately increased the predicted velocity near to the

wall.

Figure 6.25(a) shows the velocity vectors in the yz plane at x/De=0 for φ=1.0, at the exit

of the recess. A counter rotating vortex pair occurred at this plane, forming behind the

initial region of the primary jet core as shown in figure 6.25(a). The twin vortex

extended up to x/De=0.390 and then disappeared.

142


(a) (b)

x/De=0 x/De=0.078

(c) (d)

x/De=0.234 x/De=0.390

Figure 6.25 (a-d): Velocity vectors in the yz plane showing the formation of twin vortex

for φ=1.0.



Comparison of the resultant velocities for φ=3.0 at the centre plane of the primary jet

and the lower secondary jet are presented in figure 6.26 and 6.27. In figure 6.26 the

peak velocities of the simulated jet occurred slightly before the measured jet for all

locations. The magnitudes of the predicted peak velocities were higher between

y/De=3.0 to 5.0. The simulated jet showed a tendency of over predicting the velocities

near the wall and the most noticeable difference occurred at y/De=3.0. At y/De=9.0, the

simulated jet completely failed to predict the velocity profile of the measured jet.

143



0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

1.2

1.4

0 1 2 3 4 5x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

1.2

1.4

0 1 2 3 4 5x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5x/DeU

res/

Uce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5 6x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5 6x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5 6 7 8x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

1.2

0 2 4 6 8 10 12 14x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


φ=3.0

144



0

0.5

1

1.5

2

2.5

3

0 1 2 3 4 5x/De

Ure

s/U

ceUres (Expt)

Ures (SST M odel)


0

0.5

1

1.5

2

2.5

0 1 2 3 4x/De

Ure

s/U

ce

5

Ures (Expt)

Ures (SST M odel)


0

0.5

1

1.5

2

2.5

0 1 2 3 4 5x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.5

1

1.5

2

0 1 2 3 4 5 6x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.5

1

1.5

0 1 2 3 4 5 6 7 8x/De

Ure

s/U

ce

Ures (Expt)Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

1.2

0 2 4 6 8 10 12x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)


0

0.2

0.4

0.6

0.8

1

1.2

0 2 4 6 8 10 12


0

0.2

0.4

0.6

0.8

1

1.2

0 2 4 6 8 10 12 14x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)

x/De

Ure

s/U

ce

Ures (Expt)

Ures (SST M odel)

Figure 6.27: Comparison of resultant velocities at the centre plane of the lower

secondary jet for φ=3.0

Figure 6.27 shows the comparison of the resultant velocities at the centre plane of the

lower secondary jet. The simulated jet matched well with the experimental jet up to

y/De=1.0. At y/De=2.0, the peak velocity of the simulated jet shifted farther from the

145


wall and occurred after the measured peak values. This trend continued up to y/De=9.0.

At these positions (y/De=2.0-9.0), the magnitude of the peak velocities of the simulated

jet was lower than the measurements. Overall the simulated jet matched reasonably well

at this plane except downstream of the jet (y/De=5.0 and 9.0) where the measured jet

deviated more than the simulated jet form the geometric axis of the burner.


Figure 6.28 (a) shows the velocity vectors in a vertical plane through the geometric

centre of both the primary and secondary jets. Upon exiting the duct and entering the

recess, the primary jet diffused outwards and moved towards the higher momentum

secondary jet. The top and bottom wall of the recess acted as an efficient diffuser for the

secondary jet, (Massey, 1998) which attached smoothly to the upper and lower wall

with no separation. The primary and secondary jets entrained fluid into the base regions

between the jets just after the exit from the duct because of the low pressure as shown in

figure 6.28(b). Entrainment of the surrounding fluid into the jet occurred at the bottom

of the lower secondary jet (figure 6.28(a)).

Figure 6.28(a): Velocity vectors through the geometric centre of both primary and

secondary jets

146


Figure 6.28(b): Pressure contour through the geometric centre of both primary and

secondary jets

Velocity vectors in the xy plane through the centre of the primary jet are shown in

figure 6.29(a). On the short side the primary jet was completely separated from the wall

and there was a large reverse flow associated with separation. On the long side there

was also a reverse flow at the beginning of the recess, which extended some distance

along the wall before the primary jet reattached again to the sidewall. The reverse flow

region on both long and short sides was more prominent than that for φ=1.0 (figure

6.22(a)). Entrainment into the primary jet at this level occurred at the interface between

the cross-flow jet and the surrounding fluid and the vector field was normal to the

geometric axis of the burner.

Figure 6.29(a): Velocity vectors at the centre plane of the primary jet for φ=3.0

147


Figure 6.29(b) shows the velocity vectors at the centre plane of the lower secondary jet.

The recess acted as an efficient diffuser for the lower secondary jet, with the flow

attaching to the sidewall to the same extent as it did to the bottom wall (figure 6.28 (a))

due to its increase in velocity. Velocity vectors in the yz plane at a distance x/De=0.31

from the wall are shown in figure 6.29(c) and they show that no vortex pair occurred for

these jets at φ=3.0.

Figure 6.29(b): Velocity vectors at the centre plane of the lower secondary jet for φ=3.0

Figure 6.29(c): Velocity vectors in the yz plane at x/De=0.31 from the wall for φ=3.0

The pressure contours on the centre plane of the primary jet are shown in figure 6.30(a).

Due to diffusive nature of the recess, the pressure gradually increased after the jet

entered into the recess. The pressure distribution was not symmetric at the exit of the

148


recess. On the long side pressure gradually reduced to just above atmospheric pressure,

then became sub-atmospheric and remained so for a certain distance into the domain.

On the short side pressure gradually increased and at the exit of the recess there was a

high-pressure zone, which extended in to the domain. Figure 6.30(b) shows the primary,

upper and lower secondary streamlines after the jets entered into the recess. Most of the

primary streamlines mixed with the upper and lower secondary streamlines inside the

recess, exited the recess together, and penetrated the cross-flow layer. Only a few of the

primary streamlines stayed on the centre plane of the primary jet (figure 6.30(c)) and

remained with in the cross-flow. It was mainly the cross-flow that appeared at the centre

plane of the primary jet (figure 6.29(a)) and flowed along the wall, consistent with the

experimental measurements (figure 5.21 (a)) and the observations of Yan and Perry

(1994).

Figure 6.30(a): Pressure contour at the centre plane of the primary jet for φ=3.0

Figure 6.30(b): Primary, upper and lower secondary streamlines inside the recess

149


Streamlines entrained into cross-flow

Streamlines entrained into secondary jet

Figure 6.30(c): Primary streamlines viewing from the top

Figure 6.31 shows the pressure contours and lower secondary streamlines at the centre

plane of the lower secondary jet. The lower secondary jet penetrated through the cross-

flow layer due to the higher momentum. The cross-flow had little influence on the

secondary jet. Only the cross-stream pressure difference at the exit of the recess was

responsible for little deviation of the lower secondary jet from the geometric axis of the

burner.

Figure 6.31: Pressure contour and lower secondary streamlines at the centre plane of the

lower secondary jet.

150



The effect of jet velocity ratio on jet development for rectangular slot-burners in the

presence of cross-flow has been investigated numerically in this chapter. Geometries B

and D were chosen for investigation. The commercial CFD software CFX-5 was used

for numerical investigation. The mathematical model employed to obtain the predictions

solves numerically the governing equations for the ensemble-averaged values of the

components of the velocity vectors, pressure and turbulence parameters. The RANS

equations were solved and the SST turbulence model was used for turbulence closure,

which is two-layer model combination of k-ε and k-ω models.

The mechanism by which primary and lower secondary jets deviated towards the wall in

the presence of cross-flow was revealed from the numerical simulation. The numerical

results were validated against the experimental result then these results were used for

further explanation of near flow field.

For geometry B and for φ=1.0, both the cross-flow jet and the cross-stream pressure

difference across the primary and lower secondary nozzles due to the angled nozzle

geometry were responsible for the deviation of the jet towards the wall. For φ=3.0, after

exiting from the nozzle, primary jet mixed with the higher momentum secondary jets

and pierced the cross-flow layer. Only cross-stream pressure gradient caused the

primary and lower secondary jets bent slightly from the geometric axis because the

momentum of the secondary jet was very high and the primary jet was entrained into it

after exiting from the nozzle. Entrainment of the surrounding fluid into the jets for

φ=1.0 occurred only at the interface between the cross-flow jet and the surrounding

fluid and the vectors of the entrainment fluid were normal to the geometric axis of the

burner. For φ=3.0, the vectors of the entrainment field were more complex than those

for φ=1.0. Upstream of the burner the vector field was parallel to the geometric axis of

the burner while downstream of the burner the vector field was normal to the cross-flow

jet. Counter rotating vortex pairs formed for both jet velocity ratios in the middle region

in front of the primary nozzle very near to the wall. For φ=1.0, it started from

x/De=0.078, extended upto x/De=0.468, and then disappeared. For φ=3.0, this twin

vortex was more prominent than for φ=1.0 and extended upto x/De=0.938.

