The Pennsylvania State University

The Graduate School

Department of Mechanical and Nuclear Engineering

SEALING EFFECTIVENESS OF A TURBINE RIM SEAL

AT ENGINE-RELEVANT CONDITIONS

A Dissertation in

Mechanical Engineering

by

Kenneth P. Clark

2016 Kenneth P. Clark

Submitted in Partial Fulfillment

of the Requirements

for the Degree of

Doctor of Philosophy

December 2016

The dissertation of Kenneth P. Clark was reviewed and approved* by the following:

Karen A. Thole

Department Head of Mechanical and Nuclear Engineering

Dissertation Advisor

Chair of Committee

Savas Yavuzkurt

Professor of Mechanical Engineering

Steven P. Lynch

Assistant Professor of Mechanical Engineering

Robert F. Kunz

Senior Scientist and Head of the Computational Mechanics Division, PSU ARL

Professor of Aerospace Engineering

Mary Frecker

Associate Department Head for Graduate Programs

Mechanical and Nuclear Engineering

Professor of Mechanical & Biomedical Engineering

*Signatures are on file in the Graduate School

This document has been publicly released.

iii

ABSTRACT

Gas turbines are characterized by high efficiencies compared to other power generation

systems, but even small efficiency gains can result in huge reductions in fuel costs and emissions.

Overall pressure ratios for gas turbines continue to rise to achieve higher efficiencies, resulting in

increasingly extreme thermal conditions in the hot section of the engine. Secondary air, bled from

the compressor, is used to cool turbine components and seal the inter-stage cavities from the hot

main gas path to maintain component durability. Unsealed cavities lead to hot gas ingestion, which

can degrade critical components or, in extreme cases, can be catastrophic to engines. Providing

adequate sealing flow is required to maintain component life, but efficient use of the secondary air

is required to maintain the thermodynamic efficiency of the engine. To fully optimize these

competing requirements, experiments at engine-relevant conditions are required to validate new

designs and computational tools. This dissertation presents sealing effectiveness measurements for

an engine-realistic turbine rim seal operated at engine-relevant conditions.

A new facility providing continuous flow for a 1.5 stage turbine was designed, constructed,

and commissioned to study hot gas ingestion for at engine-relevant conditions. Sealing

effectiveness was determined through concentration measurements, whereby CO2 was used as a

tracer gas in the secondary air supply, and sampled throughout the turbine rim seal and rim cavity.

The turbine design was representative of a modern turbine with engine-realistic purge flow delivery

and leakage flows, resulting in complex flow fields in the cavities. The measurements indicated

that sealing effectiveness depended on the number of purge holes at and inboard of the purge

injection location, with more purge holes exhibiting higher effectiveness than fewer purge holes.

Outboard of the purge location, the sealing effectiveness was not affected by the number of purge

holes. The boundary layer on the rotating disk entrained the purge flow, and the rotor and stator

pumping distributed the purge flow throughout the cavity reducing the amount of ingestion. The

results indicated that the well-accepted orifice models used to predict sealing effectiveness

throughout the industry are successful for some cases; however, the results in this dissertation

showed that those models are unsuccessful in predicting sealing effectiveness for the complex

engine-realistic geometries and purge flow delivery methods. These findings highlight the need to

obtain sealing effectiveness data at engine-relevant conditions, using engine-realistic hardware to

develop ingestion models applicable to gas turbine engines.

iv

TABLE OF CONTENTS

List of Figures .......................................................................................................................... vi

List of Tables ........................................................................................................................... xi

Nomenclature ........................................................................................................................... xii

Acknowledgements .................................................................................................................. xv

Chapter 1 Introduction ............................................................................................................. 1

1.1 Background and Motivation ....................................................................................... 1 1.2 Objectives and Uniqueness of Research ..................................................................... 7 1.3 Outline of Dissertation ................................................................................................ 8

Chapter 2 Description of Facility and Turbine ........................................................................ 9

2.1 Review of Turbine Test Facilities .............................................................................. 9 2.2 START Facility Requirements ................................................................................... 12 2.3 Facility Design ........................................................................................................... 15 2.4 Test Section ................................................................................................................ 27 2.5 Instrumentation .......................................................................................................... 34 2.6 Control and Safety Precautions .................................................................................. 41 2.7 Summary .................................................................................................................... 43

Chapter 3 Using a Tracer Gas to Quantify Sealing Effectiveness for Engine Realistic Rim

Seals ................................................................................................................................. 44

Abstract ............................................................................................................................ 44 3.1 Introduction ................................................................................................................ 44 3.2 Review of Literature .................................................................................................. 45 3.3 Test Facility and Test Turbine ................................................................................... 47 3.4 Facility and First Vane Benchmarking ...................................................................... 52 3.5 CO2 Instrumentation and Data Acquisition ................................................................ 55 3.6 Uncertainty and Repeatability .................................................................................... 58 3.7 Validating CO2 Sampling Methods ............................................................................ 59 3.8 Mate Face Gap Leakage Effects in Rim Seal ............................................................. 69 3.9 Conclusions ................................................................................................................ 71

Chapter 4 Effects of Purge Jet Momentum on Sealing Effectiveness ..................................... 73

Abstract ............................................................................................................................ 73 4.1 Introduction ................................................................................................................ 73 4.2 Review of Literature .................................................................................................. 74 4.3 Description of Facility and Turbine ........................................................................... 76 4.4 Facility and First Vane Benchmarking ...................................................................... 81

v

4.5 Sealing Effectiveness with Purge Flow ...................................................................... 86 4.6 Sealing Effectiveness with Mate Face Gap Leakage Flow ........................................ 94 4.7 Scaling of Sealing Effectiveness for Purge ................................................................ 98 4.8 Conclusions ................................................................................................................ 102

Chapter 5 Effects of Purge Flow Configuration on Sealing Effectiveness in a Rotor-Stator

Cavity ............................................................................................................................... 104

Abstract ............................................................................................................................ 104 5.1 Introduction ................................................................................................................ 104 5.2 Review of Literature .................................................................................................. 106 5.3 Description of Facility and Turbine ........................................................................... 108 5.4 Facility and Turbine Benchmarking ........................................................................... 114 5.5 Sealing Effectiveness for Purge Flow ........................................................................ 118 5.6 Empirical Modeling ................................................................................................... 129 5.7 Conclusions ................................................................................................................ 134

Chapter 6 Conclusions and Recommendations ........................................................................ 135

6.1 Conclusions ................................................................................................................ 136 6.2 Recommendations for Future Work ........................................................................... 138 6.3 Concluding Remarks .................................................................................................. 139

References ................................................................................................................................ 140

Appendix Design-Stage Uncertainty Analysis ....................................................................... 145

vi

LIST OF FIGURES

Figure 1.1 (a) T-s (temperature-entropy) diagram of the ideal Brayton power cycle; (b) p-

v (pressure-volume) diagram of the ideal Brayton power cycle; and (c) simplified

schematic of a gas turbine engine and several applications. ............................................ 2

Figure 1.2. Cross-section of a Pratt & Whitney 4000 turbofan engine, highlighting a

simplified schematic of a secondary air (adapted from [7]). ............................................ 4

Figure 1.3. Schematic of a secondary air system for an aviation engine (adapted from [8]).

.......................................................................................................................................... 5

Figure 1.4. A model of some of the mechanisms that influence hot gas ingestion. ................. 6

Figure 2.1. START facility design envelope in terms of axial and rotational Reynolds

numbers compared to other continuous flow turbine research facilities. ......................... 11

Figure 2.2. START facility layout showing a schematic of the infrastructure. ....................... 16

Figure 2.3. Model of START facility showing the three-dimensional arrangement of the

compressor, the facility piping, and the turbine test section. ........................................... 17

Figure 2.4. Photo of the (a) inlet piping, (b) facility compressor, (c) inlet throttling valve,

(d) unloading valve and exhaust piping, (e) main control supply valve, (f) motor, and

(g) main supply piping. .................................................................................................... 18

Figure 2.5. Turbine inlet pressure as provided by three configurations: (1) the original

pneumatic actuator on the facility exit pressure valve, (2) the new electric actuator on

the facility exit pressure valve with the compressor run in automatic mode, and (3) the

new electric actuator on the facility exit pressure valve with the compressor run in

manual mode. ................................................................................................................... 19

Figure 2.6. Photo of the test bay, including (a) the upstream venturi, (b) the upstream

settling chamber, (c) the clamshell casing, (d) the turbine test section, (e) the

downstream settling chamber, (f) the fast-closing valve, (g) the downstream venturi,

(h) the facility exit pressure valve, (i) the magnetic bearing controller, and (j) the water

brake dynamometer. ......................................................................................................... 20

Figure 2.7. Facility cooling equipment: (a) outdoor heat exchanger for compressor cooling

system, (b) chiller for cooling turbine secondary air, (c) pump skid for compressor

cooling system, (d-e) piping for compresssor cooling system. ........................................ 21

Figure 2.8. Cross-section of turbine test section (in dashed outline) and adjoining parts

(generic turbine and select components): (a) radial baffle plate for bypass piping, (b)

baffle plates in settling chamber, (c) upstream settling chamber, (d) test section center

body supported by struts, (e) support structure on linear rails, (f) turbine, (g) bearing

block, and (h) annular downstream settling chamber. ..................................................... 22

vii

Figure 2.9. Schematic of the secondary air supply for the test turbine with four independent

sources, with the purge and TOBI flows supplied to the inner diameter of the turbine

at the first vanes, and the two second vane flows supplied to the outer diameter of the

turbine at the second vanes. ............................................................................................. 24

Figure 2.10. Schematic of the water brake dynamometer system, showing the water flow

loop and the hydraulic oil flow loop for the dynamometer control valves. ..................... 25

Figure 2.11. Photo of the water brake dynamometer system: (a) dynamometer, (b) water

accumulator tank, (c) hydraulic oil pump, (d) dyno water inlet valve, (e) dyno water

exit valve. ......................................................................................................................... 26

Figure 2.12. The water brake dynamometer system required the excavation of a large hole

to install the hot and cold wells, as well as the pump vault and the underground piping.

.......................................................................................................................................... 27

Figure 2.13. Cross section of turbine with particular regions and flows called out: (a) first

vane plenum, (b) front rim seal, (c) front rim cavity, (d) front wheel-space, (e) purge

flow, (f) TOBI flow, and (g) aft rim cavity. ..................................................................... 29

Figure 2.14. Cross section of turbine with select non-dimensional radii. ................................ 30

Figure 2.15. Cross-section of rotor assembly: (a) bearing tube, (b) shaft, (c) radial bearing

stator, (d) radial bearing rotor, (e) thrust bearing rotor, (f) thrust piston supply, and (g)

thrust piston exhaust. ........................................................................................................ 31

Figure 2.16. (a) thrust piston air supply stand, and (b) air supply hoses to thrust piston

through the downstream settling chamber. ...................................................................... 32

Figure 2.17. (a) Generic cross-section of turbine rotor and bearing structure; (b) XLrotor

rotor dynamic model of turbine rotor. .............................................................................. 33

Figure 2.18. Campbell, or interference, diagram of turbine rotor showing rotor dynamic

modes of test turbine. ....................................................................................................... 34

Figure 2.19. Schematic showing the approximate locations of the facility and turbine

instrumentation. ................................................................................................................ 35

Figure 2.20. Cross-section of 1.5 stage turbine, showing turbine instrumentation locations.

