You are on page 1of 113

Investigation of Fuel Nozzle Technologies to Reduce Gas

Turbine Emissions

by

Roger Neil Antony Francis

A thesis submitted in conformity with the requirements


for the degree of Masters of Applied Science
Graduate Department of Aerospace Studies
University of Toronto

Copyright c 2015 by Roger Neil Antony Francis


Abstract

Investigation of Fuel Nozzle Technologies to Reduce Gas Turbine Emissions

Roger Neil Antony Francis

Masters of Applied Science

Graduate Department of Aerospace Studies

University of Toronto

2015

With increasing requirements for reduced emissions from future gas turbines, a multitude of

research is being conducted into fuel nozzles by gas turbine manufacturers. This thesis focuses

on the development of a novel spill return nozzle, to improve combustion eciency at starting

and low power conditions -where combustion eciency is often the poorest. The spill return

nozzle has the advantage of being able to improve atomization performance and reduce internal

coking potential, all while being a simple and durable design. The spill return nozzle tech-

nology was subsequently applied to a design for an existing small gas turbine combustor, and

its improvements over the existing nozzle were demonstrated. The proposed design was also

extended to experimental testing in a simplified form. CAD drawings of the components for

testing were made, and prototypes were built in plastic using a high accuracy 3D printer. Future

work involves conducting experimental tests to validate results.

ii
Acknowledgements

I have been truly blessed to be able to meet and work with an enormous number of people,

who have assisted and guided me over the past two years as I worked on this thesis. Firstly, I

would like to thank Professor Sam Sampath, for his patience and technical guidance through-

out the project; for always trying to ensure that the project was the best it could be. I would

also like to thank my co-supervisor Professor Clinton P.T. Groth, for all his help and guidance.

My appreciation and thanks also goes out to additional members of the research assessment

committee: Professor Omer L. Gulder, Professor Adam M. Steinberg, and Professor James J.

Gottlieb.

In addition, I would also like to thank Pratt and Whitney Canada, specifically Dr. Jian-

Ming Zhou, Kian McCaldon, Enzo Macchia, Bhavan B. Patel, Jerey Verhiel, Dr. John Hu

and Dr. Hayley Ozem, for always being available, their invaluable advice, and providing any

information needed to help make this project successful. Finally, Im grateful to God, family,

and friends, for being the moral support that was so needed to ensure the success of this project;

for truly showing that life is a lot easier when you are part of a network of friends and family,

a community.

iv
Contents

Abstract ii

Acknowledgements iv

Contents v

List of Tables viii

List of Figures ix

List of Symbols xii

1 Introduction 1

1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.1.1 Semi-empirical Nozzle Design Methodologies . . . . . . . . . . . . . . . . . 4

1.2 Scope of Research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

2 Gas Turbine Technology 6

2.1 Types of Gas Turbines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

2.2 Combustion Chamber . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

2.3 Fuel Nozzle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

3 Liquid Atomization Fundamentals 13

3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

3.2 Eect of Liquid Physical Properties . . . . . . . . . . . . . . . . . . . . . . . . . . 14

v
3.3 Classical Mechanism of Atomization . . . . . . . . . . . . . . . . . . . . . . . . . 16

3.3.1 Breakup of Liquid Jets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

3.3.2 Breakup of Liquid Sheets . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

3.4 Prompt Mechanism of Atomization . . . . . . . . . . . . . . . . . . . . . . . . . . 18

3.5 Breakup of Drops . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

3.6 Liquid Atomization Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . 20

3.6.1 Discharge Coecient . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

3.6.2 Flow Number . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

3.6.3 Spray Cone Angle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

3.6.4 Patternation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

3.6.5 Droplet Size . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

3.7 Fuel Nozzle Coking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

3.7.1 Types of Coking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

3.7.2 Eect of Fuel Temperature on Coking . . . . . . . . . . . . . . . . . . . . 24

4 Fuel Nozzle Technology and Performance Characteristics 26

4.1 Fuel Nozzle Technologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

4.1.1 Simplex Nozzle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

4.1.2 Spill Return Nozzle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

4.1.3 Duplex Nozzle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

4.1.4 Air Assist Nozzle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30

4.1.5 Airblast Nozzle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

4.1.6 Hybrid Nozzle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34

4.2 Nozzle Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

4.2.1 Combustion Eciency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

4.2.2 Pattern Factor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

4.2.3 UHC and CO emissions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38

4.3 Fuel Nozzle Coking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

4.3.1 Eects of Nozzle Coking . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

vi
4.3.2 Methods of Reducing Coking . . . . . . . . . . . . . . . . . . . . . . . . . 41

4.4 Spill Return Nozzle Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

5 Nozzle Design Methodology and Requirements 47

5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47

5.2 Nozzle Feature Design Methodology . . . . . . . . . . . . . . . . . . . . . . . . . 47

5.2.1 Nozzle Design Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . 47

5.2.2 Spill Return Nozzle Design Methodology . . . . . . . . . . . . . . . . . . . 49

5.2.3 Fundamental Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49

5.2.4 Design Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50

5.2.5 Spray Uniformity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

5.2.6 Nozzle Design Check . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52

5.2.7 Methodology Validation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52

5.2.8 Nozzle Spill Line Sizing . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55

5.2.9 Nozzle SMD Calculation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56

5.3 Nozzle Control Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57

5.4 Model Combustor Test Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . 59

6 Spill Return Nozzle Design For Model Combustor 61

6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61

6.2 Nozzle Sizing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61

6.2.1 Eect of Spill Flow on Nozzle Design Choice . . . . . . . . . . . . . . . . 62

6.2.2 Spray Uniformity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64

6.2.3 Spill Line Geometries and its Size Eects . . . . . . . . . . . . . . . . . . 66

6.2.4 Nozzle Design Check . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69

6.3 Nozzle Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70

6.3.1 Spill Percentage on SMD . . . . . . . . . . . . . . . . . . . . . . . . . . . 70

6.3.2 Spill Percentage on Spray Cone Angle . . . . . . . . . . . . . . . . . . . . 70

6.4 Spilled Fuel Control Mechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71

6.4.1 Spill Return Nozzle Supply Setup . . . . . . . . . . . . . . . . . . . . . . . 74

vii
6.5 Proposed Nozzle Design Concept . . . . . . . . . . . . . . . . . . . . . . . . . . . 76

7 Next Steps and Future Work 82

7.1 CAD Drawing of the Spill Nozzle . . . . . . . . . . . . . . . . . . . . . . . . . . . 82

7.1.1 3D Printing of Spill Nozzle . . . . . . . . . . . . . . . . . . . . . . . . . . 84

7.2 Planned Experimental Testing of Spill Nozzle Design . . . . . . . . . . . . . . . . 84

8 Conclusions 86

References 89

A Equations and Experimental Nozzle CAD Drawing A-2

viii
List of Tables

4.1 Nozzle material coke deposition test . . . . . . . . . . . . . . . . . . . . . . . . . 44

5.1 Model combustor test conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . 60

6.1 Sizes of nozzle features(mm), for a nozzle with a normalized FN of 43% . . . . . 65

6.2 Spill nozzle design features against design practices from experiments . . . . . . . 69

ix
List of Figures

1.1 Description of a traditional combustor setup and the Brayton cycle . . . . . . . . 2

2.1 Setup of a generic turbo-jet gas turbine . . . . . . . . . . . . . . . . . . . . . . . 7

2.2 Setup of a generic turbo-prop gas turbine . . . . . . . . . . . . . . . . . . . . . . 8

3.1 Eect of temperature on surface tension and temperature for dierent fuels . . . 15

3.2 Eect of jet velocity on type of atomization . . . . . . . . . . . . . . . . . . . . . 18

3.3 Liquid sheet breakup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

3.4 Droplet breakup with the Weber number . . . . . . . . . . . . . . . . . . . . . . . 20

3.5 Description of the spray cone angle . . . . . . . . . . . . . . . . . . . . . . . . . 23

3.6 Eect of fuel Temperature on coke deposition rate . . . . . . . . . . . . . . . . . 25

4.1 Description of the two dierent simplex nozzle types . . . . . . . . . . . . . . . . 27

4.2 Spill return nozzle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

4.3 Description of a duplex nozzle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30

4.4 Description of air-assist nozzles . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

4.5 Description of an airblast nozzle and its features . . . . . . . . . . . . . . . . . . 33

4.6 Hybrid nozzle with a simplex primary circuit, and an airblast secondary circuit . 34

4.7 Proximity of a fuel nozzle to the combustor flame . . . . . . . . . . . . . . . . . 39

4.8 Eect of coking on spray patternation performance . . . . . . . . . . . . . . . . . 40

4.9 Implementation of dierent methods to reduce internal coking . . . . . . . . . . . 42

4.10 Fuel nozzle design conventions to reduce internal fuel temperatures . . . . . . . . 43

4.11 Description of a spill return nozzle . . . . . . . . . . . . . . . . . . . . . . . . . . 46

x
5.1 Description of spill return design parameters . . . . . . . . . . . . . . . . . . . . . 48

5.2 Spill nozzle design process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50

5.3 Eect of changing spill ratio on nozzle performance . . . . . . . . . . . . . . . . . 53

5.4 Comparison of D0 values calculated from the extended design methodology, against

experimental data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54

5.5 Normalized eect of geometric coecient (Z) and spill percentage, on the aircore

size . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55

5.6 Comparison of Normalized SMD calculated by two methods against experimental

data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56

5.7 General spill nozzle control plot . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57

5.8 Eect of spill passage size on spill line control . . . . . . . . . . . . . . . . . . . . 58

6.1 Normalized total fuel spilled (%) with spill percentage for idle conditions . . . . . 62

6.2 Percentage increase in normalized D0 and spill percentage, with increasing nor-

malized nozzle FN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63

6.3 Eect of circumferential uniformity(I) on the degree of divergence(C), for various

inlet passages. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64

6.4 Description of nozzle features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65

6.5 Dierent methods of spill line implementation in a nozzle . . . . . . . . . . . . . 67

6.6 Size of the spill passage with spill percentage, for dierent spill geometries . . . . 68

6.7 Eect of spill percentage on droplet size for various power conditions . . . . . . . 71

6.8 Eect of spill percentage on normalized droplet size(%) and normalized spray

cone angle(%), for the 7% power condition . . . . . . . . . . . . . . . . . . . . . . 72

6.9 Normalized control of spill return nozzle for low power conditions . . . . . . . . . 73

6.10 Normalized control of both circuits from starting and idle power conditions . . . 74

6.11 Implementation of the back-end fuel supply system for the spill return nozzle . . 75

6.12 Implementation of spill return nozzle with spilled fuel sent to secondary circuit . 76

6.13 Hybrid nozzle with central spill line . . . . . . . . . . . . . . . . . . . . . . . . . . 77

6.14 Hybrid nozzle with o-central spill line . . . . . . . . . . . . . . . . . . . . . . . . 78

xi
6.15 Hybrid nozzle with annular spill line . . . . . . . . . . . . . . . . . . . . . . . . . 79

7.1 3D Model of the Fuel Nozzle Showing the Various Parts . . . . . . . . . . . . . . 83

7.2 Description of the spill return nozzles for experimental testing . . . . . . . . . . . 84

A.1 Dimensions and tolerance for the front nozzle component . . . . . . . . . . . . . . A-3

A.2 Dimensions and tolerance for the orifice component . . . . . . . . . . . . . . . . A-4

A.3 Dimensions and tolerance for the core component . . . . . . . . . . . . . . . . . A-5

A.4 Dimensions and tolerance for the rear component . . . . . . . . . . . . . . . . . . A-6

xii
List of Symbols

Variables and Abbreviations


A Area (m2 )
AVGAS Aviation Gasoline
AVTUR Aviation Turbine Fuel
AFR Air/Fuel Ratio
APU Axillary Power Unit
B Constant
C Degree of Divergence
CAD Computer Aided Design
CO Carbon Mono-oxide
CO2 Carbon Dioxide
CFD Computational Fluid Dynamics
D Diameter (mm)
e Spill Ratio ( G
Gt
a
)
EI Emissions Index (g/kgf uel )
EPA Environmental Protection Agency
FAR Fuel/Air Ratio
FN Flow Number ( ppphP )
Ga Total fuel atomized ( kg
s )

Gs Total fuel spilled ( kg


s )

Gt Total fuel input ( kg


s )

g Gravity (m/s2 )
I Circumferential Uniformity (%)
ICAO International Civil Aviation Organization
IPCC Intergovernmental Panel on Climate Change

xiii
K Geometric Coecient
L Length (mm)
n Number of Inlets to Swirl Chamber
N Ox Oxides of Nitrogen (N O and N O2 )
Oh Ohnesorge number
P Pressure (M P a)
pph Pounds Per Hour
P Change in Pressure (M P a)
P/P Combustor Pressure Drop
P si Pounds Per Square Inch
Re Reynolds Number
r0 Orifice Radius
rr0 Air core radius at the orifice
RQL Rich Quench Lean
r0
S Dimensionless aircore ratio ( rr0 )
SLA Stereolithography
SMD Sauter Mean Diameter (m)
T Temperature (K)
U Axial Velocity (m/s)
UHC Unburned Hydrocarbons
v Tangential Velocity (m/s)
We Weber Number
X Ratio of Air Core to Discharge Surface Area (mm2 /mm2 )

Greek
Eciency of Filling
Density (kg/m3 )
Surface Tension (kg/s2 )
Spray Cone Angle (deg)
Kinematic Viscosity (m2 /s)
Discharge Coecients

xiv
Subscripts
0 Orifice
f Fuel
g Gas
l Liquid Fuel
p Inlet Port
ref Reference Parameter
Stoich Stoichiometric Conditions
s Swirl Chamber
sp Spill Line
t Nozzle Trumpet

xv
Chapter 1

Introduction

Modern gas turbines have evolved significantly since their inception by Sir Frank Whittle. While

still following the Brayton cycle, modern gas turbine designs have been heavily influenced by

three main factors: performance, reliability and safety. This aspect is clearly visible when

one studies the dierences between gas turbines that have been developed for small and large

thrust applications. With their high propulsive and thermal eciencies, gas turbines have found

applications primarily in aerospace, with usage to a smaller degree in powerplants, ships and

military tanks.

Despite the variability in application, all gas turbines contain a compressor, combustor and

turbine, as shown in Figure 1.1a. Figure 1.1b shows how the temperature varies throughout the

gas turbine for an ideal Brayton engine cycle. The gas turbine goes through a number of stages

where air is taken in and compressed, ignited with fuel in the combustion chamber, which is

then subsequently used to provide energy. The dierent stages in a combustor, and how they

work with respect to the Brayton cycle, are shown Figure 1.1.

1.1 Motivation

Studies by the Intergovernmental Panel on Climate Change (IPCC) have found that aircraft

contribute as much as 3.5% of the total anthropogenic radiative force caused by all human

activity, as of 2005 [3]. Given the forecasted growth of the aerospace sector, the governing

1
Chapter 1. Introduction 2

Figure 1.1: Description of: a)Traditional combustor setup b) Brayton cycle (Nguyen [43])

body that regulates emissions for aerospace - International Civil Aviation Organization (ICAO),

has implemented a set of regulations aimed at limiting the impact of the aerospace sector on

the earths climate, while allowing it to continue to be a key vehicle for economic growth.

While ICAO regulates large engines for a number of dierent pollutants, small engines are only

regulated for smoke, due to their negligible contributions in other sectors [54]. Nevertheless,

factors such as the implementation of landing fee surcharges for non-complying engines based

on local environmental laws at certain airports, and the fact that regulations in the future

could be extended to small engines, strongly motivate small gas turbine manufacturers to keep

emissions low. The dierent emissions that are currently regulated by ICAO for aircraft are:

Carbon Monoxide (CO)


Nitrogen Oxides (NOx )
Unburned Hydrocarbons (UHC)
Particulates
Combustion Noise

Improvements to a number of dierent factors such as air trac management, engine ef-

ficiency and aircraft design, are all needed to help achieve lower emissions levels in the fu-

ture. Modern demands for improvement in combustor performance, has generally been achieved

through a number of technologies such as advanced materials, new cooling technologies, higher

engine pressure ratios, and fuel nozzles. For gas turbines specifically, the emissions of NOx , CO

and UHC are strongly dependant on how the fuel and air in the combustion chamber are mixed,
Chapter 1. Introduction 3

as well as the temperature and pressure inside the combustor [54]. One way to achieve these

improvements have been through increasing the overall engine pressure ratio, leading to higher

temperatures and pressures inside the combustor. The only issue with doing this is that it leads

to higher levels of NOx emissions, since it is strongly correlated to temperature [29].

Another method has been to improve the process by which the fuel and air interact inside the

combustor, with the end goal of ensuring better mixing. The nozzle has been able to provide large

benefits due to the fact that it assists the combustion process by transforming the bulk liquid

fuel into a form that is much easier to burn. With improved methods of fuel-air ratio control

and mixing, combustor performance has greatly improved. For small gas turbine combustors,

the challenge with fuel nozzle design and implementation is more dicult, as the dispersion and

mixing of the fuel and air needs to be conducted within a smaller volume. The improvement in

fuel nozzle performance is often characterized by advances in properties such as turn-down ratio,

spray cone angle, droplet size and nozzle pressure drop. Circumferential uniformity is another

issue of importance, as it is vital that local rich and lean zones are prevented. Rich zones are

not desirable because they produce local hot spots that can reduce turbine and combustor life,

while increasing the production of soot. On the other hand, lean zones (especially at low power

conditions) can result in inecient burning, resulting in high UHCs and CO emissions. Another

aspect is that the performance of fuel nozzles inside a combustor can be quite dierent, to its

individual performance. The fuel nozzles not only need to work with each other to prevent hot

spots, but also with the internal aerodynamics such as the wall cooling and dilution jets inside

the combustor.

In addition to the above, fuel nozzles in small gas turbine combustors have a unique set of

challenges since they deal with small liquid flow rates. This often restricts the types of nozzles

that can be used (in comparison to larger combustors), and the size of nozzle itself. Size is often

an issue because the internal passages at certain locations can get quite small, becoming prone

to fouling issues from fuel contamination, gumming or carbonizing. Consequently, the method

of fuel injection and the type of fuel distribution selected are major design factors that are

determinative of the performance of small gas turbines. Thus, given the continually increasing

requirements for improved performance, it is vital that research is conducted on upcoming fuel
Chapter 1. Introduction 4

nozzle technologies that have the capability to provide the improvements in emissions and coking

performance, without giving up on both reliability and durability.

1.1.1 Semi-empirical Nozzle Design Methodologies

While we have a relatively strong understanding of the fundamentals of liquid fuel atomization

and spray formation, the same cannot be said for the design of fuel nozzles. While there are a

number of semi-empirical based methodologies by Bayvel [6] and Couto et al. [56] for design-

ing nozzles, there are often discrepancies between the dierent calculation procedures, and also

between the eventual experimental test and original design expectation. However, the more

important criteria when designing atomizers of practical applications is the simplicity of the cal-

culation method. Almost always, semi-empirical methods are used to obtain a fuel nozzle design

which is close to the desired performance point, and then either fine tuned through experimental

testing, or through computation fluid dynamic simulations - if it is considered accurate enough.

In addition to the aerodynamic requirements that are satisfied by the semi-empirical methods,

another aspect that aects the design of fuel nozzles, is the packaging of the nozzle for use in

a combustor. The final nozzle design needs to not only be aerodynamically optimized but be

inherently simple in design, in order to have good reliability and maintainability; Sometimes,

certain design features are either removed or slightly changed, in order to meet this mechanical

design requirement.

1.2 Scope of Research

The main scope of this project is to propose and develop a novel fuel nozzle design that can

help reduce emissions of CO, UHC, and PM, at starting and low power conditions. In order to

achieve this the nozzle needs to meet certain design criteria:

Consistent and low droplet sizes from idle and low power conditions.
Demonstrate good atomization over a wide liquid flow rate.
Achieve uniform circumferential fuel distribution for varying power conditions.
Improve mixing at idle and low power conditions.
Low susceptibility to coking
Chapter 1. Introduction 5

The first phase of the project involves studying dierent fuel nozzle technologies, and eval-

uating their performance in small gas turbine combustors; in particular for starting and low

power conditions. The next phase involves proposing a novel nozzle technology that is capable

of meeting the above criteria, while also showing improvements over an existing nozzle in a

model reverse flow combustor. This involves demonstrating its performance in terms of SM D,

spray cone angle, spray uniformity, and a robust control mechanism. The next phase involves

developing the selected nozzle technology for experimental testing. This involves developing

detailed CAD drawings of the nozzle technology of interest, and building the nozzles using

Stereolithography (SLA) to validate the prototypes.