151


For geometry D, the flow development inside the recess was very complex. The flow

experienced a step-change at the sidewalls as it moved from the duct to the recess. For

φ=1.0, at the centre plane of the primary jet the flow was completely separated from the

short side and it was neither separated nor completely attached on the long side. There

was a reverse flow on the long side at the beginning of the recess. The reverse flow

region was more prominent for φ=3.0. At the centre plane of the lower secondary jet the

flow was separated on the short side but attached to the wall on the long side for φ=1.0.

For φ=3.0 the flow was attached on both long and short side and acted as an efficient

diffuser. The primary jet diverged significantly from the geometric axis and remained

within the cross-flow for φ=1.0. For φ=3.0 the jet turned down further and lay close

against the wall. The mechanism of primary jet deviation was totally different for φ=1.0

and 3.0. For φ=1.0, cross-stream pressure difference as well as an external force that

came from the cross-flow was responsible for the deviation of the jet. For φ=3.0, the

primary jet split and mixed with the secondary jets inside the recess and it was mainly

the cross-flow that appeared at the centre plane of the primary jet. A twin vortex system

was formed for φ=1.0 at the same region as it was for geometry B. It started at the exit

of the recess (x/De=0) and extended up to x/De=0.390. No counter rotating vortex pair

was formed for φ=3.0.

So far single-phase flow has been discussed in the presence of cross-flow for different

jet velocity ratios. In an actual tangentially-fired furnace multiphase phenomenon is

important because of the injection of coal particles together with hot flue gases in the

primary port of the burner. Next chapter will focus on two-phase flow in the presence of

cross-flow for different jet velocity ratios.

152

NOTE


Chapter-7 Experimental investigation of two-phase flow

with cross-flow

Chapter 7-Experimental investigation of two- phase flow with cross-flow

7. Experimental investigation of two-phase flow with cross-flow

This chapter presents the results from the experimental investigation of two-phase flow

development in the rectangular slot-burners in the presence of cross-flow. Geometries B

and D have been investigated for φ=1.0 and 3.0. It was observed in chapters 5 and 6 that

the cross-flow significantly influenced the near-field flow development of jets in the

slot- burner for every burner geometry and velocity ratio considered. Two-phase

phenomenon is important in burner jets due to the interaction of the gas and solid coal

particles, and differences in particle size and loading between the main and vapor

burners, which alter their behaviour. The main purpose of the jets in a lignite

tangentially-fired furnace is to heat the coal air by mixing it with entrained hot flue

gases, and to deliver the fuel and air to the correct location in the centre of the furnace.

This raised the question as to where the pulverized fuel particle paths would lie.

Many authors have considered the scaling laws for modeling of two-phase flows where

the secondary phase consists of particles (e.g. Boothroyed, 1971). The following

groupings of relevant variables appear suitable:

π1=gd/Ue2=1/Froude number=gravity forces/inertia forces

π2=µd/(ρpUedp2)=Viscous forces/inertia forces

π3= 4mp/πρUed2= Ge=Particle/air mass flow ratio

π4=ρ/ρp= fluid pressure forces/inertia forces

π5=ρd/ρpdp= particles pressure forces/inertia forces=particle trajectory number

Using geometric and thermo-fluid flow parameters for a typical furnace (Perry and

Hausler, 1984) the dominant forces relative to the inertia force are the viscous (π2) and

particle pressure (π5) forces.

154



7.1.1. Geometry B

7.1.1.1. Comparison for φ=1.0 with cross-flow

Figures 7.1 shows the comparison between the gas velocity and the particle velocity in

the centre plane of the primary jet for φ=1.0. The measurements locations were same as

for single-phase that is y/De=0, 0.5, 1.0, 2.0, 3.0, 4.0, 5.0 and 9.0 (figure 5.3) and the

resultant velocities, Ures, were plotted for comparison. All velocities were normalized to

the centreline velocity, Uce, (figure 5.3) located at the exit of the primary nozzle

(x/De=0, y/De=0, z/De=0) for φ=1.0.

At y/De=0 and 0.5, the gas and particle-phase velocities were same at the exit of the

nozzle. However, at around x/De=1.0, the particle-phase velocity was slightly lower

than the gas-phase velocity but beyond that the velocity of both phases were the same

again. This was the mixing region between the primary jet and the cross-flow jet where

the particle velocities were less influenced by the cross-flow jet. The penetration and

spreading of the particle-phase was similar to gas-phase. At y/De=1.0, near to the wall,

the particle-phase velocities were slightly higher than the gas-phase but the peak

velocities were at the same position. At y/De=2.0, peak velocity of the particle-phase

was slightly higher than the gas-phase and this trend continued up to y/De=9.0. Higher

peak values at these positions occurred because of the higher momentum of the particle-

phase.

155



0

0.2

0.4

0.6

0.8

1

0 1 2 3 4 5 6


0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5 6x/De

Ure

s/U

ce

Gas velocity

Particle velocity

x/De

Ure

s/U

ceGas velocity

Particle velocity


0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5 6


0

0.2

0.4

0.6

0.8

1

0 1 2 3 4 5 6x/De

Ure

s/U

ce

Gas velocity

Particle velocity

x/De

Ure

s/U

ce

Gas velocity

Particle velocity


0

0.2

0.4

0.6

0.8

1

0 1 2 3 4 5 6x/De

Ure

s/U

ce

Gas velocity

Particle velocity


0

0.2

0.4

0.6

0.8

1

0 1 2 3 4 5x/De

Ure

s/U

ce

6

Gas velocity

Particle velocity


0

0.2

0.4

0.6

0.8

1

0 1 2 3 4 5 6


0

0.2

0.4

0.6

0.8

1

0 1 2 3 4 5 6

x/De

Ure

s/U

ce

Gas velocity

Particle velocity

x/De

Ure

s/U

ce

Gas velocity

Particle velocity

Figure 7.1: Comparison of resultant velocity between the gas-phase and particle-phase

in the centre plane of the primary jet for φ=1.0

156



Figure 7.2 shows the comparison in the centre plane of the primary jet for φ=3.0. The

figure shows that generally the velocities of the particle-phase were the same as

observed in the gas-phase. However, there were some minor variations between the

particle and the gas velocities. At y/De=0 and 0.5, particle velocities were slightly

higher than the gas at the exit of the nozzle, the magnitude then fell and at x/De=1.0 the

particle velocities were lower than the gas-phase, where the difference between the gas

and particle velocities was at a maximum. At these positions, y/De=0 and 0.5, the

boundaries of the particle-phase were slightly thinner in the region between

1.0≤x/De≤2.5. The peak velocity of the particle-phase was slightly higher at y/De=1.0

but occurred at the same position. This trend continued upto y/De=9.0.

157



0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5 6x/De

Ure

s/U

ceGas velocity

Particle velocity


0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5 6x/De

Ure

s/U

ce

Gas velocity

Particle velocity


0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5 6x/De

Ure

s/U

ce

Gas velocity

Particle velocity


0

0.2

0.4

0.6

0.8

1

0 1 2 3 4 5 6

x/De

Ure

s/U

ce

Gas velocity

Particle velocity


0

0.2

0.4

0.6

0.8

1

0 1 2 3 4 5 6x/De

Ure

s/U

ce

Gas velocity

Particle velocity


0

0.2

0.4

0.6

0.8

1

0 1 2 3 4 5 6 7 8x/De

Ure

s/U

ce

Gas velocity

Particle velocity


0

0.2

0.4

0.6

0.8

1

0 1 2 3 4 5 6 7 8

x/De

Ure

s/U

ce

Gas velocity

Particle velocity


0

0.2

0.4

0.6

0.8

1

0 1 2 3 4 5 6 7 8

x/De

Ure

s/U

ce

Gas velocity

Particle velocity



158


7.1.2. Geometry D


Figures 7.3 shows the comparison between the gas and the particle-phase velocity in the

centre plane of the primary jet for φ=1.0. The measurement locations were the same as

geometry B. The general shape of the velocity profiles between the gas and the particle-

phases were same for all locations. However, there were some minor differences

between the gas and particle-phase velocities. At y/De=0, particle velocities were lower

in the region between 0.6≤ x/De≤1.4, corresponding to the mixing of the primary jet

with the cross-flow jet. As a result the gas-phase velocity reached a second peak in this

region. Particle-phase velocity was less influenced by the cross-flow jet due to higher

density and hence the velocity was not increased as much as it was for the gas-phase.

The velocities near the wall were the same for both gas and particle-phases at all

locations. The magnitude of the peak velocities were also same and occurred at the

same position up to y/De=2.0. The peak velocity of the particle-phase deviated slightly

from the gas-phase at y/De=3.0 and occurred farther from the wall. At y/De=9.0 both

velocity profiles were flat indicating the self-similarity region.

159



0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5x/De

Ure

s/U

ce

Gas velocity

Particle velocity


0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4x/De

Ure

s/U

ce

5

Gas velocity

Particle velocity


0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5


0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5x/De

Ure

s/U

ce

Gas velocity

Particle velocity

x/De

Ure

s/U

ce

Gas velocity

Particle velocity


0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5x/De

Ure

s/U

ce

Gas velocity

Particle velocity


0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5x/De

Ure

s/U

ce

6

Gas velocity

Particle velocity


0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5 6


0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5 6x/De

Ure

s/U

ce

Gas velocity

Particle velocity

x/De

Ure

s/U

ce

Gas velocity

Particle velocity



160



Figure 7.4 shows the comparison between the gas and particle-phase velocity in the

centre plane of the primary jet for φ=3.0.