.......................................................................................................................................... 38

Figure 2.21. Schematic of the facility programmable logic controller (PLC). ........................ 42

Figure 3.1. START facility layout. .......................................................................................... 48

Figure 3.2. First vane only test turbine cross-section. ............................................................. 50

Figure 3.3. Test turbine nomenclature and geometric parameter definitions. ......................... 50

Figure 3.4. Test turbine instrumentation. ................................................................................. 52

viii

Figure 3.5. Range of operation for these measurements. ......................................................... 53

Figure 3.6. Turbine inlet total temperature. ............................................................................. 54

Figure 3.7. Circumferential uniformity of the first vane aerodynamic loading at 50% span.

.......................................................................................................................................... 54

Figure 3.8. CO2 injection and test turbine sampling system. .................................................. 56

Figure 3.9. Uniformity of seed concentration in first vane plenum for (a) mate face gap

leakage ṁmfg = 0.35% and (b) mate face gap leakage ṁmfg = 0.15%. .............................. 57

Figure 3.10. Concentration measurements in the first vane plenum for a range of sampling

flow rates at different leakage flow rates. ........................................................................ 60

Figure 3.11. Benchtop experiment to validate gas sampling method with measurements in

a pipe with a similar velocity to the rim cavity swirl velocity, at a Mach number of

approximately 0.3. ............................................................................................................ 62

Figure 3.12. Comparison of concentration effectiveness in the rim cavity with varying

secondary air supply CO2 concentrations for two purge flow rates. ............................... 63

Figure 3.13. Concentration effectiveness measurements in the rim cavity for a range of

sampling flow rates and mate face gap leakage flow rates. ............................................. 64

Figure 3.14. Comparison of measured concentration effectiveness for varying sampling

flow rates in the rim cavity for purge ṁp = 0.9%. ............................................................ 65

Figure 3.15. Comparison of measured concentration effectiveness for varying sampling

flow rates in the rim seal for three mate face gap leakage flows. .................................... 67

Figure 3.16. (a) Sampling flow rate sensitivity at three axial positions in the rim seal, and

(b) concentration effectiveness across rim seal axial gap; both at ṁp = 1.6%. ................ 68

Figure 3.17. Concentration effectiveness measurements on the stator and rotor sides of the

rim seal for multiple purge flow rates. ............................................................................. 69

Figure 3.18. Concentration effectiveness on the stator side of the rim seal for multiple

circumferential locations. ................................................................................................. 71

Figure 4.1. START facility layout. .......................................................................................... 77

Figure 4.2. Test Turbine cross-section. .................................................................................... 80

Figure 4.3. Test turbine nomenclature and geometric parameter definitions. ......................... 80

Figure 4.4. Test turbine instrumentation. ................................................................................. 82

Figure 4.5. First vane aerodynamic loading at 50% span compared to CFD pre-test

predictions. ....................................................................................................................... 83

ix

Figure 4.6. Pressures on vane trailing edge, platform trailing edge (22% Cx downstream of

vane trailing edge), and in rim seal (8% Cx upstream of vane trailing edge) for the no

leakage case compared to CFD pre-test predictions. ....................................................... 84

Figure 4.7. Swirl Mach number in the trench region and the rim cavity for a range of purge

flow rates. ......................................................................................................................... 86

Figure 4.8. Circumferential uniformity of concentration effectiveness for 150 purge holes

for multiple purge flows. .................................................................................................. 87

Figure 4.9. Concentration effectiveness for 150 purge holes................................................... 89

Figure 4.10. Circumferential variation of concentration effectiveness for 16 purge holes on

the stator side of the rim cavity at the purge hole radius for multiple purge flows. ......... 91

Figure 4.11. Variation of concentration effectiveness with purge flow rate on the stator side

of the rim cavity at the purge hole radius for 16 purge holes. .......................................... 92

Figure 4.12. Averaged concentration effectiveness for 16 purge holes. .................................. 94

Figure 4.13. Concentration effectiveness for mate face gap leakage flow only (no purge). .... 96

Figure 4.14. Circumferential variation in concentration effectiveness on the stator and rotor

sides of the rim cavity for 16 purge holes and the mate face gap leakage. ...................... 97

Figure 4.15. Concentration effectiveness for 150 purge holes and 16 purge holes with

varying purge flow rates. ................................................................................................. 99

Figure 4.16. Mass flux ratio, M, and momentum flux ratio, I, for 150 purge holes and 16

purge holes. ...................................................................................................................... 100

Figure 4.17. Concentration effectiveness for 150 purge holes and 16 purge holes plotted

against blowing ratio. ....................................................................................................... 101

Figure 4.18. Concentration effectiveness for 150 purge holes and 16 purge holes plotted

against momentum flux ratio. .......................................................................................... 102

Figure 5.1. START facility layout, which houses the 1.5 stage turbine. ................................. 110

Figure 5.2. 1.5 stage turbine cross-section: (a) first vane plenum, (b) front rim seal, (c) front

rim cavity, (d) front wheel-space, (e) purge flow, (f) TOBI flow, and (g) aft rim cavity.

.......................................................................................................................................... 111

Figure 5.3. Turbine cross-section with instrumentation locations. Effectiveness data will

be presented for the following locations: (A) front rim seal, (B) outer radius of front

rim cavity, (C) purge hole radius of front rim cavity, and (D) front wheel-space. .......... 113

Figure 5.4. Conditions in the turbine for a typical test. ........................................................... 115

x

Figure 5.5. Static pressure normalized by the vane inlet total pressure at 50% span for (a)

first vane and (b) second vane. ......................................................................................... 117

Figure 5.6. Static pressure normalized by the inlet total pressure on the vane trailing edge,

the platform trailing edge (22% 1V axial chord downstream of vane trailing edge),

and in the rim seal (8% 1V axial chord upstream of vane trailing edge). ........................ 118

Figure 5.7. Sealing effectiveness measurements for Configuration 1: 150 purge holes and

Configuration 2: 32 purge holes. ...................................................................................... 121

Figure 5.8. Flow schematic of secondary flows in the 1.5 stage test turbine. .......................... 123

Figure 5.9. Schematic of flows in the rim cavity for the following configurations: (a) 150

purge holes at low flow rates, (b) 150 purge holes at high flow rates, (c) 32 purge holes

at low flow rates, and (d) 32 purge holes at high flow rates. ........................................... 124

Figure 5.10. Circumferential variation in concentration effectiveness for 32 purge holes. ..... 127

Figure 5.11. Empirical models for concentration effectiveness for 150 purge holes in terms

of the net and minimum sealing flow rates, ϕ* and ϕmin. ............................................. 130

Figure 5.12. Empirical models for concentration effectiveness for 32 purge holes in terms

of the net and minimum sealing flow rates, ϕ* and ϕmin. ............................................. 131

Figure 5.13. Comparison of several rim seals in terms of the empirically determined ratio

of discharge coefficients, Γc. ........................................................................................... 133

xi

LIST OF TABLES

Table 2.1. Engine vs START Lab Operating Conditions ........................................................ 14

Table 2.2. Summary of Facility Instrumentation ..................................................................... 36

Table 2.3. Summary of Turbine Instrumentation ..................................................................... 40

Table 2.4. Uncertainty in Facility and Turbine Measurements ................................................ 41

Table 3.1. START Facility Operating Conditions ................................................................... 47

Table 3.2. Overall Measurement Uncertainty .......................................................................... 58

Table 4.1. START Facility Operating Conditions ................................................................... 78

Table 5.1. Uncertainty in Facility and Turbine Measurements ................................................ 114

Table A.0.1 Uncertainty in Vane Aerodynamic Loading ........................................................ 152

Table A.0.2. Uncertainty in Efficiency .................................................................................... 153

Table A.0.3. Uncertainty in Flow Characterization Parameters .............................................. 154

Table A.0.4. Uncertainty in Sealing Effectiveness .................................................................. 155

xii

NOMENCLATURE

𝑏 hub radius

𝑐 gas concentration

𝐶𝑑,𝑖 discharge coefficient for ingress

𝐶𝑑,𝑒 discharge coefficient for egress

𝑐𝑝 specific heat at constant pressure

𝐶𝑥 axial chord length

ℎ height

𝐼 momentum flux ratio, 𝜌𝑝𝑉𝑝2/(𝜌𝑟𝑐𝑉𝑟𝑐

2)

�̇� mass flow rate

�̇�𝑚𝑓𝑔 mfg leakage flow based on full span turbine flow rate

�̇�𝑝 purge flow based on full span turbine flow rate

𝑀 mass flux (blowing) ratio, 𝜌𝑝𝑉𝑝/(𝜌𝑟𝑐𝑉𝑟𝑐)

𝑝 static pressure

𝑝𝑖 total pressure

𝑃𝑅 pressure ratio, 𝑝𝑡,𝑖𝑛/𝑝𝑒𝑥

𝑟 radius

𝑅𝑒𝜙 rotational Reynolds number, Ω𝑏2/𝜈

𝑅𝑒𝑥,𝑖𝑛 blade inlet axial Reynolds number, 𝑉𝑟𝑒𝑙𝐶𝑥,𝐵/𝜈

𝑅𝑒𝑥 vane exit axial Reynolds number

𝑠 spacing

𝑠𝑐 seal clearance

𝑆/𝑆𝑚𝑎𝑥 percent wetted surface distance on airfoils

𝑇 static temperature

𝑇𝑡 total temperature

𝑉 velocity

𝑥 axial direction

( )̿̿ ̿̿ ̿ area-averaged quantity

Greek

𝛾 ratio of specific heats

Γ𝑐 ratio of discharge coefficients, (𝐶𝑑,𝑖/𝐶𝑑,𝑒)

xiii

ε𝑐 concentration effectiveness, (𝑐 − 𝑐∞)/(𝑐𝑠 − 𝑐∞)

ε𝑐,𝑠 vane plenum supply concentration effectiveness, (𝑐1𝑉𝑃 − 𝑐∞)/(𝑐𝑠 − 𝑐∞)

𝜁 total pressure loss

𝜂𝑇𝑇 total-to-total efficiency

𝜂𝜏 torque-based efficiency

𝜃 circumferential direction

𝜇 dynamic viscosity

𝜈 kinematic viscosity (μ/ρ)

𝜌 density

𝜏 torque

𝛷 sealing flow rate, �̇�/(2𝜋𝑠𝑐𝜌Ω𝑏2)

𝛷0 flow rate at which effectiveness crosses zero

𝛷𝑚𝑖𝑛 minimum flow rate to seal a particular location

𝛷𝑟𝑒𝑓 reference flow rate, 𝜙𝑚𝑖𝑛 for front cavity

𝛷∗ net sealing flow rate, 𝜙 −𝜙0

Φ flow coefficient

𝜓 loading coefficient

Ω shaft rotational speed

Subscripts and Abbreviations

1V first vane

1VP first vane plenum

2V second vane

AM additive manufacturing

avg indicates an average quantity

B blade

CFD computational fluid dynamics

ex indicates a turbine exit parameter

in indicates a turbine inlet parameter

∞ indicates a main gas path parameter

jet purge flow jet

min minimum

mfg mate face gap

xiv

p purge

rc rim cavity

ref reference

rel relative

rs rim seal

s indicates a secondary flow supply parameter

TE trailing edge

TOBI tangential on-board injection (pre-swirled sealing air)

xv

ACKNOWLEDGEMENTS

First and foremost, I would like to thank my advisor, Karen Thole, for granting me the

tremendous opportunity to work on the development of the START Lab. It was eye-opening to

develop an experimental laboratory, especially considering my background doing numerical

simulations. Karen helped me make the transition. Although the transition was difficult, I truly

learned a lot from her. She pushed me—sometimes more than I wanted—to become a better

researcher, engineer, and instructor than I thought I could be. I could not have done this work

without her.