Chapter 2

Gas Turbine Technology

Fuel nozzles have evolved to play a crucial role in the performance and durability of gas turbines.

As discussed in Chapter 1, It is widely accepted that the design and development of nozzles

be closely integrated with the design of the combustor itself; especially with the increasingly

stringent emissions regulations that are being put forward by ICAO. This is because the fuel

nozzle has a significant influence on a number of combustor characteristics such as starting

performance, exhaust emissions (CO, NOx and PM), hot end durability and combustion noise

[39]. Given this inter-dependency, when working on the analysis and design of fuel nozzles, it is

important to understand both the individual component (fuel nozzle) and system (combustor)

requirements, and how to achieve them.

2.1 Types of Gas Turbines

There are three types of aerospace gas turbines in use today. These three variants have come

across as a result of market demands that drive manufacturers. The three gas turbine types are:

Turbo-jet
Turbo-fan
Turbo-prop

All gas turbines generate thrust by providing a change in momentum to the air that enters

and leaves the gas turbine. The higher the dierence in momentum, the greater the thrust that

6
Chapter 2. Gas Turbine Technology 7

Figure 2.1: Setup of a turbo-jet gas turbine (Beechcraft [1])

the gas turbine produces. There are two approaches to generating this momentum change. The

first method involves taking a small amount of air, and then accelerating it to very high speeds

to create thrust. This method is generally adopted by pure turbo-jets, where the hot exhaust

gas from the combustor is used to provide all the thrust. The second method works by taking a

large mass of air, and accelerating it a small amount to achieve a large momentum change. The

approach is generally used by most commercial engines in use today, such as turbo-fans and

turbo-props. Both these combustors work by using a fan or propeller at the front of the engine

to accelerate the air. A major reason for taking this approach is that, a lower exit velocity

results in a higher propulsion eciency, which in turn leads to lower Specific Fuel Consumption

(SFC) [39].
The general setup of a turbo-fan is shown in Figure 2.1. The turbo-fan is essentially a

turbo-jet with a fan in front. Here, the combustor is primarily used to generate enough power

to turn the fan and compressor, with any remaining core energy used to provide thrust in the
Chapter 2. Gas Turbine Technology 8

Figure 2.2: Setup of a turbo-prop gas turbine (Beechcraft [1])

form of the hot exhaust. Unlike in a turbo-jet, most of the thrust is generated by the fan.

Figure 2.2 shows the second type of gas turbine that is in use. In a turbo-prop, instead of

driving a fan, the core is used to drive a propeller using a reduction gear. Turbo-prop aircraft

have high fuel eciencies and performance in the mid-range altitudes and mid-range speeds.

Because propellers become less ecient as the aircraft speed increases, they are generally used

for low speed aircraft that travel shorter distances. On the other hand, turbo-fans are used for

long distance, high speed aircraft, since they have better fuel eciency and thrust capabilities.

Higher speed aircraft like most commercial planes, use a high bypass ratio turbo-fan because

of their high fuel eciency and high thrust capability. While not always the case, turbo-props

often end up being smaller engines, in comparison to turbo-fans. This dierence in size often

results in dierent design approaches to design of the combustor.

One of the most significant dierences between the gas turbines shown in Figure 2.1 and

Figure 2.2, is the combustion chamber. Unlike in In Figure 2.1, in 2.2 we see that the airflow
Chapter 2. Gas Turbine Technology 9

changes direction in the combustion chamber (indicated in yellow). This is important, because

gas turbines with reverse-flow combustors have special design challenges, especially with respect

to fuel nozzles, that are not present in straight flow-through combustors.

2.2 Combustion Chamber

The combustor is a vital component of the gas turbine. Unlike automobiles, gas turbines have

a continuous flame inside the combustor, which is lit for as long as the engine is running. Once

ignited, the flame is maintained by constantly mixing fuel to the high pressure compressed

air from the compressor, using a fuel nozzle. The primary purposed of every fuel nozzle is to

atomize the fuel into small droplets, in order to speed up the mixing process of fuel and air. The

dierences between various fuel nozzle technologies lie in how exactly the droplets are produced.

The challenge with combustor design, lies with balancing the various competing factors of

CO, UHC, NOx and smoke emissions. Smoke is primarily produced as a result of uneven mixing

of the fuel and air in the primary zone [60], while the presence of CO and UHCs is an indication of

incomplete combustion. They are generally produced when poor fuel-air mixing produces local

regions in which the mixture strength is either too fuel-lean to provide adequate burning rates,

or so fuel-rich, that there is insucient oxygen to convert all the CO produced, into CO2 [60].

While the best method to solve this is by having higher temperatures or longer residence times,

this could lead to higher levels of NOx emissions [24]. Simultaneously, the design also needs

to keep smoke production low, by ensuring that enough air goes through the nozzle, all while

maintaining flame stability [54].

One method in which the combustor achieves the above balance, is by using the rich-quench-

lean (RQL) system. In this system, the combustion process occurs in three stages. In the first

stage, combustion is initiated in a fuel-rich primary zone. As the fuel-rich combustion products

flow out of the primary zone, they are quickly quenched using air jets to reduce their temperature.

The combustion process is then completed, by letting it run at controlled low temperatures [24].

With this system, the combustor is able to provide good flame stability, while still ensuring that

emissions of CO, UHC, smoke and NOx are kept low.


Chapter 2. Gas Turbine Technology 10

The challenge faced by combustion engineers in designing upcoming RQL combustors, is

in figuring out how to improve the combustion process that occurs in the three zones, while

reducing emissions, maintaining flame stability, and reliability. This means that most successful

solutions will require the combination of several methods, one of which is the fuel nozzle. Before

describing the requirements for a fuel nozzle, it is important to first understand the system

requirements for a combustor as a whole. The requirements for the combustor as described by

Lefebvre are [33]:

High combustion eciency

Low pressure loss (reduced fuel burn)

Low emissions - smoke, HC, CO and NOx

Relight at altitude

Good combustion stability and flameout margin

Reliable ignition

Reduced combustor tones

Uniform and low exit temperature distribution

Low weight, cost and life

As mentioned previously, these requirements have resulted in two dierent types of combus-

tors; straight flow-through and reverse-flow combustors. Reverse flow combustors are found in

small gas turbines, as they allow for good packaging with the centrifugal compressor, reduced

engine length, and lower weight. The straight through annular combustor is generally used for

large engines, since it works well the compressor type they generally employ. There are certain

drawbacks however with reverse flow combustors, in that they have a large exposed surface area

to the heat of combustion - requiring more cooling, and an increased sensitivity of combustion

eciency, due to the surface area.

Since we are interested in developing solutions for small gas turbine combustors, the study

on fuel nozzle technologies was conducted for implementation, in a RQL based reverse flow

combustor.
Chapter 2. Gas Turbine Technology 11

2.3 Fuel Nozzle

While gas turbines can work with both liquid or gaseous fuels, almost all aerospace gas turbines

work with liquid fuels [52]. Furthermore, since we are interested in a RQL type combustor, fuel-

air ratio control and mixing, are two very important parameters in order to control emissions [54].

Numerous other studies have also described where emissions, particularly CO, UHC and smoke,

depend on both the mixing rates near the nozzle, and atomization quality [31, 53, 58].

The performance of a fuel nozzle is described through a number of parameters, one of which is

the droplet size. Droplet size is important, as it is a key factor in controlling primary zone mixing,

ignition, altitude relight and emissions [54]. The droplet size impacts performance through the

evapouration rate, which is aected by factors such as contact area between the liquid and

air, vapour concentration gradient, and the mass transfer coecient [39]. While the latter two

properties are dependent on the liquid and the surrounding atmosphere, the first property is

directly controlled by the fuel nozzle. Consequently, smaller droplet sizes are generally desired

at all power conditions. The spray cone angle and circumferential uniformity are the other two

important parameters for fuel nozzles. Adequate spray cone angle and uniformity is required

to provide a satisfactory temperature distribution at the turbine inlet [52], and to also prevent

any cold or hot spots inside the combustor that could result in high UHC or CO emission [38].

Another issue, is the emissions of UHC and CO, due to quenching that occurs at the wall of the

combustor.

In addition, since the fuel nozzle has to be able to deal with low fuel flow rates for small gas

turbines, the fuel metering hole sizes can be quite small. Subsequently, fouling of the metering

passages either due to fuel contamination, gumming, or carbonizing, can have a serious impact

on the performance of the nozzle. Fouling of the nozzle can also aect the durability of the

combustor, and the rest of the engine hot end components. Coke formation inside the nozzle

occurs, because fuel inside the nozzle gets hot, and causes chemical reactions to occur forming

deposits. The main source of this heat, is from the exposure of the nozzle front face to the heat

of combustion.

Consequently, the method of fuel injection, and the type of fuel distribution selected, are
Chapter 2. Gas Turbine Technology 12

major design factors that are determinative of the performance of small combustor configura-

tions. So, in order to develop a successful atomizer design, the various requirements that fuel

nozzles need to meet can thus be summarized by the following [32]:

Small droplet size


Stable and Uniform Spray Distribution
High fuel turn-down ratio
Reasonable fuel system injection pressure requirements
Low combustor pressure drop
Good air-fuel interaction
Low sensitivity to fuel properties
Optimal internal wetted wall thermal management
Good coking abatement
Low weight, cost and maintenance
Chapter 3

Liquid Atomization Fundamentals

3.1 Introduction

Atomization is the process of transforming the bulk liquid, into sprays of small droplets. Over

time, a number of dierent atomization methods for liquids have been developed. Most common

atomizers work by either using the kinetic energy of liquid itself to create the droplets, or by

exposing the liquid to high velocity air. While there have been other atomizers that work

through mechanical or ultrasonic means, they are not as prevalent as the former in aerospace

gas turbines [33].

Spray formation is often described in two stages; primary and secondary atomization [17].

Primary atomization is considered to be the formation of droplets from the bulk fluid, which is

typically in the form of a sheet or jet. The second stage of secondary atomization occurs when

droplets interact with the ambient flow to further break up into smaller droplets. Spatially,

primary atomization occurs at or near the nozzle exit, while secondary atomization occurs

downstream of the nozzle. Within the primary stage of atomization, there are two mechanisms

that further describe the two possible atomization methods. The two methods of atomization

are often labelled as classical or prompt [33]. The classical method of atomization is generally

used to describe the atomization process when the liquid pressure or air velocities are low. For

cases of the contrary, the prompt mechanism of atomization is used to describe the atomization

process [11].

13
Chapter 3. Liquid Atomization Fundamentals 14

To have complete understanding of the atomization process, it is vital that the primary and

secondary modes of atomization are taken into account in any analysis. This is because the

combination of these two processes determine a number of spray properties such as: fuel spray

penetration, droplet number density, droplet velocity and droplet size distribution [32]. The

factors that aect the two processes of atomization are the internal nozzle geometry, liquid and

ambient gas properties.

3.2 Eect of Liquid Physical Properties

While the method in which the bulk fuel is transformed into sprays is unique to the type of

atomizer used, factors that are often independent of the atomizer, are the liquid fuel physical and

chemical properties. In a majority of nozzles, the flow and spray characteristics are influenced

by liquid properties such as density, surface tension and viscosity [32]. The significance of

density for atomization performance is often diminished by the fact that liquids often exhibit

small changes in density over their range of usage. The only time a significant density change

occurs, is when dierent fuels are being used in the same atomizer. This however is becoming

increasingly important, with the introduction of biofuel for use in future aerospace gas turbines.

The primary reason for atomizing liquid fuels, is that because it increases the net surface area

of the liquid fuel that is exposed. This generally allows a larger amount of fuel to evapourate,

and thus achieve a more complete combustion process. A fuel property that is used to measure

the capability to provide a multiplication of the net exposed surface area, is surface tension.

The surface tension represents the energy or work required to increase the surface area due to

intermolecular forces [32]. This parameter is often integrated into the Weber number (We), in

order to study its eect on atomization. The Weber number is the ratio of the inertial force to

the surface tension force, and is defined as:

a UR2 Ddrop
We = , (3.1)
L

where a is the density of air, UR is the relative velocity of the air to the liquid, L is the

surface tension of the liquid, and Ddrop is the droplet diameter. The higher We, the larger the
Chapter 3. Liquid Atomization Fundamentals 15

(a) (b)

Figure 3.1: Studies by Lefebvre on: (a) Eect of temperature on surface tension for dierent fuels, (b)
Eect of temperature on viscosity for dierent fuels (Lefebvre [33])

deforming external pressure forces are compared with the restoring surface tension forces [32].

Unlike density, the surface tension can change quite significantly as the temperature of the

liquid changes. Figure 3.1(a) shows how the surface tension changes with temperature for a

number of fuels with dierent densities. For comparison purposes, the density of Jet-A1 as per

specification, is supposed to be between 0.78 to 0.84 [9].

Another important liquid property is the liquid viscosity; it represents the fluids resistance to

flow. An increase in viscosity is generally undesirable, since it hinders the atomization processes,

especially the droplet size, and spray distribution. Figure 3.1(b) shows how the viscosity of

dierent fuels change with the temperature.

The analogous parameter of importance for the eect of viscosity on atomization, is the

Ohnesorge number (Oh). The Oh number signifies the ratio of the internal viscosity forces to

the inter-facial surface tension force, and is defined as [20]:


Chapter 3. Liquid Atomization Fundamentals 16

p
We
Oh = , (3.2)
Re

where the We is defined as in Equation (3.1), and the Reynolds number (Re) is defined as [32]:

L UR DDrop
Re = , (3.3)
L

where L is the density of liquid, and L is the viscosity of the liquid. With respect to the two

mechanisms of atomization that were mentioned before, the classical method of atomization is

used for cases when we encounter a low We number. For cases when we are dealing with large We

number (high liquid pressure or air velocities), then the process is described using the prompt

methodology. While the classical mechanism is quite strongly aected by liquid properties, the

prompt mechanism is not.

3.3 Classical Mechanism of Atomization

The classical mechanism of atomization is generally explained assuming breakup of a liquid jet

or sheet. The primary reason for this is because, when the liquid is setup as a thin sheet or jet,

it has the highest surface energy, making it highly unstable [27]. Is it for this reason that most

nozzles develop jets or liquid sheets in the first step of atomization.

3.3.1 Breakup of Liquid Jets

When a liquid jet emerges from a nozzle in a continuous cylindrical form, oscillations and

perturbations form on the surface of the jet which serve to de-stabilize the jet. Under ideal

conditions, these instabilities grow, and then cause the jet to breakup into ligaments and then

droplets. The growth and development of these instabilities is a function of the relative velocity

between the fuel jet and the surrounding air. Rayleigh was one of the first to study jet breakup,

where he studied the simple case of a low velocity laminar jet breakup in air assuming no

viscosity. According to Rayleigh, jet breakup occurred once the instabilities grew to a certain

wavelength ( opt ) of 4.51d (where d is the jet diameter) [27]. At this point a cylinder of length

4.51d then became a spherical drop of Diameter D (m), where:


Chapter 3. Liquid Atomization Fundamentals 17

D = 1.89d. (3.4)

Weber extended the work of Rayleigh, by including the eect of viscosity and air resistance

on jet breakup. These two aspects are particularly important for most modern nozzles, where

the relative velocity between the air and liquid is often quite high. He found that the eect of

increasing viscosity was to increase the optimal wavelength; meaning that we have a delay in jet

breakup, thus larger droplets. He also found that the eect of increasing the relative air velocity

was to reduce opt , and thus consequently produce smaller droplets [61]. As shown in Figure

3.2, depending on the relative velocity between the fuel and air we can have four dierent types

of atomization [47]:

Low relative velocity Rayleigh jet breakup. This regime is characterized by a linear re-

lationship between the length of the jet prior to breakup, and the jet velocity. In this

regime, the droplet size exceeds the jet diameter.


First wind-induced breakup. In this regime, the surface tension eect is enhanced by the

relative liquid-air velocity, producing a pressure distribution on the jet and accelerating

breakup. Droplet diameters in this case are about the same as the jet diameter.
Second wind-induced breakup. As the relative velocity further increases, the drops are

produced by the unstable growth of the short wavelength surface instabilities. The dier-

ence with the first wind-induced cases is that the surface tension opposes the wave growth.

Breakup still occurs several diameters downstream of the atomizer, and the average droplet

diameter is smaller than the jet diameter.


Atomization. The jet disrupts completely at the nozzle exit and the average droplet and

ligament sizes are much smaller than the diameter of the jet.

3.3.2 Breakup of Liquid Sheets

Most modern atomizers atomize fuel by dispensing the fuel in form of a sheet, since it maximized

exposed surface area. Like liquid jets, the relative velocity between the fuel and air causes the

liquid sheet to undergo surface wave instabilities. As shown in Figure 3.3, the instabilities

generated on the sheet causes slabs of fuel to break away from the leading edge, with a width
Chapter 3. Liquid Atomization Fundamentals 18

Figure 3.2: Eect of jet velocity on type of atomization (Farvardin et al. [14])

of 1
2 opt [33]. These ligaments then further disintegrate into droplets, according to the Rayleigh

mechanism. As the relative velocity increases, sheet breakup comes closer to the nozzle, as the

optimal wavelength reduces.

3.4 Prompt Mechanism of Atomization

As was briefly mentioned before, prompt atomization is when the liquid sheet or jet breakup

happens very rapidly. In these cases, there is no time for the wave like disturbances to cause

jet disintegration. As soon as the liquid jet or sheet exits the nozzle, it is torn into fragments

due to the high fuel-air relative velocity. Due to the rapid and violent mode of atomization, the

resulting droplet size is often independent of the initial jet or sheet thickness [33].

For nozzles in use today, the distinction between the prompt and classical mode of atomiza-

tion is not always clear. Nozzles often start o at low We numbers, where the dominant mode of

atomization is classical, transitioning into a prompt mode of atomization when the We number
Chapter 3. Liquid Atomization Fundamentals 19

Figure 3.3: Liquid sheet breakup (Fraser [16])

starts to get large [59].

3.5 Breakup of Drops

The final stage of atomization is predominately dominated by the breakup of larger droplets,

and is generally called secondary atomization [6]. The droplets break down when external forces

deform the liquid surface, and aect the equilibrium setting between the external and internal

forces. The external forces are generally aerodynamic, while the internal forces are related to

surface tension and viscosity. As in the case for liquid jet disintegration, there are a number

of dierent modes of droplet breakup, depending on the competing aerodynamic and surface

tension/viscosity eects. The parameter that is used to describe the dierent modes of droplet

breakup, is the We. As mentioned before, the higher the We, the larger the deforming forces

acting on the droplet, and more violent breakup is.

In addition to determining the mode of droplet breakup using the We, there is a key param-

eter known as the critical Weber number which represents the criterion for deformations, that

lead to secondary disintegration of a drop. The critical Weber number is described Bayvel [6]

and is given by:

a UR2 DDrop 8
W e = = 0 , (3.5)
L CD
Chapter 3. Liquid Atomization Fundamentals 20

1)

2)

3)

4)

5)

Figure 3.4: Droplet breakup with the Weber number(Bayvel [6])

where CD
0 is the drag coecient of the drop. For low viscosity fuels, experimental data on

0 indicate an average value for We of around 12 [33]. Thus, secondary disintegration of


CD

drops proceeds when We We . Figure 3.4 shows the dierent modes of droplet breakup.