0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5x/De

Ure

s/U

ce

Gas velocity

Particle velocity


0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5x/De

Ure

s/U

ce

Gas velocity

Particle velocity


0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5x/De

Ure

s/U

ce

Gas velocity

Particle velocity


0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5x/De

Ure

s/U

ce

Gas velocity

Particle velocity


0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5x/De

Ure

s/U

ce

Gas velocity

Particle velocity


0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5 6x/De

Ure

s/U

ce

Gas velocity

Particle velocity


0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 3 4 5 6x/De

Ure

s/U

ce

Gas velocity

Particle velocity


0

0.2

0.4

0.6

0.8

1

1.2

1.4

0 2 4 6 8 1x/De

Ure

s/U

ce

0

Gas velocity

Particle velocity



161


There was a significant difference between the profiles at y/De=0 and 0.5. In the mixing

region of the primary jet and the cross-flow jet, 0.2≤x/De≤1.2, the gas velocities were

higher than the particle velocities. The velocities near the wall were the same for both

gas and particle-phases between y/De=1.0 to 9.0 except at y/De=2.0 where there was a

tendency of separation in case of gas-phase but the particle-phase did not show any

reverse flow. The peak velocity of the particle-phase was higher between y/De=3.0 to

5.0 due to higher momentum.


This chapter presents the isothermal investigation of two-phase flow in the presence of

cross-flow. Solid spheres were used with a mean diameter of 66µm to represent coal

particles and were introduced at the centre of the primary jet. Due to physical constraint

of the experimental facility used the mass loading of the particle-phase was reduced

from typical furnace levels of order 0.2 to model value 0.01. For geometry B generally

the velocity profiles for the gas-phase and particle-phase were qualitatively similar for

both φ=1.0 and 3.0, however there were some minor discrepancies. At y/De=0 and 0.5,

in the region between 0.2≤x/De≤1.2 where the primary jet mixed with the cross-flow jet

the gas velocities were higher than the particle velocities. Further downstream the

particle velocities showed higher peaks than the gas velocities. For geometry D and

φ=1.0 in the mixing region between the primary jet and the cross-flow jet the gas

velocities were higher as geometry B. Further downstream the peak velocities of the

particle-phase were slightly deviated and occurred farther from the wall. For φ=3.0

downstream of the nozzle (y/De=2.0) there was a tendency of separation for the gas-

phase near the wall while the particle-phase was attached to the wall.

So far isothermal investigations of single-phase and two-phase flow have been

performed for different φ because in a real furnace φ varies from 1.0 to 3.0. The purpose

was to investigate the aerodynamics of the developing jets in the near field region of

rectangular slot-burners in tangentially-fired furnaces and provide more understanding

of the mechanisms of flow development. In order to obtain the near field region similar

to the tangentially-fired furnace cross-flow was introduced.

162


In the presence of cross-flow the primary jet deviated completely from the geometric

axis of the burner for both geometry B and D and lie against the furnace wall. From

these results it can be concluded that in a real furnace a large quantity of fuel will be

directed close to the wall between the burners where oxygen concentration is low. The

three dimensional flow in the furnace may then deliver these fuel rich volumes either

down the wall to the hopper region or upwards and again close to the walls. Such

characteristics may contribute to substantial wall fouling and a consequent reduction in

the heat transfer rate across the wall. Also the unsteady nature of the primary jet and the

tendency for the primary jet to turn away from the secondary jets may significantly

contribute to combustion instability and poor combustion characteristics and possibly

greater green house gas emissions.

Yallourn stage-2 power station in the Latrobe Valley of Victoria, Australia, were

designed and built in the late 1970s and much of the design at that time was based on

experience and empirical design rules. The furnace is tangentially-fired and the burners

are similar to geometry D considered in this research program. The walls of the recess

geometry D were made divergent, probably in anticipation of the recess acting to some

extent as a diffuser. The optimum divergence angle for a diffuser is around 6° while

walls of the recess diverged at 10°, and the walls were much shorter than those of a

well-designed diffuser. As a result the primary jet separated completely on the short

side of the recess wall. On the long side separation occurred at the beginning of the

recess and the burner did not act as an efficient diffuser. Hence it is not surprising that

the primary jet became unstable at φ=3.0. One of the remedies to reduce the

unsteadiness of the primary jet is the reduction of the divergence angle.

In order to obtain the complex flow field typically occurred in a tangentially-fired

furnace investigation of full furnace is required. The next chapter describes the

numerical investigation of combustion in a full-scale tangentially-fired furnace of using

conventional lignite and MTE lignite. The results obtained from the isothermal

investigations were used to aid in describing the complex aerodynamics of the

tangentially-fired furnaces.

163

Chapter-8 Combustion in a tangentially-fired furnace

Chapter 8-Combustion in a tangentially-fired furnace

8. Combustion in a tangentially-fired furnace

The combustion of conventional lignite and MTE lignite in a full-scale tangentially-

fired furnace has been simulated using CFD and the results are presented in this chapter.

The furnace was based on Yallourn stage–2 in Victoria, Australia. The commercial CFD

code CFX-4 has been used in this study. The time averaged Navier-Stokes equations

were solved with standard k-ε turbulence model. The combustion model comprised,

Shah’s discrete transfer model for radiation, single-reaction devolatilization model for

devolatilization, eddy break-up model for gaseous combustion, NOx model and Gibb’s

model for char oxidation. For devolatilization the values of pre-exponential factor and

activation temperature were taken as 2.0x 104 and 5941.0 K (Hart et al. 2000)

respectively. The activation energy for conventional lignite and MTE lignite char were

taken as 151 kJ/mol and 166 kJ/mol respectively, which were most appropriate for

Victorian brown coal investigated by Ballantyne et al. (2003). Particles were tracked

using Lagrangian equation of motion from the inlet ports until particles burn out or

leave the furnace. The detailed descriptions of these models were given in chapter 2.

After the combustion of the MTE lignite the predicted velocity vectors and temperature

contours were compared with the conventional lignite combustion process and the

temperature contours and oxygen concentration at different furnace level of the

conventional lignite combustion process were validated against the available physical

measurements.

The characteristics of the Victorian lignite before entering into the furnace are

summarized in table 8.1.

Table 8.1: Coal characteristics before entering into the furnace

Coal diameter µm 60-100 Carbon % 54.105

Volatile matter % 40.24 Hydrogen % 3.932

Moisture Content % 20.0 Oxygen % 19.636

Ash % 1.36 Nitrogen % 0.511

Fixed carbon % 38.4 Sulphur % 0.456

165


The porosity of dry lignite is 0.395 (Durie, 1991). MTE process reduces the porosity of

the Victorian lignite by approximately 50% to 0.1975 (Ballantyne, 1992). Although

some sodium is removed with the expressed water the extent of sodium removal is not

as great as anticipated for this particular lignite.

Figure 8.1 shows the proposed schematic diagram showing all the processes for one

mill before the MTE lignite (primary jet velocity 20m/s) enters in to the furnace. The

amount of hot flue gas that has to be extracted for conventional lignite is 53.0 kg/s. By

calculation it was found that if MTE lignite were used in the existing furnace, the

amount of the hot flue gas would be reduced to 13.9 kg/s. The water vapor pumped into

the furnace from each mill would be reduced from 15.9 kg/sec to 3.1475 kg/sec. These

will reduce the velocity at the primary nozzle from 20m/s to 7.47m/s. The effect of

reduced mass flow rate has also been investigated in this study and comparison has been

made with the conventional combustion process. Additional air was then supplied in the

primary duct to maintain the original velocity of 20m/s at the primary nozzle. The

extraction of hot flue gas was increased (from 13.9 kg/s to 19.35 kg/s) for heating up the

additional air from ambient temperature to inlet temperature (120°C) but still the

amount was much lower than in conventional lignite combustion (53.0 kg/s). The

results were compared again with the conventional combustion process.

Raw wet coal 27.3 kg/sec per mill Water 27.3*.667=18.2091 kg/sec

MTE Process 70% water remove

Mill water 5.456 kg/sec (37.5%) 3.1475 kg/sec evaporate in the mill

Hot flue gas 19.35 kg/sec 14.55 kg/sec Coal

oisture)11.4025 kg/sec Coal (20% m

Additional Air 46.40 kg/sec +Leakage Air 28.83 kg/sec

IB

UB

LB

Figure 8.1: Proposed schematic diagram showing all the processes before the MTE

lignite (velocity 20m/s) enters in to the furnace.

166


8.1. Numerical model verification

The MTE process is under development and has not been used in the real power station

for the commercial production of electricity, hence no experimental data was available

for comparison with the numerical results. So the numerical results for the conventional

lignite combustion have been validated against the available physical measurements.

The CFD model was first validated against an industrial scale tangentially-fired furnace

by Hossain (2001), which gave resonably good agreement. The simulation of the

Yallourn stage-2 boiler was carried out and a comparison has been made with the

physical measurements of McIntosh et al. (1985). Test were conducted at the centre

plane of the lower primary nozzle as shown in figure 8.2(a) and 8.2(b).