I would also like to thank Mike Barringer for his mentorship and friendship. Mike really

deserves the lion’s share of the credit for the lab, as he was involved before the START Lab was

even a dream. In addition to design, manufacturing, assembly, and purchasing, Mike took care of

many of the tasks for developing the lab that I would have been unable to do. This was truly a team

effort, and I thank Mike for allowing me to be a part of the team.

My PhD was different than most other doctoral students, as there were several of us

working on the lab in a team. Two research associates were also part of the START team. Andrew

Coward provided extensive work regarding the design of several facility and test section

components, as well as the programming of the data acquisition program. The lab would not have

come together as smoothly as it did without him. He also endured thousands of questions from me,

and provided vital feedback for the work I was doing. Dave Johnson joined our START team near

the end of my doctoral work, and he was instrumental in preparing the facility for rotation. We

spent many long hours together in the control room collecting data.

I would also like to thank the Department of Defense for awarding me with the National

Defense Science and Engineering Graduate (NDSEG) Fellowship, which funded the first three

years of my PhD work. If I had not been given the NDSEG Fellowship, I likely would not have

been able to work on the development of the START Lab as there was not funding for a student

early on. I also would like to thank the Department of Energy National Energy Technology

Laboratory. Although I did not directly receive funding from them, their sponsorship made the

START Lab possible.

I owe many thanks to Pratt & Whitney for funding the last couple years of my graduate

work. I’m very grateful I was given the opportunity to work on Pratt & Whitney turbine hardware

at the START Lab. I also thank the many colleagues—now friends—at Pratt & Whitney, for all the

xvi

support. There are too many of them to name, so I hope a general thanks will suffice. I am especially

excited to begin my career at Pratt & Whitney working on dependable engines.

Thanks also go to my children, Rylee and Sierra, for providing me an escape from the lab.

I knew when I came home for dinner that they would help me forget the many frustrations that I

experienced daily. They helped revitalize and reinvigorate me to face the next day’s challenges.

Because of them, I was able to lead a more balanced life. They bring more joy to my life than I

ever imagined.

Lastly, I thank my wife, Kari, for her years of sacrifice. It’s been a long and difficult road,

and I profoundly appreciate her understanding and attitude when I had to work long days, nights,

and weekends. Truly, I could not have finished this work without her. She believed in me and

motivated me when I was ready to give up. This degree belongs to her as much as it belongs to me.

1

Chapter 1

Introduction

A major challenge that gas turbine designers face is preventing hot gas from being ingested

into the cavities between stationary and rotating hardware where components do not feature the

same cooling technologies found in the main gas path. Hot gas within turbine inter-stage cavities

leads to component durability concerns. Rim seals are typically used at the boundary between the

cavity and the main gas path to reduce the amount of ingested flow, and secondary flow is used to

purge the cavities of ingested hot gas. Excessive use of the purge flow, however, has a

thermodynamic penalty that leads to a reduction in the engine efficiency. A few reduced-order

models have been developed to predict hot gas ingestion into inter-stage cavities, but these models

have been based on simplified geometries and for operating conditions not comparable to those of

an operating engine. This dissertation provides unique measurements for a turbine rim seal and

cavity design representative of modern gas turbine hardware at engine-relevant conditions.

1.1 Background and Motivation

Gas turbines account for a large portion of the U.S. energy expenditures consuming 15%

of the U.S. energy for air transportation and electricity generation [1]. For air travel in the U.S.

alone, 1.54 million barrels of jet fuel are consumed daily, equating to a fuel cost of $88 million and

emissions in excess of 600,000 metric tons of CO2 [2–4]. Although gas turbines are characterized

by high efficiencies compared to other power generation methods, small gains in efficiency are

highly desirable due to high fuel costs and high greenhouse gas emissions. Huge cost and emissions

savings can be achieved with small gains in engine efficiencies. Consider that if aircraft engines

reduced overall fuel burn by 5%, then a fuel cost savings of over $1.5 billion per year would be

realized, and the reduction in CO2 emissions would be equivalent to the removal of over 2.4 million

passenger vehicles from roads [5]. Clearly, it is important to perform research that can lead to

efficiency increases in gas turbines as these numbers will continue to grow.

2

Gas turbines can be modeled as a Brayton power cycle. The ideal Brayton cycle, shown on

a temperature-entropy diagram in Figure 1.1a and on a pressure-volume diagram in Figure 1.1b,

consists of an isentropic compression process (1 to 2), an isobaric heat addition process (2 to 3), an

isentropic expansion process (3 to 4), and an isobaric heat rejection process (4 to 1). In gas turbine

engines, the compression process occurs in a compressor, as shown in Figure 1.1c. The heat

addition occurs through a direct combustion process, whereby fuel is injected into a combustor and

burned to increase the enthalpy of the flow. The flow then passes through the turbine, and as it

expands in the turbine the enthalpy is converted into mechanical energy in the form of a rotating

shaft, as shown in Figure 1.1c. The shaft connects the turbine to the compressor, and a portion of

the work extracted from the turbine drives the compressor, creating a self-sustaining process.

Figure 1.1 (a) T-s (temperature-entropy) diagram of the ideal Brayton power cycle; (b) p-v

(pressure-volume) diagram of the ideal Brayton power cycle; and (c) simplified schematic of

a gas turbine engine and several applications.

T

s

1

2

3

4

1 2 3 4

Compressor

Combustor

Turbine

(a) (b)

Nozzle

TurbineShaftPropellerFanGenerator

p

v

1

2 3

4

(c)

3

At a point in the cycle denoted by the star symbol in Figure 1.1, there is still a substantial

amount of enthalpy remaining, which can be used for a variety of purposes as noted in Figure 1.1c.

The flow can pass through a nozzle and the enthalpy can be converted into a high speed jet to

provide thrust as in a turbojet, or the flow can pass through an additional turbine to provide

mechanical energy to rotate a power shaft, turn a propeller or a fan, or spin an electric generator.

Compared to other forms of power generation and propulsion, gas turbines provide high power

density with high efficiencies, which makes them ideal for a variety of applications ranging from

power plants to commercial passenger jets to oil drilling platforms. For power generation, high

overall efficiencies can be achieved, in excess of 61% [6], by using the high temperature exhaust

gas at state 4 in Figure 1.1 to provide heat for secondary cycles in combined cycle power plants.

The efficiency of the ideal Brayton cycle is directly related to the pressure ratio across the

compressor, commonly called the overall pressure ratio (OPR), by . As OPR increases, the

thermodynamic efficiency of the ideal Brayton cycle increases. Aircraft engines are subject to

additional constraints including propulsive efficiency, nacelle drag, and weight issues, but OPR

still determines the thermodynamic efficiency of the gas turbine core. Increasing the engine OPR

also increases the engine temperatures, causing extreme thermo-mechanical loads on engine

components, which can result in reduced component lifetimes, decreased reliability, and high

maintenance costs. Maintaining component durability is done through two methods: (1) using high

temperature materials and coatings, and (2) using compressor bleed air to cool components. High

temperature materials and coatings are crucial to the development of more efficient and durable

engines, but this topic beyond the scope of this dissertation.

The compressor bleed air is supplied to various locations in the hot section of the engine

through a complex flow network called the secondary air system. A cross-section of a Pratt &

Whitney PW4000 turbofan engine [7] is shown in Figure 1.2, and the inset image focuses on the

region near the exit of the high pressure compressor, the combustor, and the high pressure turbine.

Several arrows overlaid on the inset image schematically show a highly simplified secondary air

system. Due to the high thermo-mechanical loads on the hot section of the engine, the secondary

air system comprises as much as 25-30% of the core engine flow in a modern gas turbine [1], and

can be quite complex, as shown in Figure 1.3 [8].

The turbine is composed of alternating rows of stationary vanes and rotating blades, and

the regions between the stationary and rotating components are called rotor-stator cavities, as

shown in Figure 1.3. Flow restrictions are placed on the outer edges of the rotor-stator cavities,

which are called rim seals, to prevent hot gas from the main gas path from being ingested into the

4

rotor-stator cavities, as shown in Figure 1.3. These rim seals are intentionally complex to minimize

ingestion, but their design is subject to several constraints. As turbine engines experience thermal

transients, the rotors move axially and radially in relation to the stators, so the rim seals must

account for this movement to seal under these conditions without rubbing. In addition, as disk

rotational speed increases, the acceleration of the air causes the pressure inside the rotor-stator

cavities to decrease, drawing hot gas from the main gas path into the cavity. Work transfer from

the spinning disk also increases the temperature of the air in the cavity. A portion of the secondary

air, also called purge flow as shown in Figure 1.3, is directed into the cavities and rim seal areas to

purge the ingested hot gas and maintain acceptable temperatures and component durability.

The ingestion of hot gas into turbine rim cavities is a complex problem and is the topic of

this dissertation. A model of a turbine stage is given in Figure 1.4 to briefly describe the physics of

ingestion. The main gas path flow enters the stationary vanes from the left in Figure 1.4 where the

aerodynamic design of the vane passage determines the pressure profile at the vane exit. The

regions of high pressure drive hot gas ingestion through the rim seal into the cavity and the regions

of low pressure allow the cavity flow to egress through the rim seal into the main gas path. The

Figure 1.2. Cross-section of a Pratt & Whitney 4000 turbofan engine, highlighting a simplified

schematic of a secondary air (adapted from [7]).