The first mode is the vibrational mode, and occurs for We 12. Here, the drop disintegrates

into almost identical secondary droplets, whose number in general does not exceed four. The

second stage is called the bag mode of disintegration, and occurs for We < 20. In the third

stage (20 < We < 50), the mode of breakup is called the bag-stamen mode. In this mode, the

droplet flattens out in the direction of airflow in the form of a parachute, with a liquid jet in

the center, and then breaks into small droplets. The fourth stage is called the stripping mode

of disintegration, and is for 50 < We < 100. In this case, the fuel droplet flattens out before

disintegration, producing small droplets and liquid threads of various shapes. The last stage is

called catastrophic breakup, and it occurs for We > 100 [45]. In general, the higher the We is,

the smaller droplet sizes are expected to be. A more detailed review of droplet breakup can be

found in studies by Bayvel [6], Lefebvre [33] or Pilch [45]

3.6 Liquid Atomization Characteristics

The performance of a fuel nozzle is generally described using a number of parameters that

describe either the macroscopic or microscopic structure of the atomized liquid. The macro-
Chapter 3. Liquid Atomization Fundamentals 21

scopic structure describes the external shape of the spray such as the spray cone angle and its

penetration, while the microscopic features describe the distribution of the liquid within the

spray [6]. The importance of knowing these parameters lie in the fact that during the design

process of fuel nozzles, certain values for these parameters are set, so that when integrated with

the combustor, we get good mixing and combustion characteristics, while also avoiding issues

such as fuel hitting the combustor side walls.

3.6.1 Discharge Coecient

The discharge coecient (CD ) is an important parameter that is used to describe the perfor-

mance of a nozzle. It describes the ratio of the actual flow, to the theoretical flow from the

nozzle. The CD is influenced by a number of factors such as the Reynolds number, internal

nozzle features, injection pressure and ambient pressure. Thus, the nozzle performance charac-

teristics are not only determined by its design, but also that of the environment. In terms of

the range of CD , its value changes depending on the type of nozzle being studied. While certain

nozzles are designed to have a high CD of about 0.8 (plain orifice nozzle), others are designed

to have a CD of about 0.3 (pressure swirl nozzle) [4].

3.6.2 Flow Number

The flow number (FN), is a property that is used to describe the eective flow area of pressure

based atomizers, and is defined as [32]:

m
FN = p , (3.6)
P
where m is the mass flow rate in hr ,
lb
and P is the pressure drop in Psi. Its importance arises

from the fact that for any given pressure nozzle, the FN is a constant, and it is determined by

its physical design. This aspect can be explained through:

m = CD L A0 U, (3.7)

where A0 is the area of the orifice, and U is the velocity of the fuel leaving the nozzle. The only

parameter that changes in Equation (3.7) is the velocity (U ), which is a function of the pressure
Chapter 3. Liquid Atomization Fundamentals 22

drop across the nozzle ( P ). Furthermore, for pressure atomizers, it has been shown in the past

through experimental studies, that the change in CD is often insignificant for changing pressure

drops, and can be considered to be constant. This essentially leaves m / U / P . The FN

of a pressure nozzle, provides an indication of the nozzles performance for various flow rates.

As such, nozzles with a small FN are good at atomizing low flow rates, and vice versa. This is

because a large pressure drop is being used to atomize a small amount of fuel. While the concept

of the FN can be used for non-pressure nozzles, it is more an indication of its control (pressure

drop and mass flow rate input), than its performance. One item that needs to be taken into

consideration is the fact that theFN is a constant for a given liquid, and changes with liquids

of dierent density.

3.6.3 Spray Cone Angle

The spray cone angle () is used to describe the spread of the fuel droplets from the nozzle. It

is vital to predict this parameter because it provides the ability to properly utilize the spray.

An issue with the measurement and definition of lies in that it generally narrows due to air

interactions, as shown in Figure 3.5. However, the most widely used convention is to measure

, at a certain axial distance from the nozzle, as shown in Figure 3.5. It should be mentioned

that inside the combustor, the spray distribution will be quite dierent due to the re-circulating

flow inside the combustor.

3.6.4 Patternation

Given the heterogeneous nature of the atomization process, it should be expected that the main

and satellite drops will vary with size. In addition to drop size, another parameter of importance

is the distribution of the spray. The most widely used expression for drop size distribution is

one that was originally developed by Rosin and Rammler [32], and is given by:

q
D
1 Q = exp , (3.8)
X
where Q is the fraction of total volume contained in drops with a diameter lower than D, and X

and q are two parameters which characterize the drops composing the spray. It is important for
Chapter 3. Liquid Atomization Fundamentals 23

Figure 3.5: Description of the spray cone angle (Bayvel [6])

combustors, since spray distribution has a significant impact on combustion eciency, emissions

and life of many hot end components. The exponent q provides a measure of the spread of drop

sizes, where the higher the value of q, the more uniform the spray is. Subsequently, from

an analysis of a considerable body of data obtained from several types of pressure-swirl and

airblast nozzles, Rizk et al suggested that a modified version of the equation 3.8 was a better

fit to experimental data [32]:


q
ln D
1 Q = exp . (3.9)
ln X

3.6.5 Droplet Size

The droplet size is one of the most important parameters that describes the quality of atom-

ization. Despite the fact that fuel nozzles produce a distribution of droplets, the droplet size is

often described in terms of a single mean droplet diameter. Depending on the way the mean

droplet diameter is calculated, it can take into account dierent parameters such has number

density, surface and volume of drops [6]. As such, the type of droplet diameter selected, de-

pends on the type of application that we are interested in. The droplet parameter that is most

widely used is the Sauter Mean Diameter (SMD). SMD is widely used because it characterizes

a number of important processes for combustion analysis such as droplet penetration, heat and
Chapter 3. Liquid Atomization Fundamentals 24

mass transfer. Technically, the SMD represents the diameter of a uniform drop set with the

same total volume and the same surface area of all drops, as in the real set [6]. In general, a

small SMD is desired as it generally allows for a more complete combustion process, and lower

overall emissions.

The SMD is a function of a number of dierent parameters depending on the type of nozzle

being studied. For pressure based atomizers, the important parameters are the pressure drop,

mass flow rate and liquid properties such as viscosity and surface tension. For atomizers that

use air to help with the atomization process, the air to liquid ratio, and the relative fuel air

velocity are the two important parameters.

3.7 Fuel Nozzle Coking

In addition to atomization performance, another aspect of importance in nozzle design is coking.

Coking is the formation of solid carbon deposits, due to high fuel temperatures inside the nozzle.

Coking is important because the buildup of solid carbon deposits inside fuel nozzles can either

partially or fully bock narrow fuel passages. This can not only lead to increase pumping power

requirements, but also aect spray uniformity. This can consequently aect emissions, flame

stability, and life of many hot end components of the gas turbine.

3.7.1 Types of Coking

3.7.2 Eect of Fuel Temperature on Coking

The thermal stability of the fuel generally determines coking, as it dictates the mechanisms

that produce the solid carbon deposits, and is aected by the fuel temperature, pressure, and

amount of oxygen in the fuel [36]. Coking is generally considered to occur in two stages, and

is dierentiated based upon the type of underlying chemical reactions. as shown in Figure 3.6,

the first stage is called auto-oxidation, and occurs over temperature ranges between 450 K and

600 K.

The second stage (pyrolysis), occurs at temperature above 720 K. In between the two

stages, we have a transition stage, where both processes are going on [2]. For the autoxidation
Chapter 3.Thermal Stability of Jet
Liquid Atomization Fuels
Fundamentals 25
Deposit Formation Temperature Regime Diagram

FigureFrank
3.6: Eect of fuel Temperature
T. C. Yuen (UTIAS)
on coke deposition rate (Huang
Coking Avoidance: Experimental Studies
et al.
26 May 2014
[21])
35 / 40

process, the fuel oxygen content has a significant eect on deposit formation process. As shown

in Figure 3.6, the coking potential for oxygenated fuel is much higher than for de-oxygenated

fuels. Studies have found that the auto-oxidation rate increases until an upper limit is reached,

after which the deposit formation rate decreases for most of the transition stage [5]. After

the transition stage, the fuel starts to decompose due to excessive heat through pyrolysis reac-

tions known as hydrocarbon cracking. Pyrolysis is heavily dependant on the temperature, fuel

composition, and time exposed to the high temperatures [36].

It should be mentioned that not all the fuel degrades to form coke or deposits. Both the

auto-oxidation and pyrolysis processes are considered to compose only a small part of the fuel

degradation process, only a small portion of the fuel participates in the reactions that form

deposits [19].
Chapter 4

Fuel Nozzle Technology and

Performance Characteristics

4.1 Fuel Nozzle Technologies

A number of dierent fuel nozzle technologies have been developed to-date to produce sprays.

The driving force behind these designs, has traditionally been the the requirement to reduce

emissions while also providing stable flame characteristics inside the combustor. At the core of

any design, essentially what is needed is to provide a high relative velocity between the liquid

and the surrounding gas. Generally, this is achieved by either atomizing the fuel at a very high

velocity, or by forcing fast moving air over slow moving liquid fuel. The dierent types of nozzles

in use in modern gas turbines can be classified into the following main groups:

Pressure nozzles
Air based nozzles
Rotary nozzles (Slinger)

4.1.1 Simplex Nozzle

Simplex nozzles are the simplest type of nozzles, where a fuel spray is generated by forcing

liquid fuel through a suitable shaped nozzle orifice by mean of pressure [34]. Simplex nozzles

have little variation in design, due to their simplicity. While this allows for a relatively simple

design process, it also limits the range of features that one can expect the nozzle to display.

26
Chapter 4. Fuel Nozzle Technology and Performance Characteristics 27

(a) (b)

Figure 4.1: Schematic of a simplex atomizer with: (a) swirl chamber, (b) Schematic of a simplex
atomizer with swirling insert, (Westergaard [62])

The two main aspects of a simplex nozzle are its swirl chamber and orifice. The orifice

determines fuel mass flow, while the swirl chamber aects the spray cone angle and distribution.

The swirl chamber achieves this by imparting a tangential velocity component to the fuel, which

helps the nozzle produce a wide hollow cone spray. This is generally achieved by either using

tangential holes to the swirl chamber as in 4.1(a), or by using a grooved core insert as in 4.1(b).

The primary dierence between the two designs, is that the minimum passage size with

the swirling insert generally generally tends to be smaller, that those with the tangential inlet

slots. This aspect is important because if the passage sizes get too small, they can get clogged

and not perform as designed. The performance of pressure swirl nozzles is aected by a number

aspects such as fuel properties, internal nozzle design and the fuel pressure drop. This particular

aspect is best by Equation (4.1), which is a semi-empirical relationship developed by Wang and

Lefebvre to describe the droplet size of pressure swirl nozzles [35]:

0.25 0.25
2L 0.25 2L
SMD = 4.52 2 (tcos()) + 0.39 (tcos())0.75 . (4.1)
a L a L

From Equation (4.1), we see that that the viscosity and pressure drop are two very important
Chapter 4. Fuel Nozzle Technology and Performance Characteristics 28

Figure 4.2: Spill return nozzle (Norgren et al. [44])

parameters. We also see the importance of the spray cone angle and sheet thickness, both of

whom are strongly aected by previous parameters, along with the nozzle internal features and

flow number. Pressure swirl atomizers are vital because they help in situations where little or no

air pressure drop is available across the liner. Their simplicity in design, and ability to produce

relatively wide sprays makes them very good for use at starting and low power conditions for

small gas turbine combustors. They are however inecient at high power conditions, as the fuel

pressure drop required to atomize a larger quantity of fuel can get very high as is the case for

nozzles with small flow numbers.

4.1.2 Spill Return Nozzle

A variant of the pressure swirl nozzle that combats some of the above issues, is the spill return

nozzle. The spill return nozzle is essentially a pressure swirl nozzle with a spill line attached at

the back. As shown in Figure 4.2, the only addition for a spill return nozzle is with 5, where we

have a spill line (which is this case is an annular spill line). The nozzle works by supplying fuel

to the swirl chamber, where a certain percentage of fuel is spilled away from the nozzle to the

feeding system, and the rest is discharged through the orifice in the form of a well developed

spray. Generally, the spill line is used when atomizing low fuel flow rates, and is shut for higher

fuel flow rates. The purpose of adding a spill line, is to allow the spill return nozzle to act as a

variable flow number nozzle [8]. The starting flow number of the nozzle is set by the performance

of the nozzle with the spill line closed, and can be reduced as a given percentage of the total

input fuel is spilled away.


Chapter 4. Fuel Nozzle Technology and Performance Characteristics 29

Since liquid fuel is generally sent into the swirl chamber at a higher quantity and pressure

than is needed, this means that even for extremely low flow rates, the nozzle is able to provide

good atomization. Studies by Carey [8] have shown that good atomization can be achieved

for cases where the flow rate is as low as 1% of max flow. Spill return nozzles also have large

fuel turndown ratios, which is useful in modern combustors that tend to have high pressure

ratios. Furthermore, since the nozzle has larger internal passage due to the fact that it handles

more fuel than is atomized, coking issues are reduced. This is important for combustors with

high pressure ratios as the fuel temperature can often rise due to hot air from the compressor

imparting heat on the nozzle feed arm, or from the high heat loading on the fuel nozzle face due

to flame radiation. Furthermore, the ability to spill fuel from the front of the nozzle has been

shown to help cool the nozzle body [44], which should help to reduce coking issues due to the

fuel overheating. Other attractive features with this technology are the inherent simplicity in

design and implementation, which often results in a technology that is durable [50].

Certain issues that exist with the spill return nozzle are that the spray cone angle is a

function of the fuel spill percentage [49], and that fact that the spilled line needs to be properly

controlled. There is also the issue that extra hardware is needed to handle the spilled fuel,

which can be added weight. Unfortunately, no semi-empirical relationships exist for spill return

atomizers. However, they are aected by the same set of parameters (along with the addition

of the spill percentage) as pressure swirl nozzles; owing to their similarity in design.

4.1.3 Duplex Nozzle

The duplex nozzle is another technology that has been developed to solve some of the issues

with the simplex nozzle. It achieves this by using two simplex nozzles placed concentrically

within each other, with each nozzle (circuit) optimized for a given flow number. As shown in

Figure 4.3(a), a small flow number circuit is used for atomizing low fuel flow rates, and a larger

flow number circuit is used for larger flow rates at high power. This allows the nozzle to have

a large turndown ratio, without the need for very high fuel pressure requirements at larger fuel

flow rates, while also providing for a nozzle that is better optimized for performance than a

simplex nozzle.
Chapter 4. Fuel Nozzle Technology and Performance Characteristics 30

(a) (b)

Figure 4.3: (a) Duplex Nozzle, (b) Flow characteristics comparison of simplex and duplex nozzles,
(Lefebvre [32])

The duplex nozzle works by only allowing fuel to the primary circuit at low fuel flow rates.

At this point atomization quality is high because a fairly high pressure is needed to force the

fuel through the small ports in the primary circuit. As the fuel flow rate increases, fuel is

also admitted to the secondary circuit, which has larger ports to handle the larger fuel flows.

Certain issues with duplex atomizers are that passage sizes can be quite small, particularly

for the primary circuit, and thus coking can be an issue. Another issue with designing duplex

nozzles, is ensuring that the two circuits work together in a positive manner when they are in

use. This is often not the case at lower power conditions when the secondary circuit first starts

up, and often requires complex design fixes [25].

4.1.4 Air Assist Nozzle

An alternative approach to atomizing fuel, is to use air to assist with the atomization process.

Unlike simplex atomizers which purely depend on the fuel injection pressure, air assist nozzles

use a small quantity of high velocity air supplied from a pump to help with atomization. In

this design, a slow moving stream of liquid is exposed to a high velocity stream of air, where

a large dierence in the relative velocity results in a shearing process, tearing up the fluid and
Chapter 4. Fuel Nozzle Technology and Performance Characteristics 31

Air supply
Air supply
Fuel supply Fuel supply

(a) (b)

Figure 4.4: Description of: (a) External mixing air-assist nozzle, (b) Internal mixing air-assist nozzle,
(Bayvel [6])

atomizing it. The two dierent types of air assist nozzles are: internal and external mixing air

assist nozzles, and are shown in Figure 4.4.

In an internal mixing nozzle, the fuel and air is mixed inside the nozzle as shown in Figure

4.4(a); the opposite is the case for the external mixing air assist nozzles. In internal mixing

nozzles, generally the spray cone angle is dependant on the airflow rate. As the airflow reduces,

the spray cone angle generally tends to widen [23]. The semi-empirical relationship describing

the SMD for external mixing air assist nozzles as developed by Elkotb et al. is described as [13]

:
0.29
0.39 0.18 mL
SMD = 51D0 Re We , (4.2)
mA
where mL and mA are the liquid and air mass flow rates respectively. In addition to liquid

properties, we see the importance of the relative velocity to the performance of atomizer as it

appears in both Re and We. The air-liquid ratio (ALR) and D0 are two other parameters of

importance. The former parameter determines how much air there is in contact with the fuel,

and the latter parameter takes into account how thin the liquid is coming from the nozzle orifice.

This type of atomizer is good for highly viscous liquids, and also for atomization of low fuel

flow rates. Furthermore, the liquid and air pressure settings are interactive, as increasing the

air pressure will induce a higher liquid flow rate and will produce finer droplet sizes. External

mixing air assist nozzles have the advantage of having a constant spray cone angle for all airflow

rates, at the expense of a less ecient usage of air [32]. Also, the liquid and air setting are

not interactive, so each can be individually controlled. While this capability allows external air
Chapter 4. Fuel Nozzle Technology and Performance Characteristics 32

assist nozzle to produce finer atomization, creating a hollow cone often is an issue. A major

drawback to all air assist nozzles is that the air used for atomization, needs to be supplied via

a pump. This has traditionally made them less popular for use, in aerospace applications.

4.1.5 Airblast Nozzle

Airblast nozzles are another variant of nozzles that use air to assist with the atomization process.

The dierence between air assist and airblast nozzles lie in the quantity of air used, and its

atomizing velocity. Instead of using a pump to provide the air, the pressure dierence across

the combustor liner drives the flow of air through the nozzle. As such, while the velocity of air

might be lower in comparison to air assist nozzles, the quantity of air used is much higher. As

shown in Figure 4.5(a), airblast atomizers work by creating a slow moving stream of liquid on a

fuel prefilming surface, that is exposed to a higher velocity stream of air. A large dierence in

the relative velocity between the air and the fluid results in a large shearing process and tears

up the fluid, thus atomizing it [15]. As shown in Figure 4.5(a), the shearing stresses between

fluid and air are enhanced by sandwiching the fuel sheet between two swirlers. The swirling air

also helps with enhance the mixing of fuel and air, which has a number of benefits in terms of

emissions and combustor durability. Another advantage with airblast nozzles is that they have

a low potential for coking, since they have larger internal passages. This makes the airblast

nozzles more reliable, particularly for use over long periods of time.

Atomization performance for airblast atomizers, is often described by a number of semi-

empirical relationships, owing to the dierent types of airblast atomizers developed. One rela-

tionship that is widely used, is given in Equation 4.3 [51]:

0.5 0.425 2
3 L L DP WL 3 2 WL
SMD = (3.33 10 ) 1+ + (13 10 ) L Dp0.575 1+ . (4.3)
2A UA2 WA L L WA

The two terms in Equation (4.3) handle fluids of low and high viscosities respectively. For low

viscosity liquids such as water and Jet-A1, the SMD is aected by the surface tension, liquid

density, air density, air velocity, atomizer dimensions and ALR. At higher viscosities they carry

over, however the air velocity and density are not quite as important.

Airblast atomizers have a number of advantages over pressure atomizers in that they gener-
Chapter 4. Fuel Nozzle Technology and Performance Characteristics 33

(a) (b)

Figure 4.5: Description of: (a) Pre-filming airblast nozzle (Lefebvre et al. [33]), (b) Performance of
airblast and duplex nozzles (Mellor [39])

ally have low fuel pressure requirements, and produce a finer spray in cases when an adequate

pressure drop across the nozzle is available (Figure 4.5(b)). Furthermore, with the usage of a

larger amount of air, they are able to ensure better mixing between the fuel and air. This is

advantageous because it generally results in lower emissions, low soot formation, and flames of

lower luminosity; enabling better durability for many hot-end components of the engine [12].