Vertical Plane

Horizontal Plane

(b)(a)

Figure 8.2(a-b): Measured data level.

Figure 8.3(a) shows the comparison of the predicted temperature contours with the

available experimental temperature contours measured by McIntosh et al. (1985) at the

center plane of the lower primary nozzle., The temperature increased with the increase

of the distance into the furnace. In the middle region of the nozzle, along the geometric

axis, the predicted temperatures (200-300°C) were in resonably agreement. The

maximum temperature measured was 700°C and occured at the same position in both

predicted and measured temperature contours. Figure 8.3(b) shows the comparison of

the predicted temperature contours with the physical measurements on a vertical plane

perpendicular to the geometric axis of the burner and 1100 mm into the furnace (see

figure 8.2(b)). Near the center of the lower primary nozzle the predicted temperatures

167


(200-300°C) agreed well with the experimental values. Although the temperatures were

overpredicted at the lower right hand side of the lower primary nozzle, the upper left

hand side matched well with the measured values. Figure 8.3(c) shows the comparison

of the predicted oxygen concentration (vol %) with the measured values on the same

plane. In the region between the bottom half of the lower primary nozzle and the upper

half of the lower bottom secondary nozzle, the range of predicted oxygen levels (14-

20%) were well matched with the physical measurements. The validation shows that the

overall numerical results for the conventional lignite combustion agree reasonably well

with the measured values atthough there are some diccrepracies. The results are

therefore considered to be reliable for further comparison with the MTE lignite

combustion.

6

2

Figure 8

data’s at

9

700C 700C

.3(a): Comparison of the predicted temperature contours with experimental

the center plane of the lower primary nozzle.

168


9

6

4 2 200C

200C

Figure 8.3(b): Comparison of the predicted temperature contours with experimental

values in a vertical plane 1100 mm from the exit of the burner towards the furnace.

5

4 1

Figure 8.3(c): Comparison of the predicted oxygen concentration with experimental

values in a vertical plane 1100 mm from the exit of the burner towards the furnace.

8.2. Results and discussions

Figure 8.4(a) shows the velocity vectors on a horizontal plane for the conventional

combustion process. The plane is at the center of the lower primary nozzle. The jets

strongly penetrated into the furnace. A flow structure typical for tangentially-fired boiler

was observed at this level. A large primary vortex was formed in the middle of the

furnace, which was squeezed due to the expansion of the gas associated with

169


combustion (Ahmed et al., 2002). This vortex increased the residence time of the

incoming coal in the furnace for stable and complex combustion. Secondary

recirculations were formed near the two inactive nozzles, which resulted in increased

entrainment of fluid into the jets and enhanced spreading. These secondary

recirculations were zones where the particles might experience a long residence time

and under the correct conditions might undergo sintering and become sticky, increasing

the potential for coal to stick to the walls and cause fouling. A pair of weak recirculation

zones occurred in the middle of both sidewalls, also affecting the entrainment of the

surrounding fluid resulting in higher mixing of the coal with the jets. The jets from

nozzles 2 and 6 (figure 8.4(a)) deviated from their geometric axes and were pushed

against the furnace wall by the cross-flow, consistent with the observations of the

isothermal jets in the presence of cross-flow studied in the previous chapters.

8

7

6

4 5

3

2

1 Front Wall

Figure 8.4(a): Velocity vectors at the centre plane of the lower primary nozzle for

lignite combustion.

170


Front Wall

Figure 8.4(b): Velocity vectors at the centre plane of the lower primary nozzle for MTE

lignite combustion with velocity 7.47 m/s

Front Wall

Figure 8.4(c): Velocity vectors at the centre plane of the lower primary nozzle for MTE


171


Figure 8.4(b) shows the velocity vectors for the MTE lignite simulation with reduced

mass flow rate at the same level. The fluid flow patterns changed significantly. The

penetration of the jets was reduced and the spreading of the jets was less than in figure

8.4(a). The primary recirculation zone also shifted and due to the reduced expansion of

the jets the primary vortex was less squeezed. The secondary recirculation zone, near

two inactive nozzles, moved toward the sidewalls and there were no vortices in the

middle of the sidewalls, which were observed in the conventional combustion process.

Figure 8.4(c) shows the velocity vectors at the same level for the MTE lignite with a

primary jet velocity increased to 20m/s. The flow field was very similar to the

conventional lignite combustion (figure 8.4(a)).

Figure 8.5(a) shows velocity vectors at the center plane of the upper primary nozzle for

conventional lignite combustion. The recirculation zones in the middle of the side walls

were stronger than those of lower primary nozzle as shown in figure 8.4(a).

Front Wall

Figure 8.5(a): Velocity vectors at the centre plane of the upper primary nozzle for

lignite combustion.

172


Front Wall

Figure 8.5(b): Velocity vectors at the centre plane of the upper primary nozzle for MTE


Front Wall

Figure 8.5(c): Velocity vectors at the centre plane of the upper primary nozzle for MTE


173


At this level the existence of a recirculation zone on both sides of each jet indicated the

entrainment of the surrounding fluid into the primary and secondary jets. Figure 8.5(b)

shows velocity vectors at the same level for the MTE lignite with reduced velocity (7.47

m/s). The pair of recirculation zones in the middle of the sidewalls were not significant

here, which was an indication of poor entrainment of the surrounding fluid into the jets

resulting in less spreading of the jets. Figure 8.5(c) shows the velocity vectors for MTE

lignite combustion with a velocity of 20m/s at the same level. The vector diagram was

very similar to conventional lignite combustion of figure 8.5(a).

Figure 8.6(a) shows the temperature contours at the center plane of the lower primary

nozzle for conventional combustion process. The upper right hand side burner (figure

8.2(b)) has been chosen because the temperature contours of this burner were validated

against the physical measurements.

9 6

2

Figure 8.6(a): Temperature contours at the centre plane of the lower primary nozzle for

conventional lignite combustion

174


69

2

Figure 8.6(b): Temperature contours at the centre plane of the lower primary nozzle for

MTE lignite combustion with velocity 7.47 m/s

6 9

2

Figure 8.6(c): Temperature contours at the centre plane of the lower primary nozzle for

MTE lignite combustion with velocity 20.0 m/s

At the exit of the nozzle, temperature was low and gradually increased into the furnace.

The temperature also increased from the geometric axis towards the burner wall and at

the wall it assumed a value of 7000C. Figure 8.6(b) shows the temperature contours for

the MTE lignite with reduced mass flow rate at the same level. At the exit of the nozzle

the temperature was 2000C higher than what was observed in figure 8.6(a). Near the

furnace wall the temperature was 10000C whereas it was only 5000C in conventional

lignite combustion (figure 8.6(a)). The amount of heat required to raise the temperature

of the water vapor from 1400C (inlet temperature) to 9000C (furnace average

175


temperature) was much less because the amoumt of water vapor in the furnace was

reduced. This additional heat increased the temperature when the MTE lignite was used

with reduced velocity. In figure 8.6(c), additional air was supplied in the primary duct.

The amount of hot flue gas was reduced from 53.0 kg/s to 19.35 kg/s and water vapor

was reduced from 15.9 kg/s to 3.1475 kg/s) in the mill for MTE lignite, so additional air

(46.40 kg/s) was supplied to maintain the original velocity (20 m/s) at the exit of the

primary nozzle. The temperature contours were almost similar to figure 8.6(a) as

additional heat was used to increase the temperature of additional air from inlet

temperature to furnace average temperature. This phenomenon can be understood more

clearly by observing the shaded temperature contours at the centre of the furnace.

Figure 8.7(a-c) shows the shaded temperature contours for conventinal lignite

combustion, MTE lignite combustion with reduced velocity and MTE lignite

combustion with velocity 20m/s respectively. In figures 8.7(a) and 8.7(c), the average

temperature in the convection zone of the furnace was around 8500C whereas it was

10000C in figure 8.7(b). The maximum temperature in the furnace was 15000C which

occurred in the lower and upper main burner levels where the combustion took place. In

the case of MTE lignite combustion with reduced velocity (figure 8.7(b)), maximum

temperture zone was higher and its location shifted towards the furnace wall.

176


Figure 8.7(a): Shaded temperature contours at the centre of the furnace for conventional

lignite combustion

Figure 8.7(b): Shaded temperature contours at the centre of the furnace for MTE lignite

combustion with velocity 7.47 m/s

177


Figure 8.7(c): Shaded temperature contours at the centre of the furnace for MTE lignite

combustion with velocity 20.0 m/s

Figure 8.8(a) and 8.8(b) show the comparison of NOx concentration between the

conventional lignite combustion and MTE lignite combustion (velocity 20 m/s) in the

horizontal XY plane at a height Z=50m. The plane is just below the top exit of the

furnace. The NOx emission was slightly higher in figure 8.8(b), which is negligible,

after the introduction of additional air through the primary duct.

5

7

2

Figure 8.8(a): NOx concentration (vol ppm) for the conventional lignite combustion in

the XY plane at z=50m.

178


67

5 2

Figure 8.8(b): NOx concentration (vol ppm) for MTE lignite combustion (velocity

20m/s) in the XY plane at z=50m.