5

blade passing has a potential field that also influences the boundary conditions on the rim seal. The

unsteady interaction of the blade passing potential field with the stationary vane exit pressure

profile further complicates the flow field in the vicinity of the rim seal. High levels of turbulent

mixing occur where the low momentum flow from the rim seal interacts with the high momentum

flow in the main gas path. Radially inboard of the rim seal, the rotation of the rotor disk causes a

tangential acceleration of the fluid in the stationary reference frame, resulting in a radially outward

flow of the boundary layer fluid commonly known as disk pumping, as shown in Figure 1.4. To

satisfy continuity, fluid travels axially across the cavity toward the disk. In turbine rim cavities, the

pumping effect causes the fluid to pump radially outward on the rotor side and inward on the stator

side, which draws ingested hot gas farther inboard of the rim seal. Although reduced order models

based on simplified rim seal and cavity geometries have been developed to predict ingestion,

ingestion itself is quite complex. Moreover, the geometries in operating turbines are more complex

than the simplified geometries published in the literature. The empirical nature of the models makes

it difficult to predict the performance of new rim seal designs, which can lead to supplying

insufficient or excess purge flow to the turbine cavity.

The secondary air, which includes both the airfoil cooling as well as the purge flow,

negatively affects the engine efficiency in two ways. First, the secondary air bypasses the

Figure 1.3. Schematic of a secondary air system for an aviation engine (adapted from [8]).

6

combustor, as shown in Figure 1.2, greatly decreasing the full work potential (enthalpy) of the air.

The compressor must still impart work on the secondary air to pressurize it so it can be used for

cooling and sealing in the high pressure turbine. Second, the secondary air reduces engine

efficiency by reintroducing the low momentum secondary air into the high-momentum flow of the

main gas path causing irreversible mixing losses and entropy rise. As can be seen in Figure 1.3, the

secondary air passes through a complex flow network, but is eventually injected into the main gas

path flow resulting in mixing losses. Johnson et al. [9] showed that decreasing the rim seal purge

flow by 50% for a two-stage turbine could increase the turbine efficiency by 0.5% and reduce fuel

burn by 0.9%. A similar decrease for a four-stage turbine could increase the turbine efficiency by

1.4%. A more recent study by Glahn and Schmitz [1] showed that a savings in the secondary air

system of 10% of the core inlet flow of an aircraft engine would result in a reduction in the overall

fuel burn of 5%. The secondary air usage should thus be minimized to maintain engine efficiency,

but not at the cost of engine durability.

Figure 1.4. A model of some of the mechanisms that influence hot gas ingestion.

Blade potential

field

Vane exit pressure profile

TurbulentMixing

Hot gas ingestion

Rim seal egress

Stator pumping Disk

pumping

Main gas path flow

7

The U.S. Department of Energy – National Energy Technology Laboratory has a goal of

increasing the efficiency of land-based, combined-cycle, power plants by 3-5% [10]. As mentioned

at the beginning of this chapter, aircraft engines consume over 1.5 million barrels of jet fuel daily,

and a 5% reduction in overall fuel burn for these aircraft engines would result in a fuel cost savings

of over $1.5 billion per year and a reduction in CO2 emissions equivalent to eliminating over 2.4

million cars from roads. The motivation for the research presented in this dissertation is to minimize

the use of the secondary air in gas turbines to increase the efficiency of turbine engines without

compromising the durability of the engine components.

1.2 Objectives and Uniqueness of Research

The vast majority of research regarding hot gas ingestion into inter-stage turbine cavities

has used simplified geometries and has been performed at reduced Reynolds numbers and Mach

numbers compared to operating engines. Yet, these models present the industry standard for

predicting ingestion in actual turbines. Although several fundamental studies in the literature have

provided important learning regarding hot gas ingestion and cavity flow physics, there is a need to

perform experiments that combine all the physics simultaneously. The primary objective of this

dissertation is to present unique sealing effectiveness measurements and to further understand the

fundamental physics for an engine-realistic rim seal operating at engine-relevant axial Reynolds

numbers, rotational Reynolds numbers, and Mach numbers. To accomplish the primary objective,

this dissertation also provides a description of a newly developed, unique, continuous flow, turbine

test facility, capable of simulating engine-relevant conditions. Because significant efforts for this

doctoral research was also spent on developing this new facility, a detailed description will be given

in this dissertation.

The majority of the measurements presented in this dissertation include sealing

effectiveness measurements, which are determined by using a flow tracer gas in the secondary air

supply and measuring the gas concentration in the turbine. Concentration measurements are

obtained by sampling flow through pressure taps in the turbine rim seals and cavities, which are in

locations only made possible through the use of additive manufacturing of the airfoils. While this

particular experimental technique is not new the measurement locations are unique for an engine-

realistic geometry. Extensive validation experiments are performed to ensure the technique is able

to acquire high quality sealing effectiveness measurements. This dissertation presents an important

8

validation study for this experimental technique that has not been previously reported in the

literature. The sealing effectiveness measurements are also shown to be a powerful tool at deducing

important flow details in the rim seals and cavities.

To provide data relevant to engine designs, it is necessary to perform experiments using

engine-realistic hardware—including airfoils, rim seals, rim cavities, leakages, and purge flow

delivery methods. The turbine airfoils, rim seals, and rim cavities in this research are representative

of a modern gas turbine design. The turbine includes several engine-realistic leakages, such as

through the gaps between mating faces of adjoining airfoils or through the gaps between the

individual blades and the disk. Additionally, most studies in the literature provide the purge flow

at the center of rotation, but typically operating engines have a more complex purge flow delivery.

The purge flow in this research is provided in an engine-realistic manner through axial holes

directly into the rim cavity. Because of the engine-realistic geometry, leakages, and purge flow

delivery, the cavity flow field is more complex than is typically represented in the literature,

highlighting the importance of performing research with engine-realistic hardware.

1.3 Outline of Dissertation

This dissertation is presented in a manuscript format that includes three individual research

papers that describe this work, of which two have been peer-reviewed and already published.

Chapter 2 describes the design, implementation, and commissioning of the facility, the turbine, and

instrumentation. Chapters 3 through 5 present the methods, results, and discussions of this

dissertation through the publications. The first paper, presented in Chapter 3, provides a detailed

description of the measurement technique to quantify the sealing effectiveness of the rim seals. The

second paper, presented in Chapter 4, quantified the sealing effectiveness of an engine-realistic rim

seal with a stationary seal and cavity to isolate the effects of main gas path flow from rotational

effects. The third paper, presented in Chapter 5, provides sealing effectiveness measurements for

the same engine-realistic rim seal geometry presented in Chapter 4 but with rotation in a 1.5 stage

turbine. The well-accepted orifice model by Owen et al. [11] is also compared to the sealing

effectiveness measurements in Chapter 5. The overall conclusions and recommendations from the

work presented in this dissertation are provided in Chapter 6.

9

Chapter 2

Description of Facility and Turbine

The experiments presented in this dissertation were conducted in the Steady Thermal Aero

Research Turbine (START) Laboratory at the Pennsylvania State University. The START Lab was

developed to study aerodynamics, heat transfer, and secondary air systems using turbine hardware

from modern operating engines at continuous, engine-relevant conditions. Significant facility

infrastructure was required to achieve the required operating conditions. This chapter describes the

facility equipment and infrastructure and briefly describes the test turbine. The design of the facility

was previously described by Barringer et al. [12].

There are only a few either university or government laboratories in the world that house a

rotating test turbine for research purposes. Most of these laboratories house continuous flow rigs,

but there is only one at the present time that is able to simulate engine-relevant conditions, which

is described in this paper. A brief review of relevant turbine research facilities will be provided in

this chapter. The requirements for the facility to achieve engine-relevant conditions will be

discussed. The design of the facility, turbine test section, and instrumentation will be presented.

The chapter then concludes with a description of the facility controls and safety precautions.

2.1 Review of Turbine Test Facilities

Several rotating turbine rig facilities can be found in the open literature [12]. Most of the

research from these turbine facilities has focused on aerodynamics or heat transfer for main gas

path hardware. Secondary air systems, rotor-stator cavity flows, and rim seal performance have

also been investigated in a few rotating rigs to understand hot gas ingestion, where it is crucial to

include rotational effects. These turbine research facilities have used one of two methods to

generate the operating conditions at the turbine: (1) short duration flow or (2) continuous duration

flow.

10

Short duration rotating turbine facilities have provided engine-relevant flow conditions at

the turbine test section for a short time period, typically through a shock tube or a blowdown

facility. These rigs have delivered important data for main gas path aerodynamics, airfoil heat

transfer, and unsteady interactions at engine-corrected conditions for true scale turbine engine

hardware. Short duration rigs typically operate on the order of 200 ms to 2 seconds, so certain

questions arise regarding the development of the aerodynamic and thermal flow fields, flow

steadiness, and applicability of the data to engines. Short duration facilities will not be discussed

further in this paper, but for a more detailed description the reader is referred to Refs. [13–23].

Continuous duration rigs are capable of providing steady flow conditions at the turbine. To

operate for long periods of time, these rigs often need to make several simplifications compared to

an operating turbine, such as using simplified geometries, operating with partially loaded turbines,

or testing at reduced flow conditions by scaling the turbine hardware. Few continuous flow turbine

test facilities can simulate engine-relevant conditions using true-scale fully loaded turbines. Several

of the continuous duration turbine facilities described in the following paragraphs are shown in

Figure 2.1 for both the United States and Europe [24]. Figure 2.1 shows the published capabilities

of each facility in terms of the blade inlet axial Reynolds number versus the rotational Reynolds

number. As shown in Figure 2.1 for reference, typical high pressure turbines in operating gas

turbine engines run at a rotational Reynolds number of 2x107 to 3x107 [25].

Some continuous duration flow rigs use partially loaded blades to eliminate the need for a

dynamometer. Such rigs could be characterized as nearly full stage turbines. Roy et al. [26,27]

operated a continuous flow turbine rig with partial span airfoils using partially loaded blades to

study hot gas ingestion in a rotor-stator cavity. Two distinct turbine airfoil configurations are

highlighted in Figure 2.1 as “ASU 1” and “ASU 2”. Sangan et al. [28] developed and operated a

steady state rig with reduced span vanes and partially loaded blades. This particular rig, identified

in Figure 2.1 as “Bath”, was designed for the rapid removal and installation of different test articles

leading to the testing of several rim seal geometries, as presented in Ref. [29]. Optical access on

both the stationary and rotating sides of the cavity led to important thermal measurements and a

better understanding of cavity flows [30]. More recently Patinios et al. [31] described a near-

identical turbine rig at Bath (same operating point in Figure 2.1), but it operated with fully loaded

blades in a 1.5 stage configuration.

11

Figure 2.1. START facility design envelope in terms of axial and rotational Reynolds numbers

compared to other continuous flow turbine research facilities.