The biggest issue with airblast nozzles is that performance is poor when the pressure drop across

the nozzle is not adequate. This is often the case for small combustors at starting and certain

lower power conditions, where the amount and velocity of air available is not adequate. Gener-

ally, performance of airblast atomizers drop when the ALR drops below four, and deteriorates

quite rapidly at ALRs below about two [30]. To some extent, a low airflow rate may be com-

pensated by an increase in air velocity, but this often means increasing the pressure drop across

the nozzle, which can have other undesirable implications. Despite this, airblast atomizers have

become the mainstay for almost all aerospace gas turbines, due to their ability to provide stable

flame characteristics and low emissions, for a wide range of power conditions. The issue of a

lack of performance at low power conditions is often combated by pairing a pressure nozzle with

the airblast nozzle.


Chapter 4. Fuel Nozzle Technology and Performance Characteristics 34

Figure 4.6: Hybrid nozzle with a simplex primary circuit, and an airblast secondary circuit (Lefebvre
et al. [33])

4.1.6 Hybrid Nozzle

The hybrid nozzle, also know as the piloted airblast nozzle, is a combination of pressure and

airblast nozzles used together. The basic design features of the atomizer are illustrated in

Figure 4.6. We see that the atomizer incorporates a pressure nozzle at the centerline(pilot),

and an airblast nozzle located at a larger radius. The atomizing air for the airblast nozzle is

supplied through an outer swirler, and a number of tangential radial ports leading to the space

surrounding the pilot nozzle. The goal with this design is to overcome the airblast nozzles

inherent drawbacks of poor atomization performance, during idle and low power conditions.

The pilot nozzle works primarily for starting and low fuel flow rates, providing good atomization

and stable flame characteristics. At high power conditions, fuel is supplied to both the airblast

and pilot nozzles, with the majority of fuel being sent to the airblast nozzle.

A big challenge with hybrid nozzles is the integration of the two circuits. If certain features

of the nozzle like the pilot spray cone angle, flow rates through each circuit, and pressure drop

across the inner and outer air swirler are not optimized, each circuit could interact in a negative

manner, reducing overall nozzle performance [48]. All this makes implementation of the hybrid

nozzle more complex than pressure or airblast nozzles individually. Implemented optimally,

they enable the advantages of both technologies to be exploited, where the nozzle provides

stable combustor flame characteristics, while also emitting the lowest possible emissions.
Chapter 4. Fuel Nozzle Technology and Performance Characteristics 35

4.2 Nozzle Performance

The first aspect that was investigated as part of the thesis, was the relative performance of

dierent fuel nozzles on gas turbine emissions, in a small reverse-flow combustor. The goal here

was to learn which nozzle performed the best for a given set of conditions. One study that helps

demonstrate this aspect the best, is by Norgren et al. In their study, Norgren et al. investigated

the eect of dierent nozzles on combustion eciency, pattern factor and emissions of CO, NOx

and UHC [44]. As mentioned before, while the fuel nozzle has an impact on these parameters, it

is not the only influencing factor. Despite this, being able to study the above parameters when

everything else is fixed, is a direct indication of the strength of that particular nozzle technology.

The combustor studied by Norgren et al. was a small custom built reverse-flow combustor

by NASA, with a pressure ratio of 16 : 1 and 18 fuel nozzles. The tests were done for power

conditions simulating ground idle, low power, takeo and cruise conditions, and are given in the

original paper by Norgren et al. [44]. The nozzles studied were simplex, duplex, spill return, air

assist and splash cone airblast nozzle.

4.2.1 Combustion Eciency

Studying combustion eciency is important because it indicates what fraction of the energy

available in the fuel was released, in the combustion process. Low combustion eciency is unde-

sirable since it generally means fuel is wasted, as it has not gone through complete combustion.

Aerospace gas turbines are generally designed to have high levels of combustion eciency

because a failure to do so would mean wasted fuel; in the sense that it has not gone through

complete combustion. Combustion eciency is often measured through the amount of CO or

UHC that is emitted in the exhaust. Large amounts of soot deposition on components like

fuel nozzles, is another indicator of low eciency. To understand the influence a nozzle has on

combustion eciency, it is important to know what factors aect combustion eciency. The

various factors that are known to aect combustion eciency are [33]:

1 1 1 1
c = f (airflow rate) + + . (4.4)
mixing rate evapouration rate reaction rate

For practical combustion systems, the latter three factors of Equation (4.4), govern the eciency
Chapter 4. Fuel Nozzle Technology and Performance Characteristics 36

of heat release; with rarely all three limiting at the same time. For most modern RQL based

combustors, the mixing rate and evapouration are often the limiting factors. The mixing rate

is aected by the interaction of fuel and air from the nozzle through the spray cone angle and

their relative velocity; evapouration is aected through the atomized droplet size.

At simulated flight conditions of cruise and takeo, all the nozzles had a combustion eciency

that was greater than 99% [44]. At low power conditions however, Norgren et al. found that

eciency dropped drastically. This was primarily because of the deterioration in the fuel spray,

due to low fuel flow rates, injection pressures or an inadequate air pressure drop across the

nozzle.

For ground idle conditions, the spill return and air assist nozzles had the highest combustion

eciency of all the nozzles tested. For instance, while the spill return had a combustion eciency

of 92%, the air assist and duplex nozzles had eciencies of 74% and 52% respectively [44]; the

simplex and airblast nozzles were found to be inoperable at this power condition. Similar trends

extended to other stimulated low power conditions. While the spill return nozzle had an overall

combustion eciency range of 92% 98% from ground idle to the highest stimulated low power

condition, the air assist nozzle had a range of 75% 98%, the duplex nozzle a range of 52% 91%,

the simplex nozzle a range of blowout to 80%, and the airblast nozzle a range of blowout to

96%, over the same power conditions. The trio of the spill return, air assist and duplex nozzles

performed well, because their design allows them to provide self sustaining combustion by using

the fuel pressure in dierent methods to get good atomization and mixing. The airblast nozzle

while starting poor, starts to achieve good atomization at higher power conditions, as there is

an adequate air pressure drop across the nozzle available.

This highlights the issue with airblast atomizers in general, where the combustion eciency

is low at idle and low power conditions. This is due to the fact that at low power conditions,

the air pressure drop across the combustor is small, and it cannot assist adequately in the

atomization process. However, once a critical point (fuel flow rate and associated pressure) were

met, the atomization properties become favourable.

With respect to the air assist nozzles, Norgren et al. found that the nozzle pressure ra-

tio(pressure ratio is defined as the ratio of the pressure in the injector to the pressure in the
Chapter 4. Fuel Nozzle Technology and Performance Characteristics 37

combustor) had little eect at higher power conditions. Its biggest eect was at idle, combustion

eciency was 68%, 80% and 75% for pressure ratios of 1.2, 1.1 and 1.05 [44].

4.2.2 Pattern Factor

The importance of the pattern factor (PF) arises from the fact that it accounts for the tem-

perature of the gases discharging into the turbine. It is very important that the temperature

of these gases are always less than a limit, as it could lead to turbine blades being damaged

or even destroyed. Despite the fact that combustion gases from the primary zone are diluted,

the eux gases are strongly influenced by spray characteristics such as droplet size, spray cone

angle and spray penetration; this is because they control the fuel distribution and pattern of

burning, and hence the distribution of temperature in the primary-zone eux. The parameter

that is often used to describe this eect is the PF, and is defined as [31]:

Tmax T4
Pattern Factor = . (4.5)
T4 T3

Norgren et al. found that apart from the spill return nozzle, all the nozzles tested had a wide

range of PFs. For instance, the pattern factor range for the spill return, duplex and simplex

nozzles, were 0.1, 0.58 and 0.68 respectively [44]. A wide range is undesirable because it indicates

improper mixing and distribution of the spray with changing power conditions. Furthermore, a

wide PF range increases possibility of the eux gases damaging the turbine blades.

Another nozzle that performed well, was the airblast nozzle. At ground idle conditions, the

airblast nozzle had a large pattern factor (for power conditions without blowout), due to poor

atomization performance. However, at higher power conditions(such as the high end of low

power conditions and above), the PF improved due to the airblast nozzle being able to provide

better mixing of fuel and air, and smaller droplet sizes. Furthermore, above ground idle and

some low power conditions, the airblast nozzle produced the lowest PF, of all the nozzles tested.

More specifically, in terms of individual performance, the spill return and air assist nozzles had

a PF that was 46% lower, and 52% higher than the duplex nozzle at ground idle conditions. At

cruise conditions, the spill return and air assist nozzles had the same and 36% higher PF than

the duplex nozzle respectively, while the airblast nozzle had a PF that was 22% lower. This
Chapter 4. Fuel Nozzle Technology and Performance Characteristics 38

clearly indicates the strength of the spill return nozzle at ground idle and low power conditions.

However, due to the improved mixing aorded by the airblast nozzle, it is able to outperform

the spill return nozzle.

4.2.3 UHC and CO emissions

In terms of the emissions of UHC and CO, Norgren et al. found that the spill return nozzle

managed to outperform the duplex and air assist nozzles for almost all the power conditions.

However, at ground idle conditions, while the spill return nozzle had an emissions index of UHC

(EIUHC) that was 49% lower than the duplex nozzle, it had an emissions index of CO (EICO)

that was 30% higher [44]. This seems to indicate that while all the fuel initially combusted, not

enough time was provided for the CO to fully oxidise to CO2 , making it an issue of combustor-

nozzle optimization. For the air assist nozzle, while its performance was slightly better than

that of the duplex nozzle, it was consistently worse than the spill return nozzle for all the power

conditions. At the high end of the idle power, cruise and takeo power conditions, the airblast

nozzle had the lowest EICO and EIUHC of all the nozzles tested. For instance, at high altitude

cruise, the airblast nozzle had an EICO that was 5, 2.5, 1.3 lower than the duplex, air

assist and spill return nozzles respectively [44]. The opposite was the case for idle and low power

conditions, as for cases when the nozzle did not blowout, EICO and EIUHC were the highest of

all the nozzles that were tested.

As in the case of the PF, the spill return nozzle performs the best at idle and low power

conditions, after which it is eclipsed by the airblast nozzle, due to its ability to provide good

mixing and spray distribution characteristics. The primary reason that the splash cone nozzle

performs well for these power conditions, is due to the higher air pressure drop available. The

air assist nozzle was not able to match the performance of the airblast nozzle, because it uses a

much smaller amount of air.


Chapter 4. Fuel Nozzle Technology and Performance Characteristics 39

Figure 4.7: Placement of a fuel nozzle in a combustor (Prociw et al. [25])

4.3 Fuel Nozzle Coking

4.3.1 Eects of Nozzle Coking

Fuel nozzle coking is an important issue for small gas turbines, because they not only aect

emissions, but also their reliability. Fuel nozzles are prone to coking due to the increasing

trend of modern combustors using higher combustor inlet temperatures, leading to increased

heat loadings on the nozzle. Furthermore, in most combustors the nozzle front face is directly

exposed to the heat of combustion as shown in Figure 4.7, and has a significant eect on the

nozzle thermal properties. For instance, a study by Myers et al. found that at cruise conditions,

the heat flux entering the nozzle face from flame radiation, was more than 20 that absorbed by

conduction and convection through the nozzle fuel stem [42]. This makes thermal management of

the the frontal area of the nozzle of key importance. Other engine conditions can also contribute

to internal coke formation; during engine deceleration and hot shutdowns, fuel flow in the hot

passages can be either greatly reduced or cut-o, contributing to coking.

The eect of partially or fully blocked metering passage changes, depending on the combus-

tor and even the power condition. Generally, the eect of blocked metering passages is most

significant at starting power conditions [54]. Figure 4.8 shows a spray patternation comparison

for an airblast nozzle before and after an engine test, demonstrating the eect of a partial block-
Chapter 4. Fuel Nozzle Technology and Performance Characteristics 40

0
12.59% 12.7% 12.3% 5.05%

10.1% 13.48% 7.52%


23.4%

270 90

12.9% 15.02% 23.4% 7.68%

13.01% 9.77% 7.08%


180 12.9%

Figure 4.8: Eect of coking on spray patternation performance (Prociw et al. [25])

age on the performance of the nozzle. We can see that in certain locations the amount of fuel has

changed quite drastically, aecting the spray uniformity, and causing streaks in the spray, that

could aect the local fuel air ratio in the combustor. The streaking occurs because as some fuel

metering holes become obstructed, more fuel is being forced through the remaining nozzles or

metering holes. If this were to occur at takeo, where the fuel air ratios are already much higher

than what the turbine experiences during the rest of the cycle, it would only serve to magnify

this eect, causing dramatic hot end damage. On other occasions where coking is an issue, but

not serious enough to require replacement, they could have other undesirable side eects such

as increased smoke production, large increases in fuel pressure drop, combustor carbon growths

and torching on startup [25].

Within the class of nozzles in use for small combustors, airblast atomizers are generally less

susceptible to coking, since they use larger passage sizes to assist with the atomization process.

On the other hand, pressure nozzles are more prone to coking since the passage sizes can often

get as small as 0.3 mm. While there is no absolute evidence on how large a passage size needs

to be in order to reduce coking, a study by Mitchel et al. of pressure swirl nozzles of dierent

internal passage sizes with contaminated fuel discovered a number of aspects. For pressure swirl

nozzles they found that there are three major locations where blocking can occur: the inlet slots

(Dp ), nozzle orifice (D0 ), and the swirl chamber (Ds ) [40]. Of these three locations, the site

most likely to coke was the inlet slot. Furthermore, Mitchel et al. found that if the minimum
Chapter 4. Fuel Nozzle Technology and Performance Characteristics 41

passage size was greater than 0.45 mm for any of the above three site, coking potential was

greatly reduced [40].

Spill return nozzles that are designed to atomize low fuel flow rates, do not have the same

coking issues that similarly sized pressure swirl nozzles have, due to their larger internal passages

and improved mixing that they allow. Studies by Norgren et al. found that the spill return

nozzle had almost no carbon deposit formation inside the nozzle [44]. The only time coke

deposition on the spill return nozzle occurred, was during tests at idle power condition, where

a soft carbon deposit was found on the surface of the nozzle. This issue can be easily fixed, by

having a central airflow around the nozzle to wash the coke o.

4.3.2 Methods of Reducing Coking

A number of dierent methods have been developed in order to ensure that coking in nozzles

can be reduced. One very useful method of providing insulation for the fuel atomizer is through

using air gaps. In some instances where the air breaks down due to excessive thermal heat,

the air can be replaced by other materials generally with a thermal conductivity less than
W
0.072 mK [58]. Certain materials that fit this description and have been tested in nozzles include

Manvilles Min-K , Duponts Kapton and Polymer Techs Poly Damp. The only issue with using

other types of materials is that should the insulation gap get punctured, the materials could

react with the combustion process in a negative manner. The application of air gaps can be

either in the fuel stem, or in the front face of the nozzle. In a fuel stem, this application is

generally done through the application of an annular passage around the internal fuel pipes

in the nozzle fuel stem. Generally, if the fuel stem is exposed to compressor air (T3) that

is generally < 800 K, a single annular air gap is used. For cases where the surrounding air

temperature is 800 F < T 3 < 1556 F , a double annular air gap is used [42].

Figure 4.9(a) shows the implementation of air gaps on the nozzle front face; its eectiveness

depends on the width of air gap in the front. For instance, using simple 1D equations, taking

into account conduction, convection and radiation, if an air insulation gap between 0.2 mm to

0.3 mm is used on the front face, it can help reduce the adjacent metal wall temperatures by as

much as 20 K 35 K at takeo conditions. In both the above situations, a normal aerospace


Chapter 4. Fuel Nozzle Technology and Performance Characteristics 42

Air flow
Insulation passage
gap

Low coking
material Low
coking
material

Retracted
Retracted
tip
tip
Insulating
Insulating
material
material

(a) (b)

Figure 4.9: Description of: (a) Nozzle with front insulation to reduce internal metal temperature rise,
(b) Nozzle with surrounding air gap to reduce internal metal temperature rise

grade metal such as stainless steel-416 (SS416) was used as the nozzle material. If for instance

another material was used, the results would be dierent.

In some instances, where the addition of the air gap may not be possible due to the shape of

the nozzle, the alternative often has been to have an airflow passage as shown in Figure 4.9(b).

An airflow passage that is optimally sized, can not only help the fuel inside the nozzle stay cool,

but may also help wash any coke deposition that may occur on the nozzle surface. Furthermore,

since the flame temperature can often gets so hot, the usage of an air gap is not advised since it

deteriorates due to the increased radiative exchange between the inner and outer metal around

the air gap [42].

Another method by which the temperature inside nozzles is reduced, is by using thermal

barrier coatings (TBCs) on the surface. TBCs are generally used in other parts of the combustor

like the liner wall, since they provide a significant reduction in temperature of the metal that

is coated, with little impact on design tolerances or materials. Certain TBCs that can be used,

are Partially Stabilized Zirconia (PSZ), Alumina, and Mullite [66]. Generally, the thickness

of the TBC coating is between 25 m to 75 m, and in the case of PSZ, one can reduce the

temperature between 12 K 27 K [55]. Certain issues with TBCs are that the coatings in

some situations can prematurely fail during service. This is due to the fact that TBCs are

applied using three layers of dierent properties and functions [66]. In some cases, the bottom
Chapter 4. Fuel Nozzle Technology and Performance Characteristics 43

Figure 4.10: Eect of bends and steps in flow path on fuel temperature (Stickles et al. [58])

layer can form surface rumples with thermal cycling, and can eventually lead to failure through

local separations and delaminations with the top layer [66].

In addition to make changes to the external design of the nozzle, there are dierent methods

in which coking potential can be reduced inside nozzles, through design conventions for the

wetted wall parts of the nozzle. A study by Stickles et al. found that an eective way to reduce

coking potential was increasing the fuel velocity in sectors of the nozzle that are expected to

get hot, by narrowing the fuel passages. Furthermore, they also suggested that sharp bends

and steps in the flow path be avoided. This is because steps in the flow path tend to create

stagnation zones where the flow temperature can get quite high, increasing the potential of coke

forming [58]. Figure 4.10 compares the dierence in temperature found by Stickles et al. for

cases with and without optimized fuel flow paths.

Stickles et al. also studied the eect of metal surface finish on coke deposition [58]. The

reasoning behind this was based upon the fact that through polishing the metal, any active

sites on the metal surface that can interact with the fuel to form deposits are eliminated. For the

metal SS347, they found that reducing the surface roughness from 3.1 m to 0.25 m reduced

the deposition rate by 26%. However, when a same reduction in surface finish was was done for

Waspaloy, the reduction in deposition rate was 16% [63]. The biggest issue with surface finish is

that the coke reduction rate, is highly specific to the material. Studies by Stickles et al. found

that for three dierent types of materials, with the same surface finish, the coke deposition rate

varied greatly for the same period of time.


Chapter 4. Fuel Nozzle Technology and Performance Characteristics 44

Table 4.1: Coking tests for dierent materials for use inside nozzles (Stickles et al. [58])

Weight Change Reflection


Rank Material Comments
(mg) Change
1 SS316 0.08 -0.48 Haze of Coke
2 PSZ 1 -0.41 Stained
3 SIO broke -11.59 Stained
4 MACOR broke -53.66 Stained
5 LAS broke -52.7 Stained
6 SS416 Passive 0.9 - Patchy Black
7 SS416 3.6 -63.15 Uniformly Black

Myers et al. did a follow up study where they studied dierent materials for use in the

wetted portion of the nozzle that could inhibit coke deposition. Table 4.1 shows the results of a

test by Myers et al., where coke deposition was studied [42]. In addition to studying the change

in weight due to coke deposition, the change in reflectance of the material was also studied using

an optical densitometer. The materials that were tested include Lithium Aluminium Silicate

(LAS), Fused Silica (SIO), Machinable Glass Ceramic (MACOR), Partially Stabilized Zirconia

(PSZ), SS416 Passive - SS416 exposed to hydrogen sulfide-doped helium, and SS316 - which was

the baseline, since it is generally used in fuel nozzles. In addition to metallic materials such as

SS316 and SS416, ceramics were tested because of their low thermal conductivity properties.