Results from the simulation of the conventional lignite combustion and MTE lignite

combustion with reduced velocity have been presented in this chapter. There was a

significant change in flow pattern between the conventional lignite combustion and the

MTE lignite combustion with reduced velocity, where the penetration and spreading of

the jets were reduced. The recirculation zones near the middle of the sidewalls were not

prominent resulting in poor entrainment of the surrounding fluid into the jets.

Temperatures at the center plane of the lower primary nozzle were higher for MTE

lignite with reduced velocity. This might be the effect of introducing less water vapour

into the furnace, because the water vapor that had to be pumped into the furnace (3.1475

kg/s per mill) for MTE lignite was much less than for the conventional lignite (15.9

kg/s). The amount of heat required to increase the temperature of this water vapor from

inlet temperature (140ºC) to furnace average temperature (900ºC) was much less

compared to the conventional lignite combustion, so the removal of this heat sink

increased the temperature in the furnace. As a result the average temperature in the

convection zone of the furnace was higher (1000ºC) for MTE lignite combustion with

reduced velocity. The velocity of the fuel jet was then increased by supplying additional

air through the primary duct, after which the flow field and temperature contours for

MTE lignite combustion were similar to conventional lignite combustion. A comparison

of hot flue gas extraction from the furnace and evaporation of water in the mill has been

shown in table 8.2.

179


Table 8.2 : Comparison of extraction of hot flue gas and evaporaton of water in mill

Combustion Conventional lignite

MTE lignite with a velocity of 7.47 m/s

MTE lignite with a velocity of 20.0 m/s

Extraction of hot flue gas

53.0 kg/s 13.9 kg/s 19.35 kg/s

Evaporation of water in mill

15.9 kg/s 3.1475 kg/s 3.1475 kg/s

One of the major advantages of using the MTE lignite is the reduction in the required

flue gas extraction from the furnace (19.35 kg/s) compare to the conventional lignite (53

kg/s). If MTE lignite were used in the existing furnace, the extraction of hot flue gas

would be reduced by 65%. However, energy required in the MTE drying of the coal has

been found by Huynh et al. (2003) to be only half of the saving energy (32.5%). Lots of

other issues are under investigation stage like storage and handling of MTE product and

the possible uses for the large volumes of water that will be generated from MTE

process (Chaffee et al., 2003). So the proper utilization of MTE lignite in the existing

furnace can reduce the cost of the power plant and increase the overall efficiency.

180

Chapter-9 Conclusions and Recommendations

Chapter-9 Conclusions and recommendations

9. Conclusions and recommendations

9.1. Conclusions

The observations of this research program can provide useful information for improving

the burner design of existing boilers as well as in the design of new boilers. The

investigation started with the effect of jet velocity ratios for different rectangular slot-

burners without cross-flow. The effect of cross-flow on burner jet development was

then investigated as this has a significant effect in the near field region of developing jet

in a tangentially-fired furnace. The isothermal two-phase flow in the presence of cross-

flow was investigated to locate the possible path of the coal particles for various jet

velocity ratios. Numerical investigations of the effect of combustion in a full-scale

tangentially-fired furnace were performed and the possibility of burning MTE lignite in

the existing furnaces was assessed.

Initial numerical investigations of two recessed burner geometries revealed that burner

geometry and jet velocity ratio have a significant influence on jet development. For

geometry C, the primary jet developed along the geometric axis of the burner for φ=1.0

but tended to move away from this axis towards the long side wall of the recess with the

increase in φ. For geometry D the centreline decay rate within the recess was reduced

but increased rapidly some distance downstream as it discharged into the furnace. The

primary jet deflected strongly towards the long side of the recess for all φ. The degree of

deflection of the primary jet increased with increasing φ. For φ=3.0, the primary jet

apparently disappeared at around four and a half diameters downstream of the jet exit

and became very unstable.

In the presence of cross-flow the primary jet deviated completely from the geometric

axis of the burner for both geometry B and D and lie against the furnace wall. For

geometry B and φ=1.0, both the momentum of the cross-flow jet and cross-stream

pressure differences across the nozzles were responsible for the deviation of the jets. For

φ=3.0, only the cross-stream pressure gradient caused the primary and secondary jets

bent from the geometric axis. A counter rotating vortex pair (CRVP) formed for both jet

velocity ratios behind the initial region of the primary jet core very near to the wall

where primary jet acted as a blockage for the cross-flow. The CRVP was more

prominent for φ=3.0.

182


For geometry D, the flow development inside the recess was very complex. For φ=1.0,

on the centre plane of the primary jet the flow was completely separated from the short

side. There was a separation bubble on the long side at the beginning of the recess and

was more prominent for φ=3.0. On the centre plane of the lower secondary jet the flow

was separated on the short side only for φ=1.0. For φ=3.0 the flow was attached on both

the long and short sides and acted as an efficient diffuser. The mechanism of primary jet

deviation was totally different for φ=1.0 and 3.0. For φ=1.0, cross-stream pressure

difference and an external force from the cross-flow were responsible for the deviation

of the jet. For φ=3.0, the primary jet diffused outwards and mixed with the secondary

jets inside the recess and it was mainly the cross-flow that appeared on the centre plane

of the primary jet. The CRVP was formed for only φ=1.0 at the same location as in

geometry B.

Numerical investigations of the effect of combustion of using MTE lignite in the

existing furnaces were performed. One of the major findings was the reduction of hot

flue gas from the furnace. The amount of hot flue gas extraction from the furnace per

mill for conventional lignite was 53.0 kg/s. From energy balance it was found that if

MTE coal were is used in the existing furnace, the amount of the hot flue gas would be

reduced to 13.9 kg/s. The water vapour pumped into the furnace would be reduced from

15.9 kg/sec to 3.1475 kg/sec. These would reduce the velocity in the primary port from

20m/s to 7.47m/s. The effect of reduced velocity was also investigated in this thesis.

Additional air was then supplied in the primary port to maintain the original velocity

(20m/s). This was done to maintain the gas-dynamics flow pattern for which the furnace

was designed. The extraction of hot flue gas was increased (form 13.9 kg/s to 19.35

kg/s) for heating up the additional air from ambient temperature to inlet temperature

(120°C) but still the amount was much less compared to the conventional lignite

combustion (53.0 kg/s). Thus using of MTE lignite in the existing furnace can reduce

the extraction of the hot flue gas by 65%.

183


9.2. Recommendations for further work

Several recommendations are now made for future work to extend the completed

research. These recommendations aim to create better understanding of the near field

development of jets as well as combustion of lignite in tangentially-fired furnaces. The

recommendations are listed below.

• Two types of cross-flow exist in a tangentially-fired furnace. First where the cross-

flow favors the burner jet and second where the cross-flow opposes the burner jet.

Due to time constraints only the first type of cross-flow was investigated in detail

in this research program. Although Yan and Perry (1994) investigated to some

extent the effect of cross-flow approaches to the burner jet, detailed investigation

of mean velocities and fluctuating components as well as numerical simulation of

the near field region is needed to understand the complex aerodynamics of a

tangentially-fired furnace.

• No experimental data was taken for burner geometry D inside the recess due to

limiting movement of the traversing mechanism. So there was no scope to validate

the numerical results of geometry D inside the recess. Fluid flow inside the recess

is very complex and important in determining the near field of the developing jets.

Measurements inside the recess can validate the findings of the CFD studies.

• Steady state Navier-Stokes equations were solved for numerical investigation of

rectangular slot-burners and full-scale tangentially-fired furnaces. The solution

became unstable at higher secondary to primary jet velocity ratio suggesting the

use of Navier-Stokes equations in the transient mode.

• In this research program the two-phase flow was investigated for only one mass

loading (0.01). The effect of increased mass loading is important and needs to be

investigated.

• In investigating the combustion effects of using MTE lignite in the existing

furnaces it was found that for MTE lignite the extraction of hot flue gas from the

furnace was reduced by 65%, which would increase the steam generation capacity

184


of the boiler. In this research program numerical investigation of the furnace was

performed only. More investigation is required to determine whether the super

heater pipes in the convection zone can sustain with this increased steam

generation capacity.

185

NOTE


Chapter-10 References


10. References

Abbas, A.S. and Lockwood, F.C., 1986, Prediction of Power Station Combustors,21th

Symp. (Int.) on Combustion, The Combustion Institute, Pittsburgh, pp.285-292.

Ahmed, S. and Naser, J., 2002, Simulation of the Coal Combustion of Yallourn Stage-2

Furnace. Proceedings of the Ninth CRC Annual Conference, 27-28 June, Monash

University, Melbourne, Australia, 221-225.

Andreopoulos, J., & Rodi, W., 1984, Experimental investigation of jets in cross-flow, J.

Fluid Mech., 138, 93-127.

Andreopoulos, J., 1982, Measurements in a jet-pipe flow issuing perpendicularly in to a

cross stream, ASME J. Fluids Eng., 104, 493-499.

Andreopoulos, J., 1985, On the structure of jets in a cross-flow, J. Fluid Mech., 157,

163-197.

Badzioch, S., Hawksley P. G. W., 1970, Kinetics of thermal decomposition of

Pulverized coal particles, Industrial engineering Chemistry Process Design and

Development, pp. 521.