Other continuous duration flow rigs have operated at low speed conditions, and thus

utilized large scale hardware, to match the turbine airfoil axial Reynolds numbers. An example of

such a turbine rig was the cold flow turbine rig developed by Lakshminarayana et al. [32], which

operated at low speed conditions using large scale hardware. The use of large scale hardware

allowed for highly detailed spatial measurements not often found in rotating turbine test facilities,

but the rig did not operate at engine-relevant rotational Reynolds numbers and Mach numbers.

A few continuous duration turbine rigs have used true-scale fully loaded turbines. The rig

described by Gallier et al. [33], identified as “Purdue: LSRT” in Figure 2.1, was such a turbine

facility. A two stage axial turbine facility described by Sell et al. [34] was used to study unsteady

flows and turbine aerodynamic performance. A focus of the research was on the second stage

engine-representative low pressure turbine and the effects of multiple stages. Schmitz [35]

described a closed loop, continuous flow 1.5 stage low pressure turbine research facility, identified

as “ND – ART” in Figure 2.1. The ND-ART rig included a high-work and highly-loaded turbine

that was used to study main gas path aerodynamics and secondary air systems. A full stage turbine

rig described by Bohn et al. [36], identified as “RWTH: Aachen” in Figure 2.1, was used to study

ingestion and rim seal geometries. Although the turbine main flow path and secondary air supply

PSU START

GE HGIR

Sussex

RWTH: Aachen

ASU 1

ASU 2

Bath

Purdue: LSRT

ND - ART

Blade Inlet Reynolds Number,

∞ ∞

Rotational Reynolds Number,

,

NASA W6

105

104

106

105 106 107

This research

12

used simplified geometries compared to operating gas turbine engines, the turbine was used to

make detailed pressure measurements in the cavity and flow field measurements at the rim seal

providing an important physical understanding of the cavity egress flows [36,37].

A two-stage turbine described by Coren et al. [38], identified as “Sussex” in Figure 2.1,was

used to study inter-stage leakages and secondary air delivery. Multiple secondary air delivery

modifications allowed for systematic testing of the cooling flow delivery methods and the effects

on the cavity flows. The experiments provided valuable validation data sets for numerical studies

by Coren et al. [39], Dixon et al. [40], and Andreini et al. [41]. A continuous flow turbine rig

described by Palafox et al. [42], identified as “GE HGIR” in Figure 2.1, was developed to study

rim seal geometries. The turbine rig was located at a private General Electric research facility and

was not open to the public. The rig included a 1.5 stage turbine with a bladed disk and a modular

rim seal design. The experimental data were useful in validating reduced order computational

models by Ding et al. [43].

Many of the turbine rigs described have used simplified geometries in the main gas path as

well as the secondary flow path. Figure 2.1 shows that in terms of rotational Reynolds number the

operating envelope of most continuous flow rigs is an order of magnitude lower than typical engine

conditions. Given what was currently available, there was a need for simulating engine-relevant

Reynolds numbers and Mach numbers using engine realistic hardware. Research is needed that

reduces turbine cooling and leakage flows, minimizes hot gas ingestion, and maintains component

durability. The START facility was developed to provide the means to accomplish this research. A

major goal of the research in the START facility was to validate reduced order and physics based

design tools.

The intent for the START facility, and the work conducted for this dissertation, was to push

the operating rotational Reynolds number out to the same order of magnitude as operating gas

turbine engines while operating at an engine-relevant blade inlet Reynolds number using engine

hardware. Achieving this goal created unique test capabilities for the START facility. The design

operating envelope for the START facility, which most closely matches that of an operating gas

turbine engine, is also shown in Figure 2.1.

2.2 START Facility Requirements

Significant facility infrastructure is required to operate at engine-relevant conditions but

with reduced temperatures and pressures. Although engines operate at much higher pressures,

13

temperatures, and flow rates than turbine research facilities, if relevant non-dimensional parameters

are matched, then the physics scale from rig conditions to engine conditions [44]. These non-

dimensional parameters include Reynolds number, Mach number, rotational Reynold number,

pressure ratio, corrected mass flow rate, corrected speed, and density ratio. The START facility

was designed to push the operating envelope of turbine test rigs to engine-relevant Reynolds

numbers as shown in Figure 2.1. A summary comparison of aero-engine cruise conditions and the

START turbine design operating conditions is given in Table 2.1. Matching each of these non-

dimensional parameters required significant equipment and infrastructure which will be described

in the next section. This section describes the requirements of the facility to operate at engine-

relevant conditions.

A continuous air flow in excess of 5 kg/s (11 lbm/s) and 400 kPa (60 psia) was required to

simulate the correct axial Reynolds number, Mach number, and corrected mass flow rate for a 1.5

stage turbine. Note, however, this requirement only provided enough flow to simulate

approximately half of the main gas path flow so partial-span airfoils were used to maintain the

correct Mach number. Given the focus of this dissertation is on the cavity flows, partial-span airfoils

were relevant as will be described later in this chapter. An industrial compressor, described in the

next section, was specified and purchased to provide the required air flow. Future research in this

facility will use full span airfoils whereby a second identical compressor will be integrated into the

flow path.

To match the corrected rotational speed of the engine the START turbine rotational speed

needed to be lower than the engine due to the lower operating temperature, since corrected speed

�̇�𝑐 Ω/√𝑇𝑡/𝑇𝑟𝑒𝑓. Typical test rigs operate at least an order of magnitude lower in rotational

Reynolds number than engines (see Figure 2.1), so to operate closer to an engine-relevant rotation

Reynolds number the START turbine needed a rotational speed near 10,000 RPM. The goal was to

use engine hardware to simulate realistic features and reduce costs so aero-engine turbines of 0.4 –

0.6 m diameter could be tested in the facility. This required a significant design effort, including

high performance bearings, accurate rotor dynamics modeling, and high-tolerance manufacturing.

A substantial temperature difference is required to achieve engine-relevant density ratios.

The research presented in this dissertation focused on the secondary air flows inboard of the

platform, where matching engine density ratios was not a requirement. Future research plans for

the lab include airfoil aerodynamic and heat transfer measurements which should be performed at

a matched density ratio.

14

Since the research in this dissertation focused on the secondary air system inboard of the

platform, the airfoil span was reduced to produce engine-relevant conditions with a reduction in

the mass flow rate. This design decision allowed for one industrial compressor to provide enough

flow to produce the correct pressure ratio, flow rate, and Reynolds number for a 1.5 stage turbine.

As indicated in Table 2.1 the research presented in this dissertation was performed at the correct

Mach number (pressure ratio), but at slightly lower rotational and blade axial Reynolds numbers.

This operating point is also noted in Figure 2.1. The turbine inlet pressure was reduced from the

design value, which resulted in lower density and thus lower rotational and blade axial Reynolds

numbers for the studies conducted in this dissertation. The reason for the decrease in the turbine

inlet pressure was because, at the time the measurements were obtained, the thrust piston, which

will be discussed in more detail later, was slightly underpowered and was unable to provide

sufficient counter thrust at the design condition. (Note that the full thrust piston operability has

since been resolved, and the thrust piston is now fully operational at the time of the writing of this

dissertation.)

As was mentioned, the use of partial-span airfoils required one compressor, which used

significant electric power in excess of 1 MW (1500 hp). Future research in the START Lab will

include testing full span airfoils which will require two compressors, as indicated in Table 2.1.

Operating two large industrial compressors and their associated cooling systems will require a

Table 2.1. Engine vs START Lab Operating Conditions

Parameter Aero-engine

(cruise)

START

Phase 1

START

Phase 2

This

Research

Coolant-to-Mainstream

Density Ratio (𝝆∞/𝝆𝒑)

(at first vane exit)

2.0 1.3 2.0 1.1

Stage Pressure Ratio

(𝒑𝒕,𝒊𝒏/𝒑𝒔,𝒆𝒙) 2 1.5 - 2.5 1.5 – 2.5 ---

Rotational Reynolds

Number (𝑹𝒆𝝓)

(at first vane exit)

2x107+ ≤1x107 ≤2x107 3.8x106

Rotational Speed (rpm) 15,000+ ≤11,000 ≤13,000 ---

Mass flow rate (kg/s) 25+ 5.7 11.4 ---

Axial Reynolds

Number (𝑹𝒆𝒙)

(at blade inlet)

3x105 3x105 3x105 1.4x105

Mach (𝑴𝒂)

(at first vane exit) 0.65 0.65 0.65 0.65

Airfoil Geometry True Scale,

Full Span

True scale,

Partial span

True Scale,

Full Span

True scale,

Partial span

15

substantial power supply, greater than 2 MW (3000 hp). To accommodate the near term and future

capabilities 46 kV power line was installed at the building with a substation and transformers down

to 4160 V. The power line and substation installation are capable of providing enough power to

start and operate both compressors and their cooling systems for future testing.

2.3 Facility Design

During the course of this dissertation, the laboratory that housed the START test turbine

was designed and constructed. The facility was composed of three main rooms as shown in Figure

2.2: (1) a compressor room, (2) a test bay, and (3) a control room. The compressor room housed

the compressor and many of its subsystems, while the test bay housed the majority of the facility

piping, the test turbine, the thrust piston, the magnetic bearing system, and the dynamometer

infrastructure. The control room was separated from the test bay and enclosed by concrete-filled

cinder block walls with 2.5 cm (1 in) thick ballistic grade glass for safety. It also housed the data

acquisition equipment and personnel during testing. Additionally, several equipment items and

infrastructure shown in Figure 2.2 were installed outside the laboratory including a heat exchanger,

chiller, water cooling system for the dynamometer, and an electrical substation (not shown in the

figure). Each of the major infrastructure components of the facility will be described in this section,

including the compressor, the facility piping, the secondary air supplies, and the water brake

dynamometer. A model of the facility is given in Figure 2.3, and several features of the facility will

be referenced in this figure throughout this section.

16

Figure 2.2. START facility layout showing a schematic of the infrastructure.

Settling Chamber

Venturi

Turbine Cooling:Heat Exchanger

Coolant Pipes

TURBINE

Compressor Cooling SystemOutdoor Heat Exchanger

(800 kW)

Turbine Cooling SystemOutdoor Chiller

(200 kW)M

oto

rSt

arte

r CoolingTower

Overhead Crane

Pump System

Flow Meters

Building Back Wall

Roof Exhaust

Turbine Cooling Air

RoofIntake

COMPRESSORS

Motor

Building Exterior Wall

Motor

By-Pass

PLC

HotWell

Dynamometer Water System

Pump Chamber

ColdWell

Mo

tor

Star

ter

Controls+ DAQ System

COMPRESSOR ROOM (14m x 11m)

CONTROL ROOM(7.5m x 3m)

TEST BAY ROOM(14m x 12m)

Tank

HydraulicPump for

Dyno ValvesOil

H2O Treatment

H2O Pre-Treatment

DynoH2O FillMakeup

TP air standOil cooler

Intercooler

17

An industrial compressor capable of 5.6 kg/s (12.5 lbm/s) of air flow at a pressure of 480

kPa (70 psia) was installed in the compressor room of the START Lab as shown in Figure 2.4. The

single compressor was installed on a 1 m thick concrete pad isolated from the rest of the building

to ensure no vibrations were transmitted to or from the compressor. Air entered the compressor

through an inlet filter and weather hood, which was installed on a platform above the roof of the

compressor room as shown in Figure 2.3. The compressor inlet pressure was controlled with an

inlet throttling valve as shown in Figure 2.4. An unloading valve and exhaust piping, also shown

in Figure 2.4, were installed to allow the compressor to ramp up to its set point smoothly without

affecting the rest of the facility.