From Table 4.1 we see that SS316 performed the best of all the materials, with the lowest

weight change. Furthermore, Myers et al. found that the rate of coking for ceramic materials

such as PSZ, SIO and MACOR were much higher than for SS316. The reason for the success

of SS316, is due in part to its chemical composition. Metals that contain substances such

as Aluminium (Al) or Chromium (Cr), react with the oxygen in the fuel and form oxides on

the surface, which prevent the surface of the metal from further interacting and forming coke

deposits. This however only holds true when the fuel temperature is between 450 K 723 K,

since at higher temperatures, coke deposition occurs through pyrolysis, which is independent of

fuel and wall interactions. Since the temperature inside the fuel nozzle is almost always within

this range for small gas turbine combustors, the usage of these materials is bound to help reduce
Chapter 4. Fuel Nozzle Technology and Performance Characteristics 45

coke deposition [25].

Studies by Lai have found that materials that formed Al or Cr oxides had the lowest coke

deposition, followed by those that formed. Those that used Molybedium(Mo), Silicon (Si) or

Copper (Cu) based oxides, had deposition rates that were much higher [28]. The reason for

the dierence between dierent metals, had to do with how strongly the oxide adhered to the

surface of the metal, and how quickly the eroded oxide on the surface was replenished [28].

Accordingly as mentioned by Lai, Stickles et al. and Myers et al., the materials with the lowest

coking potential were metal with adequate Al or Cr content, and the ability to handle high

temperatures. Certain materials that fit this description were SS316, SS304, SS214, SS446,

Inconel 600 and Inconel 702.

4.4 Spill Return Nozzle Selection

Of all the nozzles that were tested by Norgren et al., the spill return nozzle was shown to be

one of the the best performing nozzles, particularly at low power conditions. The spill return

nozzle as shown in Figure 4.11, had a 100% combustion eciency at takeo, falling only to 92%

at ground idle conditions. The emissions of CO, UHC, and smoke for the spill return nozzle

at ground idle conditions, was the lowest of all the nozzles tested. The spill nozzle was able to

outperform the air assist and duplex nozzles because, in addition to producing a smaller droplet

size, the spill return nozzle also allows for slightly longer droplet residence times, allowing for

a more complete combustion process. On the other hand, at higher power conditions such as

cruise and take-o, the airblast nozzle performed the best. The ability of the airblast nozzle to

provide a greater degree of mixing, helped ensure that emissions stayed low. Furthermore, from

a coking point of view, Norgren et al. found that unlike simplex and duplex nozzles, the spill

return and airblast nozzles both had a much lower coking potential. The air assist nozzle was

not considered further, since in addition to being out performed by the spill return nozzle, the

need for an extra compressor to provide air, adds extra weight and complexity.

These factors make the airblast and spill return nozzles very attractive for use in gas turbines.

Given our initial interest in reducing emissions particularly at low power conditions, the fact
Chapter 4. Fuel Nozzle Technology and Performance Characteristics 46

Figure 4.11: Description of a spill return nozzle developed by RollsRoyce as shown by (Bayvel [6])(Rolls-
Royce [52])

that the most room for improvement in terms of emissions lies at idle and low power conditions

and coupled with the lack of proven existing design studies on spill return nozzles, the latter

was chosen for further in depth study. The goal with the in-depth study was to study the

incorporation a spill return nozzle into an existing gas turbine hybrid nozzle, where the design

of the airblast circuit would stay the same, while the primary circuit would be replaced with

a spill return nozzle. This includes creating a design for the spill return nozzle, along with a

control mechanism for our specific power conditions.


Chapter 5

Nozzle Design Methodology and

Requirements

5.1 Introduction

Given that the spill return nozzle is an extension of pressure swirl nozzles, the methodologies

used in developing the spill return fuel nozzle design, were based upon two dierent existing

methodologies. The first methodology used was based upon one developed by Bayvel [6], and

was used to calculate the sizes of the nozzle internal features. The second design methodology

used was developed by Bowen [7], to design the control system of the nozzle. Each methodology

is described in more detail in the following Chapters.

5.2 Nozzle Feature Design Methodology

5.2.1 Nozzle Design Parameters

The spill return fuel nozzle can be described by 13 dierent parameters, as shown in Figure

5.1(a). In Figure 5.1, Dp and Lp are the inlet port diameter and length respectively, Ds and Ls

are the swirl chamber diameter and length, D0 and L0 are the orifice diameter and length, Dsp

is the spill line diameter, is the inlet port angle with 90 being vertical, Lt is the length of

the trumpet, t is the trumpet angle with respect to horizontal and C is the angle of the swirl

47
Chapter 5. Nozzle Design Methodology and Requirements 48

(a) (b)

Figure 5.1: (a) Description of spill return design parameters, (b) Description of aircore parameters for
spill return nozzle

chamber conical section. One important parameter that was not shown in Figure 5.1, is the

number of inlet ports; it is described through the parameter n. An implicit assumption which

is made for the above design is the fact that the swirl chamber shape is fixed combination of a

cylinder and a cone. The determination of these parameters are done in combination with the

design methodology by Bavyel and sensitivity studies performed by Kim et al. [26], Lefebvre [32]

and Xue et al. [65].

For the purpose of this study, the fuel used was Jet-A1. The properties of the fuel used

were; L was 784 Kg , the kinematic viscosity () was 2.2 10 6 m2 , and the absolute viscosity
m3 s

() was 1.725 10 3 Ns
m2
[6]. All the parameters except t , C , lt and were designed using

the methodology by Bayvel et al.. The remaining parameters were determined using sensitivity

studies by Xue et al., Kim et al. and Lefebvre.


Chapter 5. Nozzle Design Methodology and Requirements 49

5.2.2 Spill Return Nozzle Design Methodology

As mentioned earlier, the nozzle design methodology used in this project was developed originally

by Bayvel for pressure swirl nozzles. This methodology was used because it has the ability

to do viscous calculations in determining nozzle design features, and also because the design

methodology gives a novel analysis of the axial and tangential velocities, along with the air-core

radius at various cross-sections inside the nozzle. This information is vital for locating and sizing

the spill line, so that we can avoid issues of the aircore interacting with the spill line passage.

Finally, the analytical nature of the methodology allowed for a robust extension to spill return

nozzles.

5.2.3 Fundamental Equations

For the purpose of this thesis, the design methodology developed by Bayvel for simplex atomizers

was extended to spill return nozzles. Equation (A.1) to Equation (A.4) in Appendix A, are the

set of equations that describe the performance of the nozzle from an inviscid perspective. The

equations were extended by incorporating the spill ratio where:

mt
e= , (5.1)
ma

where mt and ma are the total and atomized fuel flow rates respectively. Equation (A.1)

relates the nozzle geometric constant (K), to the eciency of filling (); where describes

what fraction of the nozzle orifice is occupied by the liquid sheet. Equation (A.2) describes

the relationship between the discharge coecient and the , K and e, Equation (A.2) relates
rr
the fraction of the aircore size S = r0 , to , K and e. Equation (A.5) to Equation (A.7) are

the second set of equations that take into account the viscous eects, and used to perform

the viscous correction. Furthermore, unlike the analytical equations used for inviscid analysis,

Equation (A.6) and Equation (A.7) which are used for the viscous analysis are semi-empirical

relationships developed for a wide range of Reynolds numbers. More details on the methodology

and some of the intermediate steps are given in [6].

It should be mentioned that a number of assumptions were made to make the methodology
Chapter 5. Nozzle Design Methodology and Requirements 50

1
6 7
Fuel%Nozzle%Design%Process%
Start Should#have#
Use#inviscid#results#to#calculate Re,#
Do,#dp,#Ds,#Ls,#
friction#coefficient#and#viscous#K.
Lp,#l#and#Beta

Design%Completion

4
2 8
Use#inviscid analysis#to#find#K#and#
Do#you#know#the# Yes Mu#from#spray#cone#angle.# Does viscous#K#have#a#
required#FN#and Yes
Geometric#coefficient#and#discharge# >#than#2%#difference#
Spray#Cone#Angle?
coefficient#respectively.# than#inviscid#K?

No
No
3
5
Determine FN#and#Spray#cone#angle#
requirement#from#application# Use#K#and#Mu#to#determine#nozzle#
requriements.#FN#means#you#know# design#parameters
flow#rate#and#Inj.#Press

9
Check#geometric ratios#
No Yes
to#ensure#they#are#
within#limits.

Figure 5.2: Spill nozzle design process

simple and robust enough to work. The assumption made were that: the angular momentum

is constant inside the nozzle swirl chamber, there is no gravity, the flow in the nozzle is stable

and axi-symmetric and there is no radial velocity component. The primary dierence between

Equations (A.1) to (A.4) for simplex and spill return nozzles, is that for spill return nozzles,

instead of only K appearing in the relationships for simplex nozzles, we now have Ke. This

highlights the main feature of the spill return nozzle, where a nozzle can be physically designed

for a give FN, and spilling the fuel allows us to eectively make the nozzle act like a nozzle of a

smaller FN. This is useful situations when the fuel flow rates are low, because small flow rates

require small FN nozzles, in order to get good atomization performance.

5.2.4 Design Process

The design process for calculating the design features of the spill return nozzle was an iterative

one. Figure 5.2 shows the process that was followed, in order to perform a complete design

iteration.

When starting the design process, it is important to know the FN and spray cone angle

() to which the nozzle needs to be designed. These two values are required inputs to our

design process, and come from requirements such as ensuring adequate fuel spray mixing, and

low emissions. Knowing , we then calculate the appropriate K and . The calculations for
Chapter 5. Nozzle Design Methodology and Requirements 51

the nozzle internal parameters are then calculated using the equations as set by Bayvel [6].

After the initial calculations are made, a second set of calculations are conducted to assess the

eect of viscosity. Using Equations (A.5) to (A.7), a viscous based geometric coecient (K )

is calculated. K is then compared to the inviscid K, and if the dierence is more than 2%,

the calculation is redone with K as new K. This process is repeated until the dierence drops

down to 2%.

After the viscous correction is done, the design features of the nozzle are checked to ensure

that they fall within certain design ranges, which were found to be stable from experimental

studies. If the features satisfy this check, the design process is then complete. If not, sensitivity

studies on the nozzle features are done to understand how to satisfy the design ratios, while

ensuring that the design still meets performance requirements.

5.2.5 Spray Uniformity

Spray uniformity (I) is an important design characteristic for fuel nozzles. Non-uniformity in

the fuel spray is undesirable since it can cause fuel streaks in the spray aecting emissions, or

in even worse conditions, damage hot end components. Similar to the pattern factor, the spray

uniformity is defined as:


qmax qmin
I= 100%, (5.2)
q
where qmax , qmin and q are the maximum, minimum and average liquid densities respectively. I

is generally studied by measuring it experimentally at a certain distance from the nozzle orifice,

at given angles along the perimeter of the spray. Since I is a function of distance from the nozzle

orifice, it is vital that all the studies be conducted at the same distance. Spray uniformity is

aected by the number of inlet ports to the swirl chamber (n), and the internal swirl chamber

features, through the degree of divergence (C), where it is defined as:

R
C= . (5.3)
r0

For gas turbine combustors, an I < 20 30% is usually required, as higher values of I end up

with sprays that are not optimal [6]. An expression that is often to study spray uniformity in

the design process, is the following semi-empirical equation developed by Bayvel et al. [6]:
Chapter 5. Nozzle Design Methodology and Requirements 52

23.7
I= 0.15 . (5.4)
2.7 1.34 1.1 Ls
(C) (n) () Ds

5.2.6 Nozzle Design Check

The nozzle design parameters were checked by ensuring the following dimensionless groups
Ap D s Ls L0 Lp
Ds D0 , D0 , Ds , D0 and Dp were kept within certain ranges. Couto et al. did a study where they

built and tested a pressure swirl nozzle, and suggested conventions for various design ratios.
Ls
They suggested that Ds should be kept as small enough to reduce frictional losses between

the fluid and wall of the nozzle. However, a limiting value was needed to achieve the liquid

flow stabilization and the generation of a uniform vortex sheet, and said that the ratio needed

to be higher than 0.5 [56]. Simultaneously, studies by Kim et al. into pressure nozzles with
Ls Ls
0.7 < Ds < 3.06, found that the air core inside the nozzle become unstable when Ds > 1.06 [26].
Ls
Thus, the range of acceptable Ds ratio was set between 0.5 and 1.06.
Ap Ds
Similarly, valid ranges for Ds D0 and D0 were chosen as 0.19 1.21 and 1.41 8.13 respec-
Ap
tively, as recommended by Couto et al. The range for Ds D0 was set as such because too high
Ds
ratios resulted in low and increased liquid film thickness. Furthermore, for D0 it was found
Ds
that increases in D0 results in a decrease in the film thickness, in the discharge coecient, and
L0
in the spray angle. The ratio D0 also needs to be kept low as increasing the orifice length can

reduce the both and the U coming out of the orifice, potentially aecting atomization. While

there is no range set for this parameter it is generally set to be 23 D0 [46] - or slightly larger if
Lp
manufacturing tolerances do not allow. Finally, the ratio Dp was set to be larger than 1.3, since

a short tangential inlet passage could generate a diuse discharge, leading to an unstable spray.

5.2.7 Methodology Validation

Figure 5.3 shows the eect of increasing e, on the performance of the nozzle. The arrows indicate

how the various parameters change, when e increases. Figure 5.3 was calculated assuming a

fixed output flow for a changing e. This means that as e increases, the amount of fuel that goes

into the swirl chamber increases, with the dierence between total and atomized fuel, being the
Chapter 5. Nozzle Design Methodology and Requirements 53

100 100

Nor maliz e d ( % of max imum f or e =2)



90 90

Nor maliz e d and ( % of max )


80 80

70 70

60 60

50 50

40 40

30 30

20 20

10 10

0 0
0 10 20 30 40 50 60 70 80 90 100
Nor maliz e d K ( % of max )

Figure 5.3: Normalized results from the extended methodology, showing the eect of spill ratio (e) on
spray cone angle (), discharge coecient (), and filling eciency (); spill ratio goes from 1 to 2 in
the direction of the arrows

quantity that is spilled.

As e varies from 1 (No fuel spill) to 2 (50% fuel spill), we see that increases; at the same

time, and decrease. These trends are in-line with both expectations, as it is well known that

with an increasing e, there is always an associated increase in [6,32]. The reason this happens

is because with a higher e, the amount of swirl velocity at the back wall of the swirl chamber

is much higher, than when we have a lower e. Furthermore, spilling the fuel from the back of

the swirl chamber only aects the axial velocity component of the fuel. This particular aspect

has been demonstrated in a number of studies by Rizk et al. [49] and Mang et al. [37]. Thus,

the eect of the higher swirl velocity results in a larger air core inside the swirl chamber, which

results in a thinner liquid sheet at the orifice.

This is the main reason why reduces, as we increase e. The spray cone angle change can

be explained as follows. The spray cone as shown in Equation (A.4) essentially describes:
v
tan = , (5.5)
2 u

where u and u are the average axial and tangential velocity components of the liquid sheet
Chapter 5. Nozzle Design Methodology and Requirements 54

100
Experimental Values

Norm aliz e d D 0 ( % of max e xp. value )


NonViscous Calculation
Viscous Calculation
90

80

70

60

50

40
20 30 40 50 60 70 80 90 100
Nor maliz e d FN ( % of max)

Figure 5.4: Comparison of normalized D0 values calculated from the extended design methodology,
against experimental data

leaving the nozzle orifice. Since the axial velocity component drops due to spilling the fuel, and

we have a higher tangential component as was mentioned before, this results in the spray cone

angle increasing as we start to spill more fuel from the nozzle.

Attempts were also made to compare the results of the design process against experimental

data of Radclie [46]. Figure 5.4 compares the normalized nozzle orifice diameter values that

were calculated for a given FN, against the respective experimental values for both viscous and

non-viscous calculations. While the results for the inviscid calculations have a relatively large

error, this is not the case for the viscous calculations. We see that the eect of friction, is to

reduce the diameter of the aircore and , thus bringing down the inviscid results. While the error

for the inviscid calculation were relatively large -on average of about 19.8%, the error with the

viscous eects taken into account, was on average 2.8%. Furthermore, we see that the viscous

calculations not only work well from a qualitative perspective, but also from a quantitative one.

Similar comparisons were conducted for Dp and the aircore diameter. Dp followed a similar

trend to D0 , where the average error was 6.2%. The error however for the aircore diameter, was

much higher. For FNs of 1, 1.1, 1.3, 1.5 and 1.7, the error was 35%, 31%, 23%, 14% and 7%

respectively. The methodology largely over predicted the aircore diameter for small FN nozzles,
Chapter 5. Nozzle Design Methodology and Requirements 55

100
0% Spill
10% Spill
30% Spill
90 50% Spill

Nor maliz e d X ( % of max)


70% Spill

80

70

60

50

40
0 10 20 30 40 50 60 70 80 90 100
Nor maliz e d Z ( % of max)

Figure 5.5: Normalized eect of geometric coecient (Z) and spill percentage, on the aircore size

only starting to approach experimental testing values by Radclie at larger spill percentages.

However, since we are using this information to find the minimum passage size, this means

that one will always end up with spill diameters larger than what is required. This may not

necessarily be a bad thing from a design perspective.

5.2.8 Nozzle Spill Line Sizing

The sizing of the spill line was done by calculating the aircore size at the rear wall of the swirl

chamber, as shown in Figure 5.1(b). For the central spill line case, the size of the spill line was

determined by summing the aircore area(under the assumption it is a cylinder), and a percentage

of the inlet port area. This percentage is proportional to the amount of fuel that we want to

spill. For the annular and the o-central cases, since we are avoiding the aircore altogether, the

size of the spill passages was based upon a percentage of the inlet port area.

Figure 5.5 is an example which shows the normalized ratio X ( Daircore


D0 ), changing with

various spill percentages. Figure 5.5 is generated by simplifying and K from the extended
Ap
Bayvel methodology in the form of X and Z, where Z = Ds D0 . This is done to help obtain

a more intuitive understanding of the relationship between the spill percentage and the size of
Chapter 5. Nozzle Design Methodology and Requirements 56

100
Lefebvre Calc.
Couto et al. Calc
95 Experimental Values

Nor maliz e d SMD ( % of max)


90

85

80

75

70

65

60

55
30 40 50 60 70 80 90 100
Nor m aliz e d Pr e ssur e D r op ( % of max)

Figure 5.6: Comparison of SMD calculated by two dierent methods against experimental data

the aircore.

5.2.9 Nozzle SMD Calculation

The original methodology by Bayvel does not have the capability to calculate SMD of pressure

or spill return nozzles. However, the methodology does calculate a number of parameters in

the intermediate steps, that can be used to get an accurate value for the SMD. Subsequently,

equations based on the basic mechanisms of atomization for pressure nozzles were used. Two

reasons that motivated this were that no existing semi-empirical relationships for calculating

SMD for spill return atomizers were found in literature. Furthermore, the validity of extending

existing simplex nozzle empirical relationships towards the spill return nozzle were unknown.

Figure 5.6 compares the results of two dierent methods of calculating the SMD, against

experimental data of Couto et al. for a pressure swirl nozzle [56]. The two methods shown in

Figure 5.6 refer to analytical relationships that were developed by Couto et al. and Lefebvre et

al. The equation of Couto et al. finds the SMD by using the sheet thickness at the orifice to

calculate the ligament diameter, and then subsequently calculating the SMD assuming Rayleigh

breakup. The second method of Lefebvre accounts for atomization in two terms, namely:
Chapter 5. Nozzle Design Methodology and Requirements 57

110

Total F
100 low

Nor maliz e d M / P ( % of max)


90

Sp
80 ill
F low
70

Spill Line Closed


60

50

40

30
ow
ut Fl
20
Outp

10

0
15 20 25 30 35 40 45 50 55 60
Ps
( %)
P

Figure 5.7: General spill nozzle control plot

0.25 0.25
2L 0.25 L
SMD = 1050 (t cos ) + 2400 (t cos )0.75 . (5.6)
A PL2 A PL

The first term as shown in Equation (5.6), accounts for the generation of surface instabilities,

due to the combined eects of hydrodynamic and aerodynamics forces. The second term takes

into account the conversion of surface perturbations into ligaments and then drops [32].