Ballantyne, T. R., Mullinger, P. J., Ashman, P. J., Smith, N. L., 2002, Combustion of

Dewatered Coal in Pulverised Fuel Furnaces. Proceedings of the Ninth CRC Annual

Conference, 27-28 June, Monash University, Melbourne, Australia, 127-130.

Ballantyne. T. R., Mullinger. P. J., Ashman. P. J., 2003, Comparison of Intrinsic

Reactivity of MTE and Non-MTE Lignite Chars. Proceedings of the Tenth CRC Annual

Conference, 12-13 June, Swinburne University of technology, Melbourne, Australia,

249-254.

Barlow, R. S., 1990, Two-phase velocity measurements in dense particle-laden jets,

Expt. in Fluids, 9, 93-104.

187


Boothroyed, R. S., 1971, Flowing Gas-Solids Suspensions, Chapman and Hall, London.

Boyd, R. K. and Kent, J. H., 1986, Three-Dimensional Furnace Computer Modelling,

21th symp. (Int.) on Combustion, The combustion Institute, Pittsburgh, pp. 265-274.

Buchhave, P., George, W. K. and Lumley, J.L., 1979, The Measurement of Turbulence

with the Laser-Doppler Anemometry, Ann. Rev. fluid Mech., Vol 11, pp. 443-503.

Catalano, G. D., Chang, K. S. & Mathis, J. A., 1989, Investigation of turbulent jet

impingement in a confined cross -flow, AIAA J, 27, 11, 1530-1535.

CFX-4–Solver, User Guide, 1997, AEA Technology, Harwell Laboratory, Oxfordshire,

UK.

Chaffee, Alan L., 2003, Water, Water Everywhere: a Drop to Drink? (An Overview of

MTE Supporting Research), Proceedings of the Tenth CRC Annual Conference, 12-13

June, Swinburne University of technology, Melbourne, Australia, 37-42.

Champagne, F. H., Wygnanski, I. J., 1971, An Experimental Investigation of Co-Axial

Jets, Int. J. Haet and Mass Transfer, 14, 1445.

Charpenay, S., Serio, M. A. and Solomon, P. R., 1992, Twenty- Fourth Symposium

(International) on Combustion, The Combustion Institute, Pittsburgh, p.1189.

Chen, J. Y., Mann, A. P. and Kent, J. H., 1992, Twenty-Fourth Symposium

(International) on Combustion, The Combustion Institute, Pittsburgh, 1992, p. 1381.

Chigier, N. A., Beer, J. M., 1964, The Flow Region near the Nozzle in Double

Concentric Jets, Trans ASME, December.

Courant, R., Isaacson, E. and Rees, M.,1952. On the Solution of Non-Linear Hyperbolic

Differential Equations by Finite Differences, Comm. Pure Appl. Math.,5, 243.

188


De Vahl Davis, G., and Mallinson, G.D., 1972. False Diffusion in numerical Fluid

Mechanics, Univ. of New South Wales, School of Mech. and Ind. Eng.

Rept.1972/FMT/1.

Durie R. A., ed. 1991, The Science of Victorian brown Coal, Butterworth-Heinemann

Ltd.

Elsharbagy, K. A., Mobbs F. R. and Cole B. N., 1974, An investigation into the effect of

the particulate phase on the spreading of air-solids jets.

Epple, B., Schneider, R., Schnell, U., Hein, K., 1995, Computerized Analysis of Low-

NOx Coal-fired Utility Boilers, Combustion Science and Technology, 108: 383-401.

Epple, B. and Schnell, U., 1992, Modeling of Coal Combustion and NOx Formation in

Industrial Furnaces, Paper presented at International Joint Power Generation

Conference, Atlanta, Georgia, USA.

Favas, G., Chaffee, A., 2002, Beneficiation of Latrobe Valley Lignite with MTE

processing, Proceedings of the Ninth CRC Annual Conference, 27-28 June, Monash

University, Melbourne, Australia, 153-158.

Ferziger, J. H. and Peric, M., 1997, Computational Methods for Fluid Dynamics,

Springer-Verlag Berlin Heidelberg New York.

Field, M. A., 1963, Entrainment into an air jet laden with particles, BCUR Members

info. Circular No. 273.

Field, M. A., Gill, D. W., Morgan B. B. and Hawksley P. G. W., 1967, The combustion

of pulverized coal, British Coal Utilization Research Association, Leatherhead, Surrey,

p 175.

Findlay, M. J., Salcudean, M., Gartshore, I. S., 1999, Jets in a Cross-flow: Effects of

Geometry and Blowing Ratio, Journal of Fluid Engineering, 121, 373-378.

189


Fiveland,,W. A. and Wessel, R. A., 1988, Numerical Model for Predicting Performance

of Three-Dimensional Pulverized-Fuel Fired Furnaces, Journal of Engineering for Gas

Turbines and Power, 110, 117-126.

Fleckhaus, D., Hishida, K. & Maeda, M., 1987, Effect of laden solid particles on the

turbulent flow structure of a round jet, Exp. in fluids, 5, 323-333.

Foss, J., 1980, Flow visualization studies of jets in a cross-flow, SFB 80 report/T/161,

Karlsruhe University.

Geroge, W. K., 1988, Quantitative Measurement with the Brust-Mode Laser Doppler

Anemometry, Expl. Thermal Fluid Sci., Vol. 1 pp. 29-40.

Gibb, J., 1985, Combustion of Residual Char remaining after devolatilization, Lecture

at course of Pulverized coal Combustion, Mechanical Engineering Department, Imperial

College, London.

Gosman, A. D., Lockwood, F. C. and Salooja, A. P., 1978, Proceedings of the

Seventeenth Symposium (International) on Combustion, Paper 76.

Harlow, F. H. and Welch, J. E., 1965, Numerical Calculation of Time-Dependent

Viscous Incompressible Flow of Fluid with Free Surface, Physics of Fluids, 8, 282-289.

Hart, J., 2001, Numerical Investigation of Isothermal Burner Jet Aerodynamics, Thesis,

School of Engineering & Science, Swinburne University of Technology.

Hart J., Naser. J., 2000, CFD Model of Pyrolysis in a Pressurised Drop tube Furnace,

Proceedings of the Seventh CRC Annual Conference, 22-23 June, University of

Adelaide, Australia, 201-206.

Haven, B. A. & Kurosaka, M., 1997, Kidney & anti-Kidney vortices in crossflow jets, J.

Fluid Mech., 352, 27-64.

190


Hetsroni, G. and Sokolov, M., 1971, Distribution of mass, velocity and intensity of

turbulence in a two-phase turbulent jet, Appl. Mech., 38, 315-327.

Hossain, M., Naser J., 2001, Modelling of Coal Combustion in Full- Scale Industrial

Furnace, 14th Australasian Fluid Mechanics Conference, Adelaide University, Adelaide,

Australia.

Hottel, H. C. and Sarofim, A. F., 1967, Radiative Transfer, (McGraw-Hill, New York).

Howell, J. R., and Perlmutter, M., 1964, Monte Carlo Solution of Thermal Transfer

through Radiant Media between Gray Walls, J. Heat Transfer, 86, 116.

Hussein, H. J., Capp, S. P. and George, W. K., 1994, Velocity Measurements in a High

Peynolds-Number, Momentum-Conserving, axisymmtric, Turbulent Jet, J. Fluid Mech.,

Vol. 258, pp. 31-75.

Huynh, S., McIntosh M. and Huynh D., 2003, Process Design Considerations for MTE

Plant, Proceedings of the Tenth CRC Annual Conference, 12-13 June, Swinburne

University of technology, Melbourne, Australia, 31-36.

Issac, K. M. & Jakubowski, A. K., 1985, Experimental study of the interaction of

multiple jets with a cross -flow, AIAA J., 23, 1679-1683.

Issac, K. M. & Schetz, J. A., 1982, Analysis of multiple jets in cross-flow, ASME J.

Fluids Eng., 104, 489-492.

Ivanov, Y. V., Laats. M. K. and Frishman, F. A.,1970, Dispersion of a heavy admixture

in a two-phaser jet, Inghenermo-fizicheskii Zhurnal, 18, 538.

Jalaluddin, A. S., 1991, Estimation of Kinetic Parameters for Char Oxidation, Fuel,

Vol. 71, pp 311-317.

Johnson, T. R., Morley W J., Thomas GR., 1976, Firing of Raw Brown coal using Swirl

Burners in a Power Station Boiler, 4th Nat. Conf. On Chem. Eng., I E Aust, Adelaide.

191


Johnston, J. P. and Khan, Z., 1997, The origins of the dominant vortex from a pitched

and skewed jet, Intl Conf. On Fluid Engng, Paper 1201 (session keynote lecture), JSME,

Tokyo, 13-16 July.

Kamotani, Y., and Greber, I., 1974, Experiments on confined Turbulent Jets in cross-

flow, NASA, CR-2392.

Kj a&&ldman,,L. and Jacobson, T., 1992, A PHOENICS based Computational

Environment for Boiler Furnace Simulations, 5th Int. PHEONICS User Conference.

Kolmogorov, A. N., 1942, Equations of Turbulent Motions of an Incompressible Fluid,

Izv. Akad. Nauk. SSR, Seria fizicheska Vi., No. 1-2, p. 56-58. (English Translation:

Imperial College, Mech. Eng.Rept. ON/6,1968).