Figure 2.3. Model of START facility showing the three-dimensional arrangement of the

compressor, the facility piping, and the turbine test section.

18

The compressor control panel allowed the user to define a set point, and the compressor

logic controller determined the inlet throttling valve and unloading valve positions needed to

produce the desired flow and pressure. The compressor simultaneously controlled both the inlet

throttling valve and unloading valve to hold the set point, but the combination of the valves caused

the compressor discharge pressure to vary approximately ±5 kPa (±0.75 psi) from the set point

pressure, resulting in undesirable unsteadiness at the turbine inlet. This variation is shown in Figure

2.5 for the original pneumatic actuator on the facility exit pressure valve, which performed poorly.

The original pneumatic actuator on the facility exit pressure valve was replaced with an electric

actuator as seen in Figure 2.6 that performed much better, as shown in Figure 2.5, but the turbine

inlet pressure was still not adequately steady. During operation the user was able to manually

override the compressor logic controller and manually set the inlet throttling valve and unloading

valve positions to stabilize the discharge pressure. Typically the compressor discharge pressure

variation decreased to within ±0.34 kPa (±0.05 psi) of the mean operating pressure during this

manual operation, allowing for a steady turbine inlet pressure as shown by the electric actuators

with the compressor in manual mode in Figure 2.5. Although not shown graphically, the turbine

Figure 2.4. Photo of the (a) inlet piping, (b) facility compressor, (c) inlet throttling valve,

(d) unloading valve and exhaust piping, (e) main control supply valve, (f) motor, and (g)

main supply piping.

19

inlet and exit mass flow rates were also steadier after installing the electric actuators and using the

compressor in manual mode. The turbine inlet mass flow rate remained steady within ±0.14% of

the mean, and the turbine exit mass flow rate remained steady within ±0.20% of the mean. Note

that the overall uncertainty of the turbine inlet pressure and mass flow rates, as well as several other

parameters, are given in Table 2.4 in Section 2.5. The compressor was typically operated by setting

the test section flow conditions with the compressor in automatic control mode. Once the desired

test section flow condition was established, the compressor was put in manual mode to ensure

steady operation during research testing. For the experiments presented in this dissertation the

compressor and facility were operated in this manner.

Figure 2.5. Turbine inlet pressure as provided by three configurations: (1) the original

pneumatic actuator on the facility exit pressure valve, (2) the new electric actuator on the

facility exit pressure valve with the compressor run in automatic mode, and (3) the new

electric actuator on the facility exit pressure valve with the compressor run in manual

mode.

0 5 10 15 20 25

Pt,in

[psia]

Time (minutes)

1 psi

Electric actuator (auto mode)

Electric actuator (manual mode)

Original pneumatic actuator

20

A 1.1 MW (1500 hp) fixed speed electric motor shown in Figure 2.4 was mounted on the

compressor to drive the impellers of the two-stage centrifugal compressor. A motor starter was

required to ensure safe and reliable start-up of the motor, and to manage the current in-rush during

start-up. The locations of the motor and starter are shown in the top left of the compressor room in

Figure 2.2.

The compressor discharge air temperature from the second stage was approximately 395

K (710° R). The air between the first and second compression stages was cooled with an intercooler

on the compressor to prevent overheating the second stage components as shown in Figure 2.2. The

compressor oil also needed to be cooled with a low temperature water-glycol mixture that was

supplied to the intercooler and oil cooler through an auxiliary pump system. The pump supplied

over 380 liters per minute (100 gallons per minute) of coolant flow to the coolers. The hot water-

glycol was pumped to an outdoor heat exchanger rated to 800 kW (2.7e6 BTU/hr) which removed

the heat load through a bank of fans as shown in Figure 2.2. Figure 2.7 also shows photos of the

facility cooling equipment, including the heat exchanger, pump, and piping.

Figure 2.6. Photo of the test bay, including (a) the upstream venturi, (b) the upstream

settling chamber, (c) the clamshell casing, (d) the turbine test section, (e) the downstream

settling chamber, (f) the fast-closing valve, (g) the downstream venturi, (h) the facility exit

pressure valve, (i) the magnetic bearing controller, and (j) the water brake dynamometer.

21

A model of the facility piping is shown in Figure 2.3, and photos of the piping in the test

bay are shown in Figure 2.6. The main facility piping was designed to withstand pressures up to

1.1 MPa (160 psia) and temperatures up to 670 K (1200° R). All piping upstream of the test section

was manufactured from stainless steel to prevent rust from entering the turbine test section, and all

piping downstream of the test section was manufactured from carbon steel and was painted to

protect it from corrosion.

The high pressure air from the compressor was sent through 0.2 m (8 in) piping to the

unloading piping and the main supply piping as shown in Figure 2.3. The unloading piping was

directed to the roof where the air passed through a silencer. The main supply piping was directed

to the turbine test bay and passed through a pneumatically-controlled flow control valve shown in

Figure 2.4, which was used to control the turbine inlet pressure. The pipe diameter downstream of

the flow control valve shown in Figure 2.4 increased to 0.3 m (12 in) to minimize losses for when

the second compressor is integrated into the facility and the turbine inlet flow rate is doubled.

A venturi flow meter was installed between flanges on the main flow supply piping as

shown in Figure 2.6 to provide the turbine inlet mass flow rate measurement. Early measurements

showed high unsteadiness in the differential pressure measurement on the venturi. The unsteadiness

was likely due to the area expansion from the 0.2 m diameter pipe to the 0.3 m diameter pipe as

Figure 2.7. Facility cooling equipment: (a) outdoor heat exchanger for compressor cooling

system, (b) chiller for cooling turbine secondary air, (c) pump skid for compressor cooling

system, (d-e) piping for compresssor cooling system.

22

well as the wake shedding from the control valve, so a flow straightener was installed downstream

of the valve and area expansion to condition the flow for the venturi flow meter.

Upstream of the test section a 1.2 m (4 ft) diameter settling chamber was installed as shown

in Figure 2.6 and Figure 2.8. The settling chamber housed several baffle plates and screens, shown

in Figure 2.8, to break up the incoming air jet from the piping and stagnate the flow before entering

the turbine test section. The settling chamber was designed with an inner diameter support structure

that was cantilevered with five equally-spaced struts shown in Figure 2.8. The struts had an

aerodynamic NACA0015 cross section to minimize wake losses entering the turbine test section.

Several internal tubes for routing cooling flows and instrumentation were designed into the struts.

The inner diameter support structure was designed to mate with the test section and help support

the inner diameter of the turbine test section.

An annular downstream settling chamber, shown in Figure 2.6, was designed to attach

downstream of the test section and direct the flow to the facility exhaust piping. Maintaining

circumferential uniformity in the test section was a high priority so the design of the settling

chamber was important. The downstream settling chamber, also shown in Figure 2.6 and in Figure

2.8, included two annular plenums separated by a radial baffle plate. The turbine exit flow was

designed to enter the inner annular plenum, and then pass through the radial baffle plate to the outer

annular plenum to reduce the effects of the potential field in the exhaust piping from affecting flow

upstream into the turbine. The flow then passed from the annular plenum to a pipe that connected

to the facility exhaust piping.

Figure 2.8. Cross-section of turbine test section (in dashed outline) and adjoining parts

(generic turbine and select components): (a) radial baffle plate for bypass piping, (b) baffle

plates in settling chamber, (c) upstream settling chamber, (d) test section center body

supported by struts, (e) support structure on linear rails, (f) turbine, (g) bearing block, and

(h) annular downstream settling chamber.

23

The pressure at the exit of the turbine test section was lower than at the inlet, which resulted

in a higher mean velocity in the piping. The piping diameter was thus increased to 0.4 m (16 in) to

minimize pressure losses in the exhaust piping. An emergency fast-closing valve was installed just

downstream of the settling chamber, as can be seen in Figure 2.6, for safety reasons to be discussed

at the end of this chapter. A second venturi flow meter, shown in Figure 2.6, was installed in the

exhaust piping to provide the turbine exit mass flow rate measurement. A flow control valve was

installed downstream of the venturi to control the turbine exit pressure, as shown in Figure 2.6, and

thus the pressure ratio across the turbine. Initial facility commissioning used a pneumatic actuator

on this flow control valve, but, as previously discussed, the valve exhibited unacceptable drift and

unsteadiness so an electrical actuator was installed on the valve resulting in a much improved,

stable operation as shown in Figure 2.5. The exhaust piping then exited the lab vertically and passed

through an exhaust silencer shown in Figure 2.3 to reduce the noise heard by the neighboring

residential areas.

A section of piping just upstream of the settling chamber bypassed the turbine test section

to the facility exhaust. Flow entered the bypass piping through a radial baffle plate around the pipe,

as shown in Figure 2.8, allowing for minimal disturbance to the main flow as it entered the settling

chamber. On the bypass piping there was a flow control valve and an emergency fast-opening valve

in parallel with each other. The bypass piping was used for two main purposes: (1) to allow for

commissioning of the compressor and its subsystems independently of the test section, and (2) to

allow the facility user to independently set the turbine inlet pressure and the mass flow rate during

research testing without stalling or surging the facility compressor. The flow control valve was

used to control the flow through the bypass piping. Similar to the facility exit pressure valve, the

bypass valve had a pneumatic actuator for initial facility commissioning. The actuator also

displayed unacceptable levels of drift, which led to the installation of an electrical actuator giving

a more stable operation. The emergency valve will be discussed at the end of this chapter.

An important feature of the facility was the ability to simulate secondary air flows in the

turbine. A secondary air supply was installed to divert up to 0.9 kg/s (2 lbm/s) of the compressor

discharge air to various locations in the turbine test section as shown in Figure 2.9. A single pass

shell and tube heat exchanger was used to cool the air from 395 K (710° R) down to 280 K (500°

R). A chiller unit rated to 200 kW (700,000 BTU/hr) was installed behind the back wall of the lab

to provide coolant to the heat exchanger. To prevent condensate from entering the turbine test

section a centrifugal moisture separator and a high flow filter were installed downstream of the heat

exchanger.

24

The secondary air supply split into four independently controlled and metered cooling

supplies as shown in Figure 2.9. Turbine flow meters and electrically-actuated flow control valves

were used to measure the volumetric flow rate and control the secondary flow rates for each supply.