Of the two methodologies, we see that the equation developed by Lefebvre performs better

over the given test range, and was subsequently used to calculate the SMD for the spill return

nozzle. Despite Equation (5.6) not having e in it, it is able to account for the spill percentage

implicitly, through the sheet thickness, spray cone angle, and the nozzle pressure drop, which

are provided through calculations done in the methodology by Bayvel.

5.3 Nozzle Control Methodology

The nozzle control method used in this project was a methodology developed by Bowen [7] specif-

ically for central spill return nozzles. Detailed information on the methodology and equations

used are shown in the paper by Bowen [7]. Certain assumptions made with the methodology
Chapter 5. Nozzle Design Methodology and Requirements 58

110

N or m a l ized F u el F l ow R a t e P er N oz zl e( %)
MAtomized 110
Total F MAtomized

N or m a l ized F u el F l ow R a t e P er N oz zl e( %)
100 lo w MSpilled
Total F
MTotal
100 low MSpilled
90 MTotal
90
Spi
80 ll F S pi
low 80 ll F
low
70
70

Spill Line Closed


60

Spill Line Closed


60

50
50

40
40

30
low 30
Output F ut Flo
w
20 Outp
20

10
10

0
40 45 50 55 60 65 70 0
85 86 87 88 89 90 91 92 93
Ps Ps
(%) (%)
P P

(a) (b)

Figure 5.8: Spill control plot for a single power condition for: (a) Small spill diameter (b) Double (a)
spill diameter

that Bowen developed, include assuming that the axial and radial velocities are small at the

back wall of the swirl chamber. This leaves the tangential component of velocity to be the most

significant one at the back wall. While this assumption is true for low spill percentages, this

is not the case for high spill percentages. Thus, the methodology becomes less accurate as the

turn-down ratio increases.

Figure 5.7 shows a plot that is used in the process of developing a control system for the

nozzle with a certain spill percentage range. On the x-axis we have the normalized pm
P
, which

is also known as the nozzle FN (normalized), and on the y-axis we have the percentage of spill

pressure (Ps ) to the injection pressure (P ).

The input values needed to generate Figure 5.7, are the design features and the spray cone

angle of the nozzle. Once that is known, the equations of Bowen are evaluated for a range of spill

percentages. For cases where we already know the spill percentage, we can directly calculate

the point on the figure we are interested in. In Figure 5.8 we have three lines representing the

total, spilled and atomized fuel flow rate per hour for each nozzle; with low spill percentages on

the right side of the figure.

While Figure 5.7 is a control plot that is specific to the nozzle design, it is not for a specific
Chapter 5. Nozzle Design Methodology and Requirements 59

power condition. To understand how to control the nozzle for a given power condition, we need

to multiply the non-normalized FN by the square root of fuel injection pressure; where the

pressure needs to correspond to the power condition that we are interested in. Doing this allows

us to generate Figure 5.8, where we have assumed a certain fuel supply pressure. While the

x-axis remains the same, y-axis variable changes to net fuel atomized per hour (normalized).

At this point, once we select the amount of fuel we want to atomize, the spilled and total fuel

are automatically determined. This process is then repeated for all the power conditions of the

combustor.

One important aspect when determining a control mechanism for the spill nozzle is the width

of spill line. The choice for the spill line width ranges from a minimum which is set by the spill

percentage, to a max determined by the size of the nozzle swirl chamber. Figure 5.8(a) depicts

the control mechanism for the minimum spill diameter (case-A), while Figure 5.8(b) is for a

spill diameter twice the minimum passage size (case-B). Case-A has a number of advantages
Ps
over case-B, in that for the same spill percentage range, we have a much larger P range to
Ps
control the spill line. For case-B, the spill percentage range is covered over a much smaller P

range. This makes control of the nozzle more dicult since a small change in the pressure ratio

might result in a relatively large spill percentage change. Furthermore, for case-A we need a

large pressure dierence in order to obtain a given spill percentage, as shown in Figure 5.8(a).

The opposite is true for case-B, where a much smaller pressure dierence is needed. The spill

passage size used in the design analysis was always the minimum passage size as determined

from the spill percentage, since it allowed for the most control.

5.4 Model Combustor Test Conditions

For the purpose of this project, an existing reverse-flow combustor which originally uses a hybrid

airblast nozzle was chosen. This particular combustor was chosen as the test case for application

through interactions with our industrial partner. The properties of this particular combustor

are shown in Table 5.1. The gray cells in the table pertain to actual combustor data, while the

rest of the cells were determined through research from other papers such as Spadaccini [57]
Chapter 5. Nozzle Design Methodology and Requirements 60

Table 5.1: Model combustor test conditions. Coloured cells indicate factual data from [64]; Other cells
indicate data developed to assist with the project

Thrust Level (% Max) 7% 30% 70% 100%

T3 (% Max) 63 78 93 100
Combustor
Fuel/Air Ratio 51.7 60 84.5 100

(% Max)

Fuel Flow Rate(% Max) 14.2 31 70 100

Primary Circuit(% Max) 65 100 100 100


Secondary Circuit(% Max) 2.9 16.4 63.2 100

and Mitchell [41].

The spill return technology is only being applied to the primary circuit, to improve atom-

ization performance at starting and low power conditions, and also reduce the coking potential

of the existing primary circuits, due to their small passage sizes. As such, the secondary circuit

in the existing dual circuit design was maintained as is. When designing the nozzle for our

combustor, there are a number of parameters that are available to us:

Combustor FAR
Primary Injection Pressure
Secondary Injection Pressure
Spill fraction of fuel

For the purpose of this project, the first three parameters were kept constant, and only the

spill percentage was allowed to change. This allowed us to understand how the application of

the spill return nozzle to an existing reverse flow combustor would look like. In addition to

maintaining the FAR, the FN of the primary and secondary circuit was also maintained, where

the primary and secondary circuit FN were maintained at 1 and 5.7 [64].
Chapter 6

Spill Return Nozzle Design For Model

Combustor

6.1 Introduction

This chapter focuses on the application of the spill return nozzle to our model combustor. As

mentioned in Chapter 5.4, this engine has a dual circuit nozzle, where the primary circuit is

a pressure nozzle, and the secondary circuit is a airblast circuit. The focus here is to only

change the primary circuit, and leave the secondary circuit as is. For the purpose of describing

improvements of the new nozzle design, a nozzle with a FN of 1 (normalized FN = 25%), was

used as a baseline.

6.2 Nozzle Sizing

When sizing the nozzle, there are 14 dierent parameters that needed to determined as men-

tioned in Chapter 5.2.1. Of these parameters, the ones with the largest impact on performance

are the orifice size (D0 ), the inlet port diameter (Dp ) and the spill port size (Dsp ). The deter-

mination of D0 is primarily influenced by the nozzle geometric constant (K), and the pressure

drop across the nozzle ( P ). As such, these parameters were the first to be solved. The sizes

of the remaining parameters follow the design constraints or the nozzle geometry.

61
Chapter 6. Spill Return Nozzle Design For Model Combustor 62

100
5%

N ormal i zed Total Fu el S p i l l ed (% of max)


90 7%
10%
15%
80

70

60

50

40

30

20

10

0
0 10 20 30 40 50 60
S p i l l Percen tage (%)

Figure 6.1: Normalized total fuel spilled (%) with spill percentage for a 52 minute period

6.2.1 Eect of Spill Flow on Nozzle Design Choice

When choosing a design for the spill return nozzle, we took advantage of the fact that the spill

return nozzle allows one to physically design the nozzle with a given FN, and then reduce the

FN of the nozzle by spilling a certain percentage of the fuel. Thus, the choice of the new nozzle

FN was aected by three main factors. The first pertained to passage sizes that had low coking

potentials; the second had to do with the eect of spill percentage on spray cone angle, and the

last had to do with the amount of fuel that would need to be spilled and consequently handled,

for a new nozzle design.

Figure 6.2 shows the eect of increasing the normalized nozzle FN on the spill percentage, for

a nozzle designed to have a spray cone angle of 90 , and after the viscosity correction was taken

into eect. While the FN is generally a number, ranging from 1 for small fuel flow rates to a large

number 4 5 for larger flow rates, it is a percentage here because it has been normalized. The

blue y-axis on the left describes the increase in normalized orifice size obtained, incomparison

to a nozzle with a normalized FN of 25%. The x-axis describes the new normalized nozzle FN,

while the red y-axis on the right describes the percentage of fuel that needed to be spilled, in

order to make it perform like a nozzle with a normalized FN of 25%.

Figure 6.1 shows the total fuel that is spilled for a 52 minute period, for nozzles with dierent
Chapter 6. Spill Return Nozzle Design For Model Combustor 63

80 200
D0

N ormal i zed Ch an ge i n D 0 (% of max)


Spill Percentage
70

60 150

S p i l l Percen tage (%)


50

40 100

30

20 50

10

0 0
40 50 60 70 80 90 100
N ormal i zed FN (% of max)

Figure 6.2: Percentage increase in normalized D0 (%) and spill percentage (with respect to the existing
nozzle), for an increasing normalized nozzle FN (%).

spill percentages; the spill percentage directly correlating to a nozzle of a given normalized FN.

The 52 minutes were chosen by doubling the ground idle time (as a safety factor), as mentioned

in the ICAO regulations [22]. From Figure 6.1, we see that at spill percentages past 50%,

the total fuel being spilled is close to the amount of fuel that the combustor uses at cruise

or take-o conditions. This is an issue for small gas turbines, because the spill percentage is

directly proportional to the extra pumping power requirements. If the goal of implementing a

spill return in an existing engine is to be met, only small increases in pumping power can be

expected. Thus, after interactions with our industrial partner, only designs associated with spill

percentages less than 45% were set as the most viable options.

Keeping this in mind, several nozzle options were studied, as shown in Figure 6.2. We see

that the biggest improvement in terms of the percentage increase in normalized D0 versus spill

percentage, occurs for normalized FN < 50%. This aspect is clear from the changing slope of the

blue line versus increasing FN in Figure 6.2. Given the constraint of 45% spill percentage, the

design chosen was one with a normalized F N = 43%. With this choice, we get a design where

D0 is 15% larger than the existing design nozzle orifice, while still ensuring that the increasing

in pumping requirements remain low at 30%. Despite not being shown yet, a additional benefit
Chapter 6. Spill Return Nozzle Design For Model Combustor 64

160 8 Inlets
6 Inlets
4 Inlets

Ci r cu mf e r e n t i al Un i f or mi t y ( I %)
140 3 Inlets
2 Inlets

120

100

80

60

40

20

0
0 10 20 30 40 50 60 70 80 90 100
Nor mal i z e d At omi z e r D i v e r ge n c e ( C , % of max )

Figure 6.3: Eect of circumferential uniformity (I) on the normalized degree of divergence(C), for
various inlet passages.

with this design choice was the fact that all the other internal nozzle features were now larger

than the 0.45 mm Mitchel et al. found in their studies [41]. The studies leading to this aspect

are shown in the following chapters.

6.2.2 Spray Uniformity

While spray uniformity was tested for all power conditions, only the results for the 7% power

condition are shown in Figure 6.3. Given the low nozzle P , the 7% is the power condition is the

most likely power condition that would be exposed to spray uniformities. Figure 6.3 shows the

eect of the number swirl chamber inlet on spray uniformity, for our design. From Figure 6.3 we

see that for each line corresponding to the number of inlets, there is a certain value of normalized

C, after which the value of I falls within the acceptable range of less than 30%. For our particular

nozzle design, the value of normalized C is 99%. Ideally, when choosing the number of inlets,

we need to ensure that we have as many inlets as possible to allow for uniformity. However, we

also need to ensure that the size of the inlet are not so small that they become a coking hazard.
Chapter 6. Spill Return Nozzle Design For Model Combustor 65

Ls 72% Lt 37%

Ds 100% t 61%

L0 17.5% c 50%

D0 19.7% 100%

Lp 30% Dsp T BD

Dp 21% n 2

Table 6.1: Normalized sizes of nozzle features(%)


Figure 6.4: Description of nozzle fea- for a nozzle with a normalized FN of 43%. Features
tures are normalized against Ds or

For our particular case, the size of the inlet ports ranged from being 53% larger with 1 inlet

port, to 9% larger for 2 inlet ports, to 12% smaller for three inlet ports, than the minimum

passage size found from the coking studies of Mitchel et al. Taking into account the three main

factors as mentioned in Chapter 6.2.1, the best choice was the design with 2 inlets ports, with a

diameter that was 9% larger. This choice ensures that the nozzle is designed to provide a spray

that has a low I, while also maintaining a low coking potential. As Dp is a function of the nozzle

FN, the larger FN is, the larger Dp is for a fixed number of ports. This was another reason

why a nozzle with a normalized FN of 43% was chosen; as it was the design which required the

smallest increase in spill percentage, to reach the targeted minimum passage size. Given that

D0 and Dp are now sized, the rest of the parameters are sized using the equations of Bayvels

methodology, and are shown in Table 6.1. The only parameter that has yet to be sized is the

spill diameter Dsp .


Chapter 6. Spill Return Nozzle Design For Model Combustor 66

6.2.3 Spill Line Geometries and its Size Eects

The final design parameter to be sized, is the nozzle spill diameter, Dsp . The four dierent

methods to spill the fuel are shown in Figure 6.5. The geometry that is studied most often,

is the central spill method. In this design, the spilled fuel returns through a passage in the

center of the rear swirl chamber wall, as shown in Figure 6.5a. Studies by Lefebvre et al. [49]

and Radclie [46] have primarily focused on this spill return type in research purposes; reasons

being that it is the simplest implementation of spilling the fuel from a mechanical design aspect,

and also because this design always has the largest spill diameter of all the spill geometries.

This is because the size of the spill passage has to be large enough to not only spill the fuel, but

also encompass the aircore that is present inside the nozzle [7]. One potential issue with this

design is that air from the aircore, could travel down the spill line. While previous studies by

Lefebvre et al. have shown this to not aect atomization performance, it could be an issue for

mechanical equipment such as pumps and the fuel line manifold [8]. Furthermore, since the air

inside the nozzle is essentially air from the combustor, there might be cases where soot or larger

particles may clog up the spill line system. The eect of spill percentage on the size of the spill

passage is shown in Figure 6.6.

The annular spill geometry is second type of spill geometry, and is shown in Figure 6.5b.

This geometry has the distinction of being the only spill geometry that has been developed for

use in aerospace gas turbines in the past. This spill geometry was used by Rolls-Royce, for the

Pegasus V gas turbine prototype. Although the spill return nozzle was not implemented in the

final design due to issues with a valve that controlled the amount of fuel spilled, it otherwise

performed well from an atomization point of view [10]. The main feature here is that we have

an annular spill line, that is located outside the maximum air core diameter, while also being

a reasonable distance below the inlet ports. This is done to ensure that the liquid coming in

through the inlet ports is not removed, before having a chance to contribute to developing and

maintaining the swirl velocity inside the swirl chamber. Unlike the central spill design concept,

the size of annular spill line is much smaller, due to the fact that we no longer have to account

for the air core diameter in sizing the spill line. The issue however with this design is that for

most spill percentages, it is dicult to make the normalized spill passage width larger than 45%,
Chapter 6. Spill Return Nozzle Design For Model Combustor 67

2 Central(Spill(Line

1 Annular(Spill(Line
Air(Gap

Inlet(Fuel

Spill(Fuel

Retracted(
Orifice 3 Off/Central(Spill(Line
Ceramic(
component

Figure 6.5: Spill return nozzle with: (a) Annular spill line, (b) Central spill line, (c) O-central spill
line

in order to prevent coking; we see this in Figure 6.6 where for spill percentages < 100%, the

width of the annular spill line is < 45%. This is because our nozzle is implemented in a small

engine and thus deals with very small fuel flow rates. This issue did not exist for the Pegasus

V gas turbine, since it was a much larger engine, producing almost 5 more thrust than our

model combustor.

The third type of spill geometry that is available, is the o-central type. The goal with

this design is to avoid the aircore, by placing the spill line o the central axis of the nozzle.

Since we are not using an annular design, we have a much larger passage size, however it is

smaller than the central spill line since we do not need to account for the aircore size. While

the normalized passage diameter is slightly less than the 45% target for the 30% spill scenario,

we avoid the issue of pulling in combustion air, and aecting the pump. We did not attempt to

reach a passage size of 0.45 mm with this design, because a spill percentage outside the limits

was needed to satisfy this criteria.

There are two issues with the o-central design that could potentially hinder its use in

fuel nozzles. The first involves construction of the nozzle, in that it is more complex to build.

Secondly, there is no experimental data on the o-central spill return nozzle to show that it
Chapter 6. Spill Return Nozzle Design For Model Combustor 68

100
Central Spill

N ormal i zed S p i l l Passage S i ze (% of max)


90 OffCentral Spill
Annular Spill

80

70

60

50

40

30

20

10

0
10 20 30 40 50 60 70 80 90 100
S p i l l Percen tage (%)

Figure 6.6: Normalized size of spill passage (%) with spill percentage, for dierent spill geometries

works well. A reason behind this could be the fact that spilling fuel in a non-symmetrical

fashion, could lead to an uneven fuel distribution at the outlet. However, given the large degree

of divergence, and also the high angular velocity inside the swirl chamber, it is expected that

taking fuel from an o-central axis, will not eect the uniformity of the spray. The only method

however to validate this aspect, would be through experimental tests.

From Figure 6.6, we see that the designs with the largest spill passages sizes, were the central

and o-central spill return designs, making the former the most likely design solution. The three

competing factors in terms of the design chosen, was based upon the coking potential of the spill

line, ease of manufacturability, and the eect on spray uniformity. While we can answer the

first two questions analytically, the last aspect can only be determined through experimental

testing. The simplicity in the design and implementation, along with its relatively well defined

history, makes the central spill design the most attractive option, followed by the o-central and

the annular spill line for further study. The corresponding passage sizes for our nozzle design

for a 30% spill percentage, were 24% larger, 15.5% smaller and 56% smaller than the minimum

passage size found by Mitchel et al. respectively. Despite not being a vital, one aspect that

would further benefit the central spill design, would be a method to arrest any air coming down

the spill line.


Chapter 6. Spill Return Nozzle Design For Model Combustor 69

Table 6.2: Normalized spill nozzle design feature values against design practices from experiments [65]
[26] [32]. Requirement indicates whether the parameter need to be below a limit, and Design value
indicated by how much.

Parameter Requirement Design Value


Ap
Ds D0 < 78.5%
Ls
Ds < 32%
L0
D0 < 10%
Lp
Dp > +15.5%
Ds
D0 < 38%
t < 45%
c < 30%
R
D0 < 43%

It should be mentioned that all the spill passage sizes shown in Figure 6.6, were the minimum

size needed for the corresponding spill percentage. While one can increase the spill passage size

beyond what was mentioned, there is a cost that is be paid in terms of being able to control the

spill line, when this is done. This aspect was addressed in Chapter 5.3.

6.2.4 Nozzle Design Check

While a number of nozzle features were sized according to equations provided by Bayvel et

al. [6], the remaining features were obtained by following best practices that have been laid out

by the combination of studies by Xue et al. [65], Kim et al. [26] and Lefebvre [32]. The rational

for the design ratios, and their respective ranges, are described Chapter 5.2.6.

Table 6.2 shows how the design matches up against the requirements from the previous

studies. We see that the design of the nozzle matches favourably for all the parameters. Nozzle

features that were solved using equations as by Bayvel [6] were D0 , Dp Ds , Ls , R and Dsp . The

only features that were solved by using the design convention ratios were L0 and Lp .
Chapter 6. Spill Return Nozzle Design For Model Combustor 70

6.3 Nozzle Performance

In addition to design characteristics, another important aspect is the performance of the nozzle.

Two such important nozzle performance characteristics are the SMD, and spray cone angle () of

the nozzle. Both of these parameters are strongly aected by the spill percentage. It is vital that

once designed, the performance of the nozzle meets adequate performance criteria. Otherwise,

the nozzle features would need to be reworked to ensure that they are met. The performance

criteria include a reduction in SMD, formation of low SMD at low power conditions, and an

that was always < 90 .