Laats, M. K. and Frishman F. A., 1970, Scattering of an inert admixture of different

grain size in a two-phase axisymmetric jet, Heat Transfer Soviet Research 2, 7-11.

Laats, M. K., 1966, Experimental study of the dynamics of an air dust jet, Inghenermo-

Fizicheskii Zhurnal 10, 11-15.

Lendt, B., 1991, Numerische Berechnung der Stickoxidkonzentration in

Kohlenstaubflammen, VDI-Fortschritt-Berichte Reihe 6, Nr.254, Dusseldorf, Germany.

Leschziner, M. A., 1980, Practical Evaluation of Three Finite Difference Schemes for

the Computation of Steady State Recirculating Flows, Comput. Meth. Appl. mech. Eng.,

23, 293-312.

Lester, L. Yuan, Robert, L. Street & Joel, H. Ferziger., 1999, Large-eddy simulations of

a round jet in cross- flow, J. Fluid Mech., 379, 71-104.

Lim, T.,T., New, T. H. & Luo, S.C., 2001, On the development of large-scale structure

of a jet normal to a crossflow, Physics of Fluid, 13, 3, 770-775.

192


Lockwood, F. C. and Shah, N. G., 1981, A New Radiation Solution Method for

Incorporation in General Combustion Prediction Procedures, Eighteenth Symposium

(International) on Combustion, The Combustion Institute, Pittsburgh, 1405-1414.

Lockwood, F.C., Salooja, A. P. and Syed, S. A., 1980, A Prediction Method for Coal-

Fired Furnaces, Combustion and Flame, 38, 1-15.

Lockwood, F. C., Mahmud, T. and Mohammed, A. Y., 1998, Simulation of pulverized

coal test furnace performance, Fuel, Vol. 77, No 12, pp 1329-1337.

Magnussen, B. F., 1989, Modelling of NOx and Soot Formation by the Eddy Dissipation

Concept, First Topic Oriented Technical Meeting- International Flame Research

Foundation, The Netherlands.

Makihata, T. & Miyai, Y., 1983, Prediction of the trajectory of triple jets in a uniform

cross -flow, ASME J. Fluids Eng., 105, 91-97.

Mann, A. P., Kent, J. H., 1994, A Computational Study of Heterogeneous Char

Reactions in a Full-Scale Furnace, Combustion and Flame, 99, 147-54.

Marsters, G. F.,1977, Interaction of Two Plane Parallel Jets, AIAA J., 15, 1756-1762.

Massey, B., 1998, Mechanics of Fluids, Seventh Edition, Stanley Thornes (Publishers)

Ltd, Cheltenham United Kingdom.

Matsumoto, R., Kymoto, K., Tsuchimoto, N., 1973, A Study of Doubble Concentric

Jets, Bul. J. JSME, 16, 529.

McComb, W. D. and Salih S. M., 1978, The calculation of three dimensional turbulent

flow jets, Proceedings of the symposium on Turbulent Shear Flows, Penn. State

University.

193


McIntosh, M. J., Ellul, W. M. J., Norman J. A., 1985, Measurements in a Burner Jet of

Yallourn W unit 4 boiler, Research and Development Department, State Electricity

Commission of Victoria.

Melville, W. K. and Bray K. N. C., 1979, A model of the two-phase turbulent jet, In .J.

Heat and Mass Trans., 22, 647.

Miller, D. R., Comings, E. W., 1960, Force Momentum field in a Dual Jet Flow, J.

Fluid Mechanics,7, 237-256.

Mitchell, R. E., and McLean, J. W., 1982, Nineteenth Symposium (International) on

Combustion, The Combustion Institute, Pittsburgh, p. 1113.

Modaress, D., Tan, H. & Elghobashi, S., 1984, Two-component LDA measurement in a

two-phase turbulent jet, AIAA J, 22, 5, 624-630.

Mostafa, A. A., Mongia H. C., McDonell V. G. & Samuelsen G. S., 1990, An

experimental and numerical study of particle-laden coaxial jet flows, Int. J. Heat and

Fluid Flow, Vol. 11, No.2.

New, T. H., Lim T. T. & Luo S.C., 2003, Elliptic jets in cross-flow, J. Fluid Mech., 494,

119-140.

Park, C. J., & Chen L. D., 1989, Experimental investigation of confined turbulent jets

Part-1, Single-phase data, AIAA J., 27, 11, 1506-1509.

Park, C. J., & Chen L. D., 1989, Experimental investigation of confined turbulent jets

Part 2, particle-laden flow data, AIAA J., 27,11, 1511-1516.

Patankar, Susas V., 1983, Numerical Heat Transfer and Fluid Flow, hemisphere

Publishing Corporation.

Patankar, Susas V., 2001, CFD Short Course-Lecture Notes, Victoria University of

Technology.

194


Perry, J. H., and Yan, X. H., 1992, Isothermal Two-phase Development from a Model

Slot- Burner, Eleventh Australasian Fluid Mechanics Conference, University of

Tasmania.

Perry, J.H., 1981, Aerodynamics of Burner Jets Designed for Brown-Coal Fired Boilers

– Part- I Literature Survey, SECV Engineering Research Division Report No G0/82/53.

Perry, J.H., Hausler T., 1982, Aerodynamics of burner jets designed for brown-coal

fired boilers–part III, SECV Engineering Research Division Report No GO/83/57,

December.

Perry, J.H., Hausler T., 1984, Aerodynamics of Burner Jets Designed for Brown-Coal

Fired Boilers–Part-IV, SECV Engineering Research Division Report No ND/84/004,

March.

Perry, J.H., Somogyi M., Rezk M., 1986, Aerodynamics of Burner Jets Designed for

Brown-Coal Fired Boilers–Part-V, SECV Engineering Research Division Report No

ND/86/029, August.

Peterson, S. D., & Plesniak M. W., 2004, Evolution of jets emanating from short holes

into cross-flow, J. Fluid Mech., 503, 2004, 57-91.

Pleasance, G., 1980, Self-Induced Recirculation Characteristics of a Swirl Burner, 7th

Aust. Hyd. and Fluid Mech. Conf..

Popper, J., Abvaf, N. and Hetsroni, G., 1974, Velocity measurements in a two-phase

turbulent jet, Int. J. Multiphase Flow 1, 715-726.

Prandtl, L., 1945, U&&berein neues Formelsystem fu&&rdie ausgebildete Turbulenz, Nacr.

Akad. Wiss. G o&&ttingen, Math-Phys, KI., 6-19.

Rhie, C. M., and Chow, W. L., 1983. Numerical Study of Flow Past an Airfoil with

Trailing Edge Separation, AIAA Journal, Vol. 21, No. 11, 1525-1532.

195


Rhie, C. M., 1981., A Numerical Study of the flow Past An Isolated Airfoil with

Separation, Phd Thesis, Department of Mechanical and Industrial Engineering,

University of Illinois at Urbana-champaign.

Ribero, M. M., and Whitelaw, J. H., 1980, Co-Axial Round Jets With and Without Swirl,

Journal of Fluid Mechanics, 96, 769-795.

Ricou, F. P., Spalding, D. V., 1961, Measurements of Entrainment by Axisymmetric

Turbulent Jets, J. Fluid Mech., 11, 21.

Rodi, W., 1984., Turbulence Models and Their Applications in Hydraulics- A State of

the Art Review, International Association for Hydraulic Research, The Netherlands.

Runchal, A. K., and Wolfshtein, M.,1969, Numerical Integration Procedure for the

Steady State Navier-Stokes Equations, J. Mech. Eng. Sci., 11, 445.

Savory, E. & Toy, N., 1991, Real time video analysis of twin jets in a cross- flow,

ASME J. Fluids Eng., 113, 68-72, 1991.

Schefer, R. W., Hartmann, V., and Dibble, R. W., 1986, Conditional sampling of

velocity in a turbulent nonpremixed propane jet, AIAA Journal, vol.25, No.10, pp.1315-

1329.

Sherif, S. A., & Pletcher, R. H., 1989, Measurements in the flow and turbulence

characteristics of round jets in cross- flow, ASME J. Fluids Eng., 111, 165-171.

Smoot, L. D., Allred, L. D., 1975, Particle Gas Mixing Effects in confined non-parallel

co-axial Jets, AIAA J. 13, 721.

Spalding, D.B., 1972, A novel Finite Difference Formulation for differential

Expressions Involving both First and Second derivatives, Int. J. Num. Methods Eng., 4,

551.

Stambuleanu, A., 1976, Flame Combustion Processes in Industry, ABACUS Press, UK.

196


Stanmore, B.R. and Visona S. P., 1998, The Contribution to Char Burnout from

Gasification by H2O and CO2 during Pulverized-Coal Flame Combustion, Combustion

and Flame, 113:274-276.

Sykes R. I., Lewellyn, W. S. & Parker, S. F., 1986, On the vorticity dynamics of a

turbulent jet in a cross- flow, J. Fluid Mech., 168, 393-413.

Tanaka, E., 1970, The Interface of Two-Dimensional Parallel Jets, Bulletin of Journal

of ASME, 13, 56.

Ubhayakar S. J., Stickler D. B., Von Rosenburg C. W., Ganon R.E., 1976, Rapid

devolatilization of pulverized coal in hot combustion gases, 16th Symposium

(International) on Combustion, The Combustion Institute, pp. 426.

Walker, P.L., Rusinko, F. and Austin, L. G., 1959, Advan. Catal., 11,133-221.

Wall, T. F., Nguyen, H., Subramanian, V., Mai-Viet, T., 1979, Entrainment of Single,

Double Concentric and particle laden Jets Close to the nozzle, Nat. Conf. on Chem.

Eng., IE, Aust., 84-88.

Wall, T. F., Nguyen, H., Subramanian, V., Mai-Viet, T., Howley, P., 1980, Direct

Measurement of entrainment by single and double Concentric Jets in the Regions of

Transition and Flow Establishment, Trans. I. Chem. Eng., 58, 4, 237.

Wheeler R., Tiu C., 2002, A fundamental study of a Continuous Mechanical and

Thermal Expression (MTE) Process for Lignite, Proceedings of the Ninth CRC Annual

Conference, 27-28 June, Monash University, Melbourne, Australia, 149-52.

Wicke, E., 1955, Fifth Symposium on Combustion, The Combustion Institute,

Pittsburgh, p.245.

Winter, F., Prah, M. E., Hofbauer. H., 1997, Temperatures in a Fuel Particle Burning in

a Fluidized Bed: The Effect of Drying, Devolatilization and Char Combustion,

Combustion and Flame, 108, 302-314.

197


Wirtz, S., 1989, Mathematische Modellierung der Kohlenstaubverbrennung,

Dissertation Universitat Bochum.

Yan, H., Perry, J.H., 1994, Two–Phase Flow Development in Slot-Burners - Part 2

Detailed Flow Measurement and Numerical Model Validation, ESAA Report No.

ES/94/01.

Yan, H., Perry, J.H., 1994, Two–Phase Flow Development in Slot-Burners - Part 1

Flow Visualization Study, ESAA Report No. ES/93/01.

Yeh, Y. and Cummings H.Z., 1964, Localized Fluid Flow Measurements with an He-Ne

Laser Spectrometer, Appl. Phys., Lett 4, 176-178, 1964.

Zhang, Y.P., Jun Mou, and Wei, Biao Fu., 1990, Method for Estimating Final Volatile

Yield of Pulverized Coal Devolatilization, FUEL, Vol.69, pp 401-402.

198

NOTE


Appendix-I

Appendix I

Appendix I – Reynolds Stresses in a Plane Jet Simplifications Under a Plane Jet Assumption The Reynolds Stress transport equation is shown in (A.1). Under the simplifications

provided by assuming a two-dimensional plane jet situation, the stress transport

equation reduces to (A.2) to (A.5) for the individual stress components

General Stress Transport Equation

⎥⎥⎦

⎤

⎢⎢⎣

⎡

⎟⎟⎠

⎞⎜⎜⎝

⎛

∂

∂+

∂

∂+

∂

∂−++

∂∂

−

⎟⎟⎠

⎞⎜⎜⎝

⎛

∂∂∂∂

+⎟⎟⎠

⎞⎜⎜⎝

⎛

∂

∂

∂∂

+∂

∂

∂∂

−∂∂∂∂

−⎪⎭

⎪⎬⎫

⎪⎩

⎪⎨⎧

∂∂

+∂∂

=

k

ji

i

jk

j

jikj

iij

kkii

j

ki

ik

i

j

j

i

k

j

j

i

jj

ki

j

kji

j

ijk

ki

xu

uxu

uxuupupu

uuux

xxuup

xu

xu

xu

xu

xxuu

xUuu

xUuu

DtuDu

µδρ

δρ

ρµµ2

(A.1) uu Normal Stress

[ ]22

121

111121

21 222 uu

xxup

xu

xu

xUuu

DtDu

jjj ∂∂

−∂∂

+∂∂

∂∂

−∂∂

−=ρ

µ (A.2)

vv Normal Stress

⎥⎦

⎤⎢⎣

⎡+

∂∂

−∂∂

+∂∂

∂∂

−=ρρ

µ 232

22

2222

2 222 puuxx

upxu

xu

DtDu

jj

(A.3)

ww Normal Stress

[ ]22

323

3332

3 22 uu

xxup

xu

xu

DtDu

jj ∂∂

−∂∂

+∂∂

∂∂

−=ρ

µ (A.4)

uv Shear Stress

⎥⎦

⎤⎢⎣

⎡+

∂∂

−⎟⎟⎠

⎞⎜⎜⎝

⎛

∂∂

+∂∂

+∂∂

∂∂

−∂∂

−=ρρρ

µ puuuxx

upxup

xu

xu

xUu

DtuDu

jj1

221

21

2

2

121

2

122

21 2

(A.5)

200

Appendix I

Explanation of each term on the right hand side of (A.1) is as follows, from left to right.

• Production of stress by the working of components of the stress tensor against

the main strain tensor

• Dissipation of stress due to viscous action on the small scale turbulent motion

• Diffusion of stress arising from three mechanisms

ο Turbulent velocity fluctuations

ο Turbulent pressure fluctuations

ο Molecular transport (neglected in fully turbulent flow)

• Pressure-Strain interaction which acts both to promote reversion to isotropy of

the stress field and to smear out the effects of stress generation over other

components of the stress tensor, this is a re-distributive process. This term

makes no direct contribution to the level of turbulence energy; it merely serves

to alter the relative levels of the fluctuations in three mutually orthogonal

directions.

• In the shear stress equations the pressure-strain correlation will act as a source or

sink of that stress component

The following can be noted for the two-dimensional plane jet case;

• All the energy extracted from the mean flow enters turbulence in the stream-

wise normal stress. There are no mean flow terms in the vv and ww equations.

• The levels of vv and ww can only be maintained through the re-distributive

action of the pressure fluctuations.

• Diffusion by pressure fluctuations is absent from the equations for uu and ww .

201

Appendix I

• The rate of creation of shear stress is strongly dependent on vv , the mean square

of the fluctuating velocity in the direction of the momentum flux (U1-mean

velocity in the stream-wise direction).

Turbulence is maintained in the presence of shear by the triangular interaction indicated

in figure A.1. Cross-stream velocity fluctuations, uv transfer momentum and its

momentum directly intensifies stream-wise fluctuations,uu ; finally vv is replenished by

energy transfer from uu and the process repeats itself.

vv

uv

uu

Figure A 1. Interaction of stress components in a 2D plane jet

202

NOTE


List of personal publications


Journal Publications

Ahmed, S., Naser, J., 2004, Numerical Investigation of Aerodynamics and Combustion

of Conventional Brown coal and MTE coal in tangentially-fired furnace, Submitted in

Fuel.

Ahmed, S., Hart, J., Naser, J., 2004, CFD Modelling of Rectangular Slot-Burners by

varying Jet Velocity Ratio, Submitted in Applied Mathematical Modelling.

Conferences Proceedings

Ahmed, S., Naser, J., Nikolov, J., Solnordal, C., Yang, W., Hart, J., 2004, Experimental

Investigation of a Rectangular Slot-Burner in the presence of Cross-flow for different

jet velocity ratios, 13-17th December, The University of Sydney, Australia.

Ahmed, S., Naser, J., Hart, J., 2004, The Validation of Numerical Investigation of a

Rectangular Slot-Burner in the presence of Cross-flow, Eleventh CRC Annual

Conference, 8-9 July, Monash University, Melbourne, Australia, pp. 205-210.

Ahmed, S., J Hart, J. Naser, 2003, The effect of jet velocity ratio on aerodynamics of

rectangular slot-burners in tangentially-fired furnaces, Third International Conference

on CFD in the Minerals & Process Industries, 10th-12th December, Melbourne,

Australia, pp 41-46.

Ahmed, S., Naser, J., 2003, Combustion of MTE Coal in a Full Scale Industrial

Furnace., 12th International Conference on Coal Science., 2nd-6th November, Cairns,

Australia.

Ahmed, S., J. Naser, J. Hart, 2003, The Comparison of Exit Velocity Profiles on

Aerodynamics of a Rectangular Slot-Burner in Tangentially-Fired Furnaces, Tenth

CRC Annual Conference, 12th-13th June, Swinburne University of Technology,

Melbourne, Australia, pp 233-236.

204


Ahmed, S., Naser, J., 2002, Simulation of the Coal Combustion of Yallourn Stage-2

Furnace, Proceedings of the Ninth CRC Annual Conference, 27-28 June, Monash

University, Melbourne, Australia, pp. 221-225.

Ahmed, S., Naser, J., 2002, Modelling of Coal Combustion in YallournStage-2 Full

Scale Industrial Furnace., 2002 Australian Symposium on Combustion and The

Seventh Australian Flame Days, 7-8 February, Adelaide University, Adelaide,

Australia, pp.150-154.

Presentations

Presented paper in 5th European Conference on Coal Research and its Application held

at 6th –8th September, Edinburgh, Scotland, 2004.

205

Documents

Aerodynamics of rectangular slot-burners and combustion … · Aerodynamics of Rectangular Slot-Burners and Combustion in Tangentially-fired Furnace . A Thesis Submitted for the Degree