Each of the secondary flow supplies entered a plenum with a splash plate before splitting into

multiple supply hoses to reduce the pressure losses. Two of the secondary air supplies—the purge

flow and the TOBI flow—entered the turbine through the five struts in the upstream settling

chamber and passed to the upstream side of the turbine disk through the inner diameter casings.

The flow through the supply hoses then recombined and passed through three baffle plates to ensure

uniform flow entered the turbine. The two other secondary air supplies entered the turbine test

section through the outer casing to the second vane. For the experiments presented in this research

the secondary air supplies to the second vane were not used.

A water brake dynamometer maintained the speed and load on the rotor. The dynamometer

also dissipated the generated power by flowing water through two rotating perforated disks. The

water pressure in the dynamometer and the water flow rate determined the rotational speed and

required braking torque. The dynamometer was rated to a maximum rotational speed of 11,000

RPM, a power of 895 kW (1200 hp), and a maximum torque of 1200 N-m (900 ft-lbf). The

dynamometer was equipped with a torque transducer with a range suitable for the turbine operating

conditions to reduce torque measurement uncertainty.

The operation of the water brake dynamometer required significant infrastructure and

equipment. In addition to pumps, tanks, bypass loops, back flow regulators, and numerous manual

valves there were strict requirements on the dynamometer water purity. A water treatment system

Figure 2.9. Schematic of the secondary air supply for the test turbine with four

independent sources, with the purge and TOBI flows supplied to the inner diameter of the

turbine at the first vanes, and the two second vane flows supplied to the outer diameter of

the turbine at the second vanes.

Heat Exchanger

Filter

Turbine Flow

Meters

Plenums Delivery Hoses to Turbine

Flow Control Valves

P/T Sensors

Test Turbine

Purge Flow

TOBI Flow

2V Flow A

2V Flow B

Facility Compressor

1V2VB

25

was implemented that included water softening, rust monitoring, and a reverse osmosis system to

protect the equipment from corrosion and ensure stable performance.

Figure 2.10 shows a schematic of the overall water brake dynamometer system, and Figure

2.11 shows a photo of the some of the indoor portion of the water system. The dynamometer is

shown in Figure 2.11. The water was supplied to the dynamometer from a supply pipe with a

pressurized inline accumulator tank shown in Figure 2.10. In case of a power or pump failure the

accumulator tank was designed to supply sufficient water to the dynamometer to perform a

controlled shutdown. After exiting the dynamometer, the heated water was gravity fed to an

underground tank, or hot well shown in Figure 2.10. The accumulated water in the hot well was

pumped to a cooling tower, where the hot water was injected in the top of the cooling tower and

ambient air was forced upward through the water by a large blower allowing the water to cool to

ambient conditions. The cold water then flowed into a cold well before being pumped back into the

facility to the accumulator tank and the water brake dynamometer.

Figure 2.10. Schematic of the water brake dynamometer system, showing the water flow loop and

the hydraulic oil flow loop for the dynamometer control valves.

Inlet Valve

Exit ValveCold well

Hot well

Evaporative Cooling Tower

Accumulator Tank

Dyno Turbine

Tower Pump

Dyno Pump

Hydraulic Oil Pump

Forced Air

Water Treatment

26

A controller provided by the dynamometer vendor set and maintained the dynamometer

operating conditions. At the inlet and exit of the dynamometer two water control valves shown in

Figure 2.10 and Figure 2.11 were installed and controlled by the dynamometer controller. The two

valves were hydraulically-actuated, so a hydraulic oil pump, shown in Figure 2.10 and Figure 2.11,

was used to provide the high-pressure oil to the valves. As shown in Figure 2.10, a small portion

of the water in the dynamometer water system was diverted to cool the hydraulic oil pump.

Installing the dynamometer equipment required a large hole to be excavated behind the

laboratory as seen in Figure 2.12. The largest equipment, including the hot well, the cold well, and

a pump room, were installed in the large hole. Since glycol could not be added to the water due to

Figure 2.11. Photo of the water brake dynamometer system: (a) dynamometer, (b) water

accumulator tank, (c) hydraulic oil pump, (d) dyno water inlet valve, (e) dyno water exit valve.

27

the potential for foaming in the dynamometer, the piping to and from each of the tanks needed to

be below the local frost line to ensure the system would never freeze.

The dynamometer vendor assisted in the commissioning, calibration, and tuning of the

water brake [45]. After a tuning process the water brake dynamometer was shown to hold the

rotational speed of the turbine rotor constant within a standard deviation of ±0.2% of the mean

speed for a full day of testing.

2.4 Test Section

The turbine test section consisted of over 100 components, both stationary and rotating.

This section describes the test section design, including the turbine, the magnetic bearing system,

and the rotor dynamics calculations. A cross-section of the test section and the adjoining parts is

given in Figure 2.8. Note that most of the turbine test section parts were drawn generically to protect

the intellectual property of the sponsor.

Figure 2.12. The water brake dynamometer system required the excavation of a large hole

to install the hot and cold wells, as well as the pump vault and the underground piping.

28

The test section was composed of outer casings, a center body support structure, and inner

and outer flow path components. The facility and the components interfacing with the turbine were

designed by the START team. The test turbine was designed by Pratt & Whitney and Belcan [46].

A thick outer casing around the turbine section was designed to contain a liberated blade and a

burst disk. Upstream of the disk the rig center body shown in Figure 2.8 was supported by the struts

through the settling chamber as well as ten thin struts upstream of the first vane (not shown). The

portion of the rig center body aft of the disk was supported by an 18 cm (7 in) thick bearing block

with several components that housed the rotor as shown in Figure 2.8. The parts composing the test

section center body were line drilled to ensure concentricity.

The upstream portion of the test section was securely mounted to a support structure

mounted on linear rails shown in Figure 2.8. A clamshell design shown in Figure 2.6 was used for

the casing that attached the upstream settling chamber to the test section. The center body also had

a similar design that allowed radial removal of the components. By removing these clamshell

components the turbine test section could be axially split and moved forward on the rails allowing

0.5 m (18 in) of axial clearance to access the turbine components including the rotor and the test

section instrumentation.

The upstream portion of the flow path to the turbine included a 20:1 area contraction to

ensure uniform flow at the turbine inlet. Two axial locations for turbulence grids were designed

into the test section, although the experiments presented here used no turbulence grid. One grid

location was one first vane axial chord upstream of the first vane leading edge, and the other was

located upstream of the area contraction.

The design of the turbine used for this research was a 1.5 stage (vane-blade-vane) turbine

with partial span airfoils as shown in Figure 2.13. Non-dimensional radii for select locations in the

turbine are provided in Figure 2.14. The use of partial span airfoils was justified for experiments

focusing inboard of the platform, such as rim seal ingestion measurements like this research, as

shown by several previous studies [26,28,47]. This research presents effectiveness measurements

for the front (vane-blade) cavity, but not the aft (blade-vane) cavity. A modern turbine design was

implemented for the airfoils, the rim seal and cavity geometries, and the secondary air delivery

methods. A portion of the research presented in this dissertation was for a half stage (vane only)

geometry with the same rim seal and rim cavity geometry as the 1.5 stage turbine, which will be

discussed in more detail in Chapters 3 and 4.

Both the vanes and blades were uncooled for this particular test turbine. The blades were

solid single crystal castings from a nickel alloy. Both the first and second vanes were manufactured

additively using a direct metal laser sintering process from Inconel 718. A distinct advantage of

29

additively manufactured vanes was the inclusion of integrated pressure taps and routing tubes for

instrumentation. More details on the test turbine and instrumentation are given in Section 2.5 and

in later chapters in which the research findings are presented.

Figure 2.13. Cross section of turbine with particular regions and flows called out: (a) first

vane plenum, (b) front rim seal, (c) front rim cavity, (d) front wheel-space, (e) purge flow,

(f) TOBI flow, and (g) aft rim cavity.

Main gas path(MGP)

First Vane(1V)

Second Vane(2V)

Blade(B)

ae

d

c

b

g

f

30

The turbine rotor was installed in a bearing tube, shown in Figure 2.15, that was mounted

within the bearing block. Magnetic bearings were used as they had certain advantages over

conventional bearings, including reduced bearing friction and losses, better control of bearing

stiffness and damping, and thus better control of rotor dynamics. The turbine rotor was overhung

with two radial bearings downstream of the turbine disk. Two radial magnetic bearings, shown in

Figure 2.15, were capable of supporting the 120 kg (250 lbm) rotor, and the bearings were capable

of maintaining control of the rotor up to a speed of 20,000 RPM.

Figure 2.14. Cross section of turbine with select non-dimensional radii.

r/b

=1.1

r/b

=0.9

1

r/b

=0.9

3 r/b

=0.9

5

b

r/b

=0.6

8

r/b

=0.9

6

r/b

=0.8

0

31

The magnetic bearings held the rotor in a levitated state through an electromagnetic field.

Passive rotors were fixed on the shaft as shown in Figure 2.15, and the stators, which were actively

controlled by an external control unit shown in Figure 2.6, sustained the necessary electromagnetic

field to levitate the shaft. Auxiliary radial bearings were designed to catch the rotor in case the

magnetic bearings failed. Power failures were to be mitigated by an uninterrupted power supply for

the magnetic bearing controller. A radial clearance of ±0.18 mm (±0.007 in) was designed between

the levitated rotor centerline and the auxiliary bearings. When the magnetic bearings were inactive

the turbine rotor rested on the auxiliary bearings. Although not used for the research presented here

the radial clearance of the auxiliary bearings allowed for a radial shift of the rotor centerline or

testing with a controlled whirl orbit within a ±0.08 mm (±0.003 in) radial clearance.

Figure 2.15. Cross-section of rotor assembly: (a) bearing tube, (b) shaft, (c) radial bearing

stator, (d) radial bearing rotor, (e) thrust bearing rotor, (f) thrust piston supply, and (g)

thrust piston exhaust.

32

A thrust bearing, shown in Figure 2.15, was used to provide counter thrust and maintain

the nominal axial position of the rotor. The axial thrust force from the turbine rotor exceeded the

6.7 kN (1500 lbf) capability of the thrust bearing so a two-stage pneumatic thrust piston system

was designed to provide supplementary counter thrust. As shown in Figure 2.15 high pressure air

was supplied to each of the thrust pistons to provide up to 8.9 kN (2000 lbf) of additional counter

thrust. The additional counter thrust provided by the thrust piston system was designed to be

capable of future testing of full span airfoils, which could result in an additional 2.2 kN (500 lbf)

of axial thrust on the turbine rotor. The thrust bearing had an axial clearance of ±0.25 mm (±0.010

in) and allowed an axial movement of ±0.13 mm (±0.005 in). The leakage across each thrust piston

was mitigated by a stationary brush seal, and the leakage between the thrust piston stages was

mitigated by a multi-stage labyrinth seal. Several high flow regulators were used for the thrust

piston air supply as shown in Figure 2.16a. The thrust piston air supplies and exhaust tubes were

routed through the downstream settling chamber through the bearing block as shown in Figure

2.16b. As was mentioned, at the time the experiments for this dissertation were performed, the

thrust piston was underpowered, making it necessary to reduce the turbine inlet pressure. The thrust

piston operability has since been resolved, and the thrust piston is presently able to supply the full

designed value of 11 kN (2500 lbf) of counter thrust at the time of the writing of this dissertation.

The magnetic bearing parameters, including bearing stiffness and damping, could be tuned

to provide stable operation at any speed from 0 - 20,000 RPM. The magnetic bearing parameters

were tuned for optimal operation by the vendor at several operational speeds during the rotor

Figure 2.16. (a) thrust piston air supply stand, and (b) air supply hoses to thrust piston

through the downstream settling chamber.

33

commissioning. The shaft orbits at both radial bearings were tracked with sensors in the magnetic

bearing system and displayed in real time to the magnetic bearing user interface. During normal

operation the shaft centerline was maintained within ±2.5 µm (±0.0001 in) of the true centerline by

the magnetic bearings, although at lower speeds the centerline was observed to have a slightly

higher orbit at ±10 µm (±0.0004 in).

Magnetic bearings were a non-traditional choice for bearings, so detailed analyses was

required to ensure safe and stable operation of the turbine. An in-house rotor dynamics analysis

was performed using XLRotor [48,49]. The same rotor dynamics analysis was independently

checked by both the industry sponsor and the magnetic bearing vendor [50,51]. A generic cross

section of the turbine rotor is shown in Figure 2.17a, and the rotor dynamic model is shown in

Figure 2.17b. The end of the shaft connected to a flexible coupling that connected to the

dynamometer that was assumed to have infinite stiffness. The bearing support structure was also

assumed to have infinite stiffness. These assumptions were sufficient because the magnetic bearing

stiffness was much lower than the stiffness of the support structure.

Figure 2.18 shows the Campbell [52,53], or interference, diagram of the rotor dynamic

model. The natural frequencies are plotted against the turbine rotational speed. Two rigid body

modes were shown to exist at low speeds, and the bending modes were shown to exist at higher

speeds. The turbine was designed to operate between these modes. The black dashed line represents

the synchronous line, which crosses the first bending mode at a speed of 14,800 RPM, which was

the critical speed for the design of the turbine. Commissioning and research testing showed that the

turbine was able to operate safely at the design speed with ample margin.

Figure 2.17. (a) Generic cross-section of turbine rotor and bearing structure; (b) XLrotor

rotor dynamic model of turbine rotor.

34

2.5 Instrumentation

A major part of the work for this dissertation was designing, specifying, and installing the

instrumentation for both the facility and the turbine. The facility instrumentation was composed of

venturi flow meters, pressure transducers, and resistance temperature detectors. The turbine

instrumentation was composed of pressure probes, pressure taps, a pressure scanner and calibration

system, thermocouples, a gas analyzer, mass flow controllers, and turbine flow meters. A

description of the instrumentation for both the facility and turbine will be given in this section. The

overall uncertainty for key measurements, as computed per the method of Figliola and Beasley

[54], is also provided at the end of this section. A design-stage uncertainty analysis was performed

to select the instrumentation before testing, and this analysis is contained in the Appendix.

An overview of the facility instrumentation and the layout of some of the turbine

instrumentation is given in Figure 2.19. The facility instrumentation was installed to provide

reliable monitoring of the facility operation, and was therefore designed to be accurate and robust.

Table 2.2 provides the facility instrumentation ranges and accuracies. The measurements from the

facility instrumentation were sent to the facility programmable logic controller (PLC—to be

discussed in Section 2.6), and were sampled at 100 Hz to allow for quick detection of errors and

initiate emergency shutdowns.

Figure 2.18. Campbell, or interference, diagram of turbine rotor showing rotor dynamic

modes of test turbine.

Bending modes

Rigid body modes

Design speed

Nat

ura

l Fre

qu

en

cy [

1/m

in]

Rotor speed [RPM]

35

Venturi flow meters were used to measure the turbine inlet and exit flow rates as shown in

Figure 2.19. The flow meters were purchased with a factory calibration rated to an accuracy of

±0.75% of reading, but a design-stage uncertainty analysis showed that an accurate mass flow rate

measurement was critical to testing. Both the upstream and downstream venturi flow meters and

the connected pipes were thus sent to a NIST certified calibration facility [55] to increase the

accuracy of the calibration to ±0.34% to ±0.47% of the reading, as shown in Table 2.2. The venturi

flow meters were both installed with the recommended ten length-to-diameters of straight pipe

upstream of the flow meters and five length-to-diameters of straight pipe downstream.

Three kinds of pressure transducers were used for the facility measurements. The ranges

and accuracies of the pressure transducers are given in Table 2.2. A barometric pressure transducer

provided the lab barometric pressure. Most of the transducers were gage pressure transducers, and

combined with the barometric pressure measurement, they provided the absolute pressure at various

locations in the facility. The locations of the gage pressure transducers are shown in Figure 2.19.

Differential pressure transducers were installed on the venturi flow meters to allow for computation

of the turbine inlet and exit mass flow rates. Pressure taps with a diameter of 1.0 mm (0.04 in) were

drilled into the facility piping, and the pressure tubing was routed to the transducers. Since these

measurements were critical to the operation of the facility each pressure measurement location had

redundant pressure transducers to ensure safe and reliable operation.

Figure 2.19. Schematic showing the approximate locations of the facility and turbine

instrumentation.

Venturi

Building Back Wall

Roof Exhaust

RoofIntake

X

P

XP

T

P

P P

PT

T

T

Facility Instrumentation

Gage pressures

Differential pressures

Temperatures

Venturi flow meters

X

T

PPLC

T

T

Research DAQ

Wire routing

path

Secondary air supply turbine flow meters

Gas analyzer

Pressure scanner & calibrator

36

The facility temperature measurements were provided by robust industrial probes equipped

with resistance temperature detectors (RTDs). Two RTDs, offset by 180°, were installed at each

measuring location to provide redundant measurements. The RTD probes had a diameter of 6.4

mm (0.25 in) and penetrated approximately 75 mm (3 in) into the flow. The RTDs were installed

at the locations shown in Figure 2.19 to provide temperatures at the compressor inlet and exit, the

test section inlet and exit, and upstream both venturi flow meters. The RTDs for the venturi flow

meters were located five length-to-diameters upstream of the venturis to minimize flow

disturbances and ensure accurate flow rate measurements.

Table 2.2. Summary of Facility Instrumentation

Parameter Range Accuracy

Barometer 26 to 32 mm Hg

(12.8 to 15.7 psia) ±0.05% Full Scale

Gage pressure transducers 690 kPa

(100 psig) ±0.05% Full Scale

Differential pressure transducers 35 kPa (5 psi diff)

17 kPa (2.5 psi diff) ±0.05% Full Scale

Class A resistance temperature detectors -200 to 600°C

(-238 to 1112°F)

±0.15°C at 0°C

(±0.3°F)

Venturi flow meters 2.3 to 6.0 kg/s

(5.0 to 13 lbm/s)

±0.35% to ±0.47%

of reading [55]

The turbine was heavily instrumented to provide a variety of measurements. Figure 2.20

shows the turbine instrumentation locations on a cross-section of the 1.5 stage turbine, including

the pressure probes, static pressure taps, and thermocouples. The approximate locations of the

pressure scanner and calibration system, gas analyzer, and the turbine flow meters for the secondary

air supply are shown in Figure 2.19.

Kiel pressure probes were used to measure the turbine inlet pressure. The probes were

inserted one axial chord upstream of the first vane leading edge to approximately mid span, as

shown in Figure 2.20, and were oriented in the axial direction (note that the kiel probes were

insensitive to flow direction to ±40° yaw and pitch). The stem diameter of the probe was 3.2 mm

(0.13 in), the kiel head diameter was 1.6 mm (0.063 in), and the sensing tube diameter inside the

kiel head was 0.5 mm (0.02 in).

37

The turbine contained nearly 250 static pressure taps at a variety of locations shown in

Figure 2.20. The first and second vanes were made through additive manufacturing (AM), and

nearly 220 of those pressure taps were integrated into the design of the vanes through AM, as

indicated in Figure 2.20. Many of them were not possible without the use of AM, as traditional

manufacturing methods would have been too difficult or too costly to execute.

There were several pressure taps at three spanwise locations (10%, 50%, and 90% span)

on both the first and second vane airfoil surfaces. Additional taps at several circumferential

locations near the front rim seal were integrated through additive manufacturing. Pressure taps at

multiple circumferential locations in the front rim cavity were manufactured traditionally. Each of

those groups of pressure taps were duplicated at two circumferential locations to ensure

measurement redundancy and to check for circumferential uniformity of the test section. Pressure

taps in the first vane plenum, the front wheel-space, the TOBI supply, the aft cavity, and at the

second vane exit were located at four circumferential locations equally-spaced around the annulus.

The pressure probes and static pressure taps in the turbine test section were all routed to a

pressure scanner with 112 channels and a variety of transducer ranges. As indicated in Table 2.3,

the 200 and 350 kPa (30 and 50 psi) transducers provided gage pressure measurements, and the 35

and 200 kPa (5 and 15 psi) transducers provided differential pressure measurements, with all

transducers having an accuracy of ±0.05% FS. The pressure scanner used silicon piezoresistive

transducers for each channel, and two 16-bit analog-to-digital converters electronically scanned the

transducers at up to 625 Hz. An onboard processor performed real time temperature compensation

from 0°C to 55°C, as well as averaging and conversion to engineering units.

A pressure calibration system was set up to perform calibration checks and calibrations of

the pressure scanner system. The pressure calibrator had an accuracy of ±0.01% full scale with a

390 kPa (56 psi) range and a 120 kPa (17.5 psi) range. The calibration system was automated to

perform multi-point calibrations for all of the transducers in a single batch ensuring pressure

measurement accuracy. The pressure scanner calibrations were initially checked before each test,

and once the stability of the system was established the calibrations were checked at least monthly.

To minimize the length of the pressure tubing the pressure scanner and calibrator were located near

the test section as shown in Figure 2.19.

The temperature measurements in the turbine test section were provided by thermocouples.

Type T thermocouples were used to minimize the measurement uncertainty to ±0.5°C (±0.9°F) as

shown in Table 2.3. The thermocouples in the vane plenum, TOBI supply, and front wheel-space

shown in Figure 2.20 were small (AWG 30) to minimize any disturbances to the flow. Additionally,

fifteen kiel temperature probes were located on the leading edges of the first vanes. These kiel

38

probes had an outer diameter of 2 mm (0.079 in), two vent holes with a diameter of 0.8 mm (0.03

in), and a type T thermocouple with a bead diameter of approximately 0.4 mm (0.015 in). These

fifteen temperature probes were spaced around the annulus on six different vanes, with three

measurements each at five spanwise