6.3.1 Spill Percentage on SMD

Figure 6.7 shows the eect of spill percentage on SMD for a nozzle with a normalized FN of 43%,

for low power conditions. From Figure 6.7, we see that for low spill percentages the reduction

in SMD is relatively small. For a spill percentage of 30%, and at the 7% power condition, the

SMD is 6.5% smaller than if the spill line was closed. Larger drops in the value of SMD are only

obtained for spill percentages in the 60% plus range. However, as was mentioned in Chapter

6.2.1, these large spill percentages are not viable options for this particular design study. One

aspect that we also found was that the starting SMD for the 7% power condition was 45% lower,

than that found in previous studies at this power condition [38].

6.3.2 Spill Percentage on Spray Cone Angle

The eect of spill percentage on is shown in Figure 6.8. Also shown, is the SMD for the

corresponding spill percentage. The blue line on the plot corresponds to the blue axis on the

left, and similarly for the green line on the plot. From Figure 6.8 we see that changes in a

relatively significant manner with the spill percentage. This is one of the biggest issues with

spill return nozzles, since too large an can cause the spray to hit the combustor walls. For our

particular test case, the combustor geometry required that normalized < 80% for all power

conditions. This eectively set our spill percentage at 30%, since the normalized spray cone

angle increases from 71% (0% spill), to 79.5% (30% spill). This was another major factor a spill
Chapter 6. Spill Return Nozzle Design For Model Combustor 71

100
7% Power
10% Power
90 12% Power

N ormal i zed S MD (%) of max


15% Power
80

70

60

50

40

30

20

10
0 20 40 60 80 100 120 140
S p i l l Percen tage (%)

Figure 6.7: Eect of spill percentage on normalized droplet size (%), for various power conditions
(7% 15%)

percentage of 30% was chosen for our design.

6.4 Spilled Fuel Control Mechanism

In-addition to the design of nozzle features, another major component is the establishment of

how to control the spill return nozzle. As mentioned in Chapter 4.2, the spill return nozzle has

the biggest impact at starting and low power conditions. As such, the spill return nozzle must

be designed to take full advantage of its spill capabilities at these conditions. Figure 6.9 shows

a possible method to control the nozzle at starting and low power conditions (0% to 25%). The

method of control is listed as a possible method because the way in which the spill percentage

reduces from 30% to 0%, depends on the engineer implementing the control system. The control

mechanism in Figure 6.9 is implemented such that the spill percentage gradually reduces, while

still taking full advantage of the spill capabilities at starting and ground idle conditions. Also

on Figure 6.9, we have the pressure required in the spill line to ensure that we are spilling the

appropriate percentage of fuel. This pressure was calculated using the procedure mentioned in

Chapter 5.3. As mentioned before, the pressure in the spill line is a function of the size of the
Chapter 6. Spill Return Nozzle Design For Model Combustor 72

100 100

SMD

Nor maliz e d Spr ay Angle ( % of max )


95 90

Nor maliz e d SMD ( % of max )


90
80

85
70

80
60

75
50

70

40

65
0 10 20 30 40 50 60 70 80 90 100
Sp ill Pe r c e nt age ( %)

Figure 6.8: Eect of spill percentage on normalized droplet size(%) and normalized spray cone angle(%),
for the 7% power condition

spill passage and the percentage of fuel that is being spilled.

One thing we notice is that the required pressure in the spill line is a small fraction of

the nozzle pressure drop for high spill percentages, and vice versa for low spill percentages.

Furthermore, when the spill line is fully closed, the spill line pressure is not the same as the

pressure drop across the nozzle. This indicates that there is a pressure loss associated with

pulling the fuel into the spill line, in the case for our spill line, it is approximately 11% of the

nozzle pressure drop. In this design we are using the minimum required passage size for the

spill line for optimal control purposes.

Another important aspect with the nozzle control, is what happens when one looses control

over the nozzle. It is vital that in these cases, the nozzle still performs adequately, and combus-

tion is not severely impacted. For the case where the pressure in the spill line goes cannot be

controlled, in the worse case the nozzle performs like it has a normalized F N = 33% or 43%.

For the former condition, performance of the nozzle should improve as shown previously. This

could however become an issue at high power conditions, where the pump maybe not be able

to provide a 30% increase in fuel flow; creating issues such as a drop in power, or reduced flame

stability. For the latter case the nozzle would perform like a nozzle with a normalized FN of
Chapter 6. Spill Return Nozzle Design For Model Combustor 73

30% Reducing Spill Percentage 0%

100 89%

N ormal i zed N ozzl e P ressu re Drop (% of max)


MAtomized Primary

90
Spill Closes
80 77%
72%
70

60

50

40 62%

30
51% Normalized Spill
20 Pressure

10

Spill Opens
0
20 30 40 50 60 70 80 90 100

N ormal i zed Atomi zed Fu el Fl ow R ate (% of max)

Figure 6.9: Normalized control of spill return nozzle for low power conditions. Normalized spill pressure
Pspill
is Pinjection 100

43%, and again, while there might the emissions levels might be aected, there should be no

reason to expect the combustor to flame out. Finally, one aspect is that on shutdown, the spill

line needs to be drained like the other fuel manifolds. Else, there exists the possibility that

fuel in the spill line could enter the swirl chamber when the engine is next started, due to the

pressure gradient in the nozzle.

Figure 6.10 shows how the two circuits in the nozzle will work with each other. In this

particular case, we have a 30% spill percentage at startup, going to 0% spill percentage at the

25% power condition. Due to constraints by the fuel pump, after the 25% power condition, the

amount of fuel that the primary circuit atomizes remains constant. The secondary circuit how-

ever does not have this constraint, and increases the amount of fuel it atomizes with increasing

power conditions. For this particular setup, we have the spill percentage going to 0% at the 25%

power condition. However, there might be cases where one might want the spill line to always

be running. This is done to ensure that there is no time lag on requests for quick increases in

power.
Chapter 6. Spill Return Nozzle Design For Model Combustor 74

30% Reducing Spill Percentage 0%

100

Normal i zed Fu el Mass Fl ow R ate (% of max)


Primary Fuel Circuit
90 Secondary Fuel Circuit
Fuel Spill Circuit

80

70

60

50

40

30

20

10

0
4 6 8 10 12 14 16 18 20 22 24 26
P ower (%)

Figure 6.10: Normalized control of both circuits from starting and idle power conditions

6.4.1 Spill Return Nozzle Supply Setup

Another factor aecting the design of the spill return nozzle, is with how the spilled fuel was

used. There are two methods in which to handle the spilled fuel. The first method involves

sending the spilled fuel from the nozzle to a separate tank and reusing it -as shown in Figure

6.11. With this setup, there would need to be another fuel manifold to handle the spilled fuel

from the nozzle, in addition to the two manifolds for the primary and secondary circuit. The

spill valve as indicated in Figure 6.11, is there to ensure that we are spilling the correct amount

of fuel, for a given power condition. To reduce complexity and weight, we have only added a

single valve to control the amount of fuel being spilled. Depending on other requirements such

as reliability and safety, one might need to add another valve as a backup, to the one being

used. The spill valve as currently planned, is expected to be controlled by the fuel control unit

(FCU). Since most modern FCUs are electronic, one should be able to control the spill valve

without any additional weight.

The fuel that is spilled to the tank, is subsequently connected to the fuel metering unit

(FMU). The FMU of the gas turbine is controlled by the FCU, and determines the fuel quantity

that must be sent to the flow divider -which splits the fuel into primary and secondary fuel.
Chapter 6. Spill Return Nozzle Design For Model Combustor 75

Fuel Nozzle

Fuel
Control
Unit

Spill
manifold

Primary &
P3 Secondary
Auto Shutoff manifold
Valve
Flow
Divider

Spill Spill Valve


Tank

Figure 6.11: Implementation of the back-end fuel supply system for the spill return nozzle

The only issue with this setup is that in adding the spilled fuel to the main fuel line going to

the flow divider, it would change back end control, since the fuel quantities and pressures from

the tank would need to be recalibrated. One advantage though with this method is that when

the fuel is in the spill manifold, it should allow it to cool down, since the manifold is generally

exposed to cooler air from the surrounding atmosphere.

The next method as shown in Figure 6.12, involves spilling the fuel from the primary circuit

to the secondary circuit, at the nozzle itself. In this particular case, a Reed valve is used at each

nozzle to help ensure that the correct amount of fuel is being sent to the secondary circuit. This

valve was chosen because earlier studies by Stickles et al., had showed the Reed valve to work

reliably for a similar purpose, over a large number of tests in an experimental nozzle.

The biggest issue with choosing the second method has to do with the quality of the spilled

fuel once it leaves the nozzle. This is because the fuel temperature in the swirl chamber for similar

sized combustors gets hot enough, that coking could be an issue. At this fuel temperature, there

is the possibility that fuel may form coke deposits, which can not only clog up spill passages, but

also passages in the fuel supply system. This issue is not expected to be there for the previous

method design as the fuel should be able to cool down when in the spill manifold. For this
Chapter 6. Spill Return Nozzle Design For Model Combustor 76

Airblast circuit
Spill%Return%
Reed
Nozzle
Valve Secondary%
Fuel%
Entry
Secondary%
fuel%from%
manifold
Spill%Fuel

Fuel%Reservoir
Primary%Fuel%
Entry
Spill%Fuel%Line

Figure 6.12: Implementation of spill return nozzle with spilled fuel sent to secondary circuit

reason, the first method was assumed to be the best solution, until further study showing the

quality of the spilled fuel was available. The second method while holding a lot of promise, is

quite complex in that it requires the use of a valve for each nozzle, that all need to be controlled.

6.5 Proposed Nozzle Design Concept

All the above features were incorporated into a design that could be implemented in the model

combustor. This combustor uses a hybrid nozzle, where the primary circuit is a pressure nozzle,

while the secondary circuit is an airblast nozzle. Three dierent designs were proposed, where

each design implemented a dierent method of spilling the fuel. These designs are shown in

Figure 6.13 to Figure 6.15. It should be mentioned that in all three proposed designs, only the

primary circuit of the nozzle has changed, as shown in the black-boxed section in Figure 6.13.

The proposed fuel nozzle designs have been colour coded to indicate the appropriate flow of fuel

in the nozzle.

The orange parts in Figure 6.13 indicate places where the secondary fuel flows, while green

represents spilled fuel, and red represents primary fuel; the same convention is followed for the

other two nozzles. Size wise, the outer diameter of the nozzle is 5.77 the outer diameter of

the primary circuit. The relative sizes of the two circuits provide an indication of the amount
Chapter 6. Spill Return Nozzle Design For Model Combustor 77

Tangential
Air Tubes
High
Temperature
Material

Central Spill Passage

Straight Flow
Primary Passages
Circuit
Retracted
Front Tip

Return
Fuel

Primary Fuel

Secondary Fuel

Airflow

Figure 6.13: Hybrid nozzle with central spill line

of fuel that each circuit handles. The nozzles as shown have an inner and outer airflow passage,

where the inner air passage flows almost half the amount of air as the secondary circuit. The

secondary air is used with the airblast circuit, to help with atomization and mixing of the fuel

and air.

The outer swirler in the design is used because it plays a vital role in atomization, spray

distribution, and flame stability. It is expected that the nozzle will operate with an AFR in the

range of 2 3, which can be as much as 6.5% of the total compressor airflow. Since we are forcing

such a relatively large amount of air through a small nozzle, the momentum concentration of

the injector flow relative to the remainder of the air entering the combustor primary zone is

much higher. The nozzle uses this to create a central recirculation zone along the axis of the

injector flow field, and also influence flame stability, ignition and emissions characteristics. The

nozzle outer swirler consists of radially drilled angled tubes that is used to force air over the

airblast circuit and help improve atomization. Angled holes are used as opposed to vanes, since

it is a much simpler design, and there are also design constraints due to the small overall size

of the nozzle [25].


Chapter 6. Spill Return Nozzle Design For Model Combustor 78

Off-Central
Spill Passage Primary Fuel

Flat Front
Nozzle Face
High
Temperature
Material

Figure 6.14: Hybrid nozzle with o-central spill line

All three proposed fuel nozzle designs also have a central air passage, which separates the

primary circuit from the secondary circuit. The passage width between the primary circuit

outer body, and the secondary circuit inner body was maintained from the initial design. This

was done to prevent any changes to the pressure drop across the primary circuit, consequently

aecting the amount of air flowing between the two circuits. The air around each primary

circuit is swirled, as it helps in developing a stable recirculation zone that is generated by the

injector. The central air is also used to cool the nozzle tip, and ensure that any coke deposition

on the nozzle surface is washed o. The reason for swirling the core air is because studies in

the past have shown that unswirled core air flows have resulted in improper mixing of fuel and

air, leading to excessive smoke emissions, and high wall temperature gradients [25]. Another

potential benefit with air around the pressure nozzle, is that it plays a role in neutralizing eects

of streaky sprays and drooling [25, 32]. Like the swirler around the secondary circuit, the air

around the primary circuit is supplied through six air tubes, with feed multiple tangential jets

around the primary circuit.

To reduce coking potential, the front face of the primary circuit has been placed further

back than the front face of the secondary circuit. This is done to reduce the radiation exposure

of the primary circuit from the combustion process. For all the primary circuits, we have have
Chapter 6. Spill Return Nozzle Design For Model Combustor 79

Annular Spill
Passage Primary Fuel

Flat Front
Lip and
High Retracted Tip
Temperature
Material

Figure 6.15: Hybrid nozzle with annular spill line

a conical section in front of the orifice. The conical section is there to not only force a certain

spray cone angle, but to also further pull back the orifice. One potential issue with this design

is that fuel could potentially coke in the conical section during soakback conditions.

The internal features of the primary circuit, namely the swirl chamber and number of inlet

passages are shown in Table 1.1. The only change between the designs, is the implementation

of the spill line at the back wall of the swirl chamber. All the designs proposed were for a nozzle

with a maximum spill percentage of 30%. In Figure 6.13, we have the implementation of the

central spill line. This return fuel passage was given high priority in terms of location, because

it carries hot fuel from the front. Eort was taken to ensure that we have no steps or bends in

this fuel flow path. The same was the case for the o-central and annular spill lines.

This was done because sudden flow direction changes can cause stagnation zones. At these

stagnation zones, the fuel temperature can get quite high, and cause coking to occur [58].

Furthermore, since front of the nozzle gets hot due to exposure to heat, the size of the fuel

passages are reduced to increase the fuel velocity as it approaches the front. The size of the fuel

passages are balanced between achieving the appropriate increase in fuel velocity and getting

too small, where coking is an issue. The goal here is to ensure that the fuel spends a small

period of time in these hot zones, reducing the likelihood of coking occurring. Eort was also
Chapter 6. Spill Return Nozzle Design For Model Combustor 80

taken to ensure that fuel input lines are straight, with no sudden area changes. Unfortunately,

this was not fully possible for the primary circuit flow as other design constraints required an

initial L-shaped passage. However, given that fuel is much cooler in the rear and the much wider

passages, coking may not be a significant issue. Furthermore, at places where the passage sizes

are reduced, a higher surface finish could be used to reduce the possibility of coke adhering to

the surface.

Finally, the nozzle was designed to be built in multiple parts to help maintenance, and also

to help use more expensive materials in the front, while cheaper materials that just need to be

used for handling loads can be used in the rear. The front part which is shown in a darker shade

of grey in Figure 6.13 needs to be able to handle heat, have a low coking potential, and ability

to handle wear and tear. For our particular case, this can be achieved by having materials like

SS446, SS316 or SS216. If need be, a thermal barrier coating like Partially Stabilized Zirconia

(PSZ can be used to help with reducing thermal gradients. Generally, a TBC that is 25.4 m

thick can help reduce the surface wall temperature by as much as 4 K 10 K. For our case,

we need to ensure that we design the nozzle to have maximum insulative capabilities at takeo

and soakback conditions. Thus a TBC thickness of about 50 m to 75 m would be advisable.

In summary, by replacing the old primary circuit of the model combustor, with the new spill

return nozzle we should expect to see the following changes:

Improved atomization starting at low flow rates. A 30% spill percentage should ensure

at least a 7% improvement in SMD. Further increases in spill percentage will allow for a

greater reduction in SMD.

Reduced nozzle internal coking potential due to internal passage sizes that are as much as

15% larger.

Increased nozzle cooling due to the fact that hot fuel is being spilled from the front of the

nozzle. This aects both internal coking and nozzle durability, since the nozzle does not

reach the high temperatures it used to.

A simple design that is simple and robust, with no moving parts. Since the only change
Chapter 6. Spill Return Nozzle Design For Model Combustor 81

from a simplex nozzle is the addition of a spill passage at the rear wall, previous experience

with pressure atomizers to solve certain issues that arise in combustor design, are still valid.

Ability to obtain good atomization rates with dierent fuel types, or with fuels of lower

temperatures. This is an advantage for cold starting conditions.

There is an increase in spray cone angle with spill percentage. This could be advantageous

at starting conditions, if we decide to use fuel staging. If need be, the spray cone angle

can be reduced by using central air.

Increase in fuel pumping requirements, and the need for hardware to handle the spilled

fuel.
Chapter 7

Next Steps and Future Work

The next steps for this project are the validation of the designs that were studied in the previous

chapters. This is planned to occur in two phases, where the first phase would be to perform

experimental studies on a simplified spill return nozzle, to validate its performance and control

aspects. The goal here would be to only study the spill return nozzle in a simplified design

setup, while ignoring other aspects such as features to reduce coking potential, or the secondary

airblast circuit. The nozzle designs that were chosen for further study were the central, annular

and o-central design variants. The design of the simplified nozzle components were developed

in this thesis; work for experimental testing is to be conducted in the future as a separate

project.

7.1 CAD Drawing of the Spill Nozzle

Figure 7.1 shows the 3D CAD Model of the simplified nozzle designed to be used for experimental

testing. The nozzle was designed such that the dierent spill geometries could be tested by only

changing one component. Standard O-Rings were placed a two locations inside the nozzle to

ensure leakages would not occur. Finally, the input and output fuel pipes were connected to the

nozzle through a special National Pipe Taper Fuel (NPTF) thread of standard size to ensure

leak free connections.

In terms of fit, components two and three are supposed to be high precision components.

82
Chapter 7. Next Steps and Future Work 83

Fuel Inlet
(NPTF Thread)

O-Ring
Two Inlet
Passages

O-Ring
Fuel Spill Outlet
(NPTF Thread)

4 Screw-in Type
Cover to
Reduce
Central Spill Internal Gaps
Line
1
3

Nozzle Orifice &


Chamber Component

2
Nozzle Sealed
Using
O-Ring

Figure 7.1: 3D Model of the Fuel Nozzle Showing the Various Parts

Consequently, the orifice and the inlet passages have very small tolerances, since small deviations

could have an impact on the expected nozzle performance. Parts one and four have larger toler-

ances since they hold everything together and do not have any internal nozzle features. While

all parts are expected to be held together by screwing component 4 and 1 tightly, component 1

and 2 will be brazed with each other to ensure proper fit, and prevent any leakages. The CAD

drawings with the dimensions of the various features of the components are shown in Figures

A.1 to A.4 in the Appendix. Similar detailed dimensions were also done for the annular and

o-central design geometries.

Important points with respect to the design and manufacture of s based atomizers are that

the tangential entry ports must be located symmetrically around the swirl chamber, to ensure

uniformity in the resulting spray. Furthermore, it is also vital that the orifice is in line with the

axis of the swirl chamber. These demands put a very high accuracy limit when it comes to the

manufacture of pressure atomizers, as a lack o, would result in performance deterioration. For

instance, for Dp , if the diameter is o by 0.0254mm, it results in a 7% change in the area of the

inlet, and an even more significant change for the amount of fuel entering the swirl chamber.
Spray angles +10/-0 from nominal rating.@300 psig Example: 200.0
operating pressure, with spill-back valve closed.
Test fluid: heavy oil.

Chapter 7. Next Steps and Future Work 84


Spill-Back Nozzle Assembly Instal

Installation:
A pressure regulati
Back pressure. Wit
nozzle discharges a
rate is adjusted by c
(a) (b)

Figure 7.2: (a) 3D printed nozzle parts, (b) Schematic of the ordered commercial spill return nozzle [18]

Hago Manufacturing Co., Inc.


1120 Globe Avenue Mountainside, NJ USA 07092
Tel 908.232.8687 Fax 908.232.7246 Sales 1.800.71
7.1.1 3D Printing of Spill Nozzle E-Mail: sales@hagonozzles.com www.hagonozzles.c

Figure 7.2(a) shows a 1 : 1 3D printed scaled model of the central spill nozzle. The 3D printer

had a resolution of 10 m, which was smaller than the design tolerances of 12.7 m. The purpose

of 3D printing the spill return nozzle was to see if there were any issues in fitting or design that

may have been missed during the design process. The 3D printed nozzles could not be used

for experimental testing for two reasons. The first was because the surface finish of the nozzles

was very rough, and experimental testing would have resulted in larger than normal internal

pressure losses. Secondly, since the nozzle was printed out of plastic, it is not strong enough

to withstand fuel travelling at high velocities in the internal fuel passages. In order to prevent

these issues, the experimental testing prototype needs to be built in metal.

7.2 Planned Experimental Testing of Spill Nozzle Design

The purpose of the planned experimental testing is to validate the following aspects:
Droplet Size
Spray Cone Angle
Spray Uniformity
Hysteresis
The experimental testing is expected to validate the analytical calculations for the droplet size

and spray cone angle, as shown in Chapter 4.2. With droplet size and spray cone angle, the

goal would be to find out how they change with injection pressure, for a given number of spill
Chapter 7. Next Steps and Future Work 85

percentages. Since the nozzles were designed for a maximum of 30% spill, the testing should

be able to cover a range from 0% to 30% from a theoretical point of view. Spray uniformity

was not studied in Chapter 4.2, since the central and annular spill return nozzles remove the

fuel from the nozzle in a symmetric manner, they should produce sprays which are uniform.

With respect to the o-central nozzle, the lack of any studies on this particular concept makes

it dicult to predict how the nozzle will perform. However, if one extends the study on spray

uniformity in Chapter 5.2.5, given the large atomizer divergence that was selected for this nozzle,

the o-central spill line is not expected to aect spray uniformity. However, if the o-central

spill location does aect spray uniformity, one way to fix it would be by adding another spill

passage directly opposite it to ensure symmetric spilling of the fuel. This would however, half

the size of the eective spill passage, making them more prone to coking. With hysteresis, the

focus is to see if the nozzle performs the same way when powering-up, versus powering-down.

Tests in certain rigs have shown hysteresis to be an issue, during power-up and power-down

situations [19, 58].

Despite not being important for preliminary testing, the next item of consideration would

be to see how the primary and secondary circuit interact together. It is vital that each nozzle

does not interact with each other in a negative manner, since trading performance at high power

conditions, for improved performance at low power conditions would not be an acceptable trade-

o. To validate the performance of our spill return nozzle design, a commercially designed spill

return nozzle was ordered. As shown in Figure 7.2(b), this nozzle use a central spill line to spill

the fuel. While the nozzle has the same FN when the spill line is fully closed, the FN when the

spill line is fully open, is o by 11%. Nevertheless, testing the designs created in this thesis,

allow us to study the eect of the spill geometry on the nozzle performance. The comparison

between our central spill line and the commercially designed spill line, stand to inform us of

how much better (or worse) our design is, compared to a commercially optimal designs.
Chapter 8

Conclusions

The objective of this thesis was to study dierent fuel nozzle technologies for small gas turbines,

that could particularly reduce emissions at low power conditions. The primary reason for this

focus, is because combustors generally have combustion eciencies close to 99% at high power

conditions, often independent of the nozzle being used. This however is not the case at starting

and low power conditions, where combustion eciency is much lower [44]. The selection of

the spill return nozzle to improve emissions at low power conditions, was based upon a study

by Norgren et al.. They found that a novel fuel nozzle technology for starting and low power

conditions, was the spill return nozzle. The main advantage with this design is that, it is a

variable FN nozzle. This enables one to design the nozzle with a large FN, and then reduce the

FN of the nozzle through spilling fuel.

The next step involved developing a new design for the spill return nozzle for a specific

combustor that was selected by our industrial sponsor. This combustor was a small reverse

flow combustor, with a hybrid airblast nozzle. Given that the spill return nozzle was shown to

work best at starting and ground idle conditions, only the primary circuit of the hybrid airblast

circuit was changed. It is hoped that by replacing the old primary circuit with the new spill

return nozzle, we should be able to get better SMD at low power conditions, improved mixing,

and lower coking potential.

Before designing the new primary circuit, the first aspect that was worked on, was a method-

ology that could be used to design, and determine the spill return nozzles performance. This

86
Chapter 8. Conclusions 87

aspect was covered in Chapter 5, where the methodology of Bayvel et al. for simplex nozzles

was extended to spill return nozzles. This methodology was subsequently used to design the

internal features of the nozzle, as shown in Chapter 6. Some improvements include, a 15%

increase in the orifice size, and a 13% increase in the swirl chamber inlet ports. This is impor-

tant because previous studies have shown these two parts of the nozzle to have a large impact

on performance, while also being the two parts that are the most prone to coking [41]. The

methodology of Bayvel et al. was also extended to help calculate SMD, which for the spill return

nozzle was shown to reduce by 6.5% for a 30% spill percentage. The spill percentage is the most

important parameter in designing this nozzle, because it has a hand in determining atomization

performance, and internal passage sizes. As mentioned previously, the spill percentage chosen

was a maximum of 30%, due to a number of factors such as pumping power requirements, spray

distribution, and spray uniformity. One aspect that was also covered, was a validation of the

extended methodologies. When the methodology for sizing nozzle features was validated against

experiential data for simplex nozzles, the error was less than 3%. Similarly, for calculation of

the SMD, the error was less than 12%.

A method to control the nozzle was also used, and was based upon a methodology developed

by Bowen. The nozzle control mechanism designed, involved spilling a maximum spill percentage

of 30% at starting, going down to 0% at the 25% power condition. The implementation of the

spill return nozzle in the hybrid airblast nozzle is shown in Chapter 6.5, in the form of three

dierent geometries. In addition to implementing the spill return nozzle, we have also added

various coking reduction solutions into the design.

The last part of this project involved developing a simplified version of the three dierent

nozzles geometries that were proposed for experimental testing purposes. The main goal of this

component, is to find out the eect of the three dierent spill methodologies on SMD, spray

uniformity and internal pressure loss. To do this, a new nozzle geometry was developed to

only test the spill return aspects of the proposed nozzle geometries. CAD drawings of this new

geometry are shown in the Appendix. Finally, these nozzle components were also build using

a 3D printer to validate their design. The 3D printed parts could not be tested, since they

were built out of plastic. Plastic printed nozzles cannot be used for tests because the internal
Chapter 8. Conclusions 88

passages erode due to the high fuel velocity inside. Future work involves building the nozzles in

metal, and then testing it to validate our results.


Bibliography

[1] Turboprops vs. turbofans, January 2014.

[2] Andrsen, J. M., Strohm, J. J., Sun, L., and Song, C. Relationship between the formation
of aromatic compounds and solid deposition during thermal degradation of jet fuels in the pyrolytic
regime. Energy & Fuels 15, 3 (2001), 714723.

[3] A.S. Grossman, J.M. Haywood, D. R., and Subbaraya, B. IPCC Special Reports on Climate
Change. Michael Prather and Rober Sausen, November 2000.

[4] Ballester, J., and Dopazo, C. Discharge coecient and spray angle measurements for small
pressure-swirl nozzles. Atomization and Sprays 4 (1994), 351367.

[5] Balster, W. J., and Jones, E. G. Eects of temperature on formation of insolubles in aviation
fuels. Journal of Engineering for Gas Turbines and Power 120, 2 (04 1998), 289293.

[6] Bayvel, L. Liquid Atomization. Combustion (Hemisphere Publishing Corporation). Taylor &
Francis, 1993.

[7] Bowen, I. The spill pressure jet atomizer. Journal of the Institute of Fuel 35 (November 1962),
484495.

[8] Carey, F. The development of the spill flow burner and its control system for gas turbine engines.
Journal of Royal Aeronautical Society 58, 527 (November 1954), 737753.

[9] Department, T. World jet fuel specifications. Tech. rep., ExxonMobil Aviation International Ltd.,
Leatherhead, UK, 2005.

[10] Dow, A. Pegasus, The Heart of the Harrier: The History and Development of the Worlds First
Operational Vertical Take-o and Landing Jet Engine. Pen & Sword Aviation, 2009.

[11] urdina, L. Measurement of spray characteristics using optical measurement methods. Masters
thesis, Brno University Of Technology, Antonnsk 548/1, 601 90 Brno, Czech Republic, 2012.

89
Bibliography 90

[12] Eatock, H., and Sampath, P. Low emissions combustor technology for small aircraft gas turbines.
Tech. rep., A Division of United Technologies Corporation.

[13] Elkotb, M., Madhy, M., and Montaser, M. Investigation of external-mixing airblast at-
omizers. In Proceedings of the Second International Conference on Liquid Atomziation and Spray
Systems. (1982), ILASS, pp. 105 115.

[14] Farvardin, E., and Dolatabadi, A. Numerical simulation of the breakup of elliptical liquid jet
in still air. Journal of Fluids Engineering 135, 7 (05 2013), 071302071302.

[15] Feras Z. Batarseh, I. V. R., and Tropea, C. Characterization of a spray generated by an


airblast atomizer with a prefilmer. Atomization and Sprays 21, 10 (2010), 887903.

[16] Fraser, R. P., Eisenklam, P., Dombrowski, N., and Hasson, D. Drop formation from
rapidly moving liquid sheets. AIChE Journal 8, 5 (1962), 672680.

[17] Guildenbecher, D. R., Lpez-Rivera, C., and Sojka, P. E. Secondary atomization. Exper-
iments in Fluids 46, 3 (2009), 371402.

[18] Hago Manufacturing Co., I. Technical data sheet (return flow nozzles). Online, 2010.

[19] Hazlett, R. N. Thermal Oxidation Stability of Aviation Turbine Fuels. American Society for
Testing and Materials, 1916 Race St. Philadelphia, PA 19103, 1991.

[20] Hinze, J. O. Fundamentals of the hydrodynamic mechanism of splitting in dispersion processes.


AIChE Journal 1, 3 (1955), 289295.

[21] Huang, H., Spadaccini, L. J., and Sobel, D. R. Fuel-cooled thermal management for advanced
aeroengines. Journal of Engineering for Gas Turbines and Power 126, 2 (06 2004), 284293.

[22] ICAO. International civil aviaition organization annex 16: Environmental protection volume ii
aircraft engine emissions, ammendment 7. Tech. rep., U.S. Government, November 2011.

[23] Inamura, T., and Nagai, N. The relative performance of externally and internally mixed twin-
fluid atomisers. In Proceedings of the 3rd International Conference on Liquid Atomization and Spray
Systems (Imperial College, July 1985).

[24] Janicka, J., Sadiki, A., Schfer, M., and Heeger, C., Eds. Flow and Combustion in Ad-
vanced Gas Turbine Combustors. No. 102 in Fluid Mechanics and Its Applications. Springer Nether-
lands, 2013.
Bibliography 91

[25] K. McCaldon, L. P., and Sampath, P. Design aspects in small aircraft gas turbine fuel injectors.
In Fuels and Combustion Technology for Advanced Aircraft Engines (Fiuggi, Italy, May 1993),
no. 536 in AGARD Conference Proceedings, AGARD.

[26] Kim, S., Khil, T., Kim, D., and Yoon, Y. Eect of geometric parameters on the liquid film
thickness and air core formation in a swirl injector. Measurement Science and Technology 20, 1
(2009), 015403.

[27] Kreith, F. Fluid Mechanics. Mechanical/Chemical engineering. Taylor & Francis, 1999.

[28] Lai, G. High-Temperature Corrosion and Materials Applications. ASM International, 2007.

[29] Lanewala, H. Multi-dimensional nox emissions predictor for preliminary gas turbine combustor
design optimization. Masters thesis, University of Toronto Institute of Aerospace Studies, 2012.

[30] Lefebvre, A. H. Airblast atomization. Prog. Energy Combust. Sci. 6 (1980), 233261.

[31] Lefebvre, A. H. Fuel eects on gas turbine combustion-linear temperature, pattern factor, and
pollutant emissions. Journal of Aircraft 21, 11 (Nov. 1984).

[32] Lefebvre, A. H. Atomization and Sprays. Hemisphere Publishing Corporation, 1989.

[33] Lefebvre, A. H., and Ballal, D. R. Gas Turbine Combustion: Alternative Fuels and Emissions,
third ed. CRC Press, 2010.

[34] lefebvre, A. H., and Suyari, M. Film thickness measurements in a simplex swirl atomizer.
Journal of Propulsion and Power 2, 6 (2015/07/31 1986), 528533.

[35] Lefebvre, A. H., and Wang, X. F. Mean drop sizes from pressure-swirl nozzles. Journal of
Propulsion and Power 3, 1 (2015/07/23 1987), 1118.

[36] Liu, Z. J. Chemical kinetic and thermal numerical simulation of coking process of jet fuels in thin
nozzle sections within autoxidation temperature regime, 2014.

[37] Loffler-Mang, M., and Leuckel, W. Atomization with spill controlled swirl pressure jet
nozzles. In International Conference on Liquid Atomization and Spray Systems (Gaithersburg,
MD, USA, July 1991), Combustion Technology Section, Univeristy of Karlsruhe, pp. 431 440.

[38] Marchand, M. D. Multi-dimensional carbon monoxide emissions predictor for preliminary gas
turbine combustor design optimization. Masters thesis, University of Toronto Institute of Aerospace
Studies, 2012.

[39] Mellor, A. Design of modern turbine combustors. Combustion treatise. Academic Press, 1990.
Bibliography 92

[40] Mitchell, J. P. Final report for study of nozzles to spray contaminated fuel. Tech. rep., Depart-
ment of Navy Burearu of Naval Weapons, Florida Research and Development Center, 1962.

[41] Mitchell, J. P. Final report for study of nozzles to spray contaminated fuel. Tech. rep., Depart-
ment of Navy Burearu of Naval Weapons, Florida Research and Development Center, 1962.

[42] Myers, G. D., Armstrong, J. P., White, C. D., Clouser, S., and Harvey, R. J. Devel-
opment of an innovative high-temperature gas turbine fuel nozzle. Journal of Engineering for Gas
Turbines and Power 114, 2 (04 1992), 401408.

[43] Nguyen, T. K. Gas turbines for aircraft propulsion chapter 9.9-10.

[44] Norgren, C., and Riddlebaugh, S. Eect of fuel injector type on performance and emissions
of reverse-flow combustor. Tech. rep., National Aeronautics and Space Administration, 1981.

[45] Pilch, M., and Erdman, C. A. Use of breakup time data and velocity history data to predict the
maximum size of stable fragments for acceleration-induced breakup of a liquid drop. International
Journal of Multiphase Flow 13, 6 (1987/12// 1987), 741757.

[46] Radcliffe, A. The performance of a type of swirl atomizer. In Proceedings of the Institution of
Mechanical Engineers (June 1955), vol. 169, pp. 93106.

[47] Reitz, R. Atomization and Other Breakup Regimes of a Liquid Jet. PhD thesis, Princeton Uni-
versity, 1978.

[48] Rizk, N., and Mongia, H. Performance of hybrid airblast atomizers under low power conditions.
American Institute of Aeronautics and Astronautics, 2015/07/21 1992.

[49] Rizk, N. K., and Lefebvre, A. H. Drop-size distribution characteristics of spill-return atomizers.
Journal of Propulsion and Power 1, 1 (2015/07/18 1985), 1622.

[50] Rizk, N. K., and Lefebvre, A. H. Spray characteristics of spill-return atomizers. Journal of
Propulsion and Power 1, 3 (2015/07/31 1985), 200204.

[51] Rizkalla, A. A., and Lefebvre, A. H. Influence of liquid properties on airblast atomizer spray
characteristics. Journal of Engineering for Power 97, 2 (04 1975), 173177.

[52] Rolls-Royce. The Jet Engine, 6th ed. Wiley, Derby, England, 1966.

[53] Rosfjord, T., and Briel, D. Evaluation of fuel injection configurations to control carbon and
soot formation in small GT combustors. American Institute of Aeronautics and Astronautics,
2015/06/23 1982.
Bibliography 93

[54] Sampath, P., Hu, T., and Ozem, H. Combustion technologu challenges for small aviation. Tech.
rep., Pratt and Whitney Canada, 1801 Courtney Park Drive, Mississauga, Ontario, Canada, 2002.

[55] Schilke, P. Advanced gas turbine materials and coatings. Tech. rep., General Electric Energy,
Schenectady, NY, July 2004.

[56] Silva Couto, H., Lacava, P. T., Bastos-Netto, D., and Pimenta, A. P. Experimental eval-
uation of a low pressure-swirl atomizer applied engineering design procedure. Journal of Propulsion
and Power 25, 2 (2015/07/10 2009), 358364.

[57] Spadaccini, L., Tillman, G., Gentry, J., Chen, A., and Edwards, T. Fuel Stabilization
Unit Proof of Concept Engine Demonstration. American Institute of Aeronautics and Astronautics,
2015/08/02 2008.

[58] Stickles, R. W., Dodds, W. J., Koblish, T. R., Sager, J., and Clouser, S. Innovative
high-temperature aircraft engine fuel nozzle design. Journal of Engineering for Gas Turbines and
Power 115, 3 (07 1993), 439446.

[59] Stiesch, G. Modeling Engine Spray and Combustion Processes. Heat and Mass Transfer. Springer
Berlin Heidelberg, 2003.

[60] Sturgess, G. J., Zelina, J., Shouse, D. T., and Roquemore, W. M. Emissions reduction
technologies for military gas turbine engines. Journal of Propulsion and Power 21, 2 (2015/08/08
2005), 193217.

[61] Weber, C. Disintegration of liquid jets. Zeitschrift fr Angewandte Mathematik und Mechanik 11,
2 (1931), pp. 13659.

[62] Westergaard, V. Milk powder technology. Tech. rep., GEA Process Engineering, Gladsaxevej
305 - 2860 Sborg - Denmark, 2000.

[63] William H. Edwards, John A. Harris, E. S. S. Inhibiting coke formation by heat treating in
nitrogen atmosphere. Patent, US 5298091 A, 1994.

[64] Wright-Patterson Air Force Base. Alternative fuels combustion research. Tech. rep., Aero
Propulsion Laboratory, July 1984. AFWAL-TR-84-2057.

[65] Xue, J., Jog, M. A., Jeng, S. M., Steinthorsson, E., and Benjamin, M. A. Eect of
geometric parameters on simplex atomizer performance. AIAA Journal 42, 12 (2015/07/18 2004),
24082415.
Bibliography 94

[66] Zhang, S., and Zhao, D. Aerospace Materials Handbook. Advances in Materials Science and
Engineering. CRC Press, 2012.
Appendix A

Equations and Experimental Nozzle

CAD Drawing

The equations shown below are the final equations that were re-derived in order to model spill return
nozzles, and do not show any intermediate equations. Please see section Fundamental Equations for a
description on the equations.

p
(1 ) 2
K= (A.1)
(3/2 )e
s
1
= (A.2)
1 + (Ke)2

p !
p p 1 + 1 (Ke)2
= 1 (Ke)2 S S2 (Ke)2 (Ke)2 log p (A.3)
S + S 2 (Ke)2
!
2Ke
= 2 arctan p (A.4)
(1 + S)2 4(Ke)2

4m
Re = p (A.5)
iDp

25.8
log( ) = 2 (A.6)
log(Re)2.58

K
K = 2 ! (A.7)
R
rp
1+ 2 i K

A-2
Appendix A. Equations and Experimental Nozzle CAD Drawing A-3

Figure A.1: Dimensions and tolerance for the front nozzle component
Appendix A. Equations and Experimental Nozzle CAD Drawing A-4

Figure A.2: Dimensions and tolerance for the orifice component


Appendix A. Equations and Experimental Nozzle CAD Drawing A-5

Figure A.3: Dimensions and tolerance for the core component


Appendix A. Equations and Experimental Nozzle CAD Drawing A-6

Figure A.4: Dimensions and tolerance for the rear component

You might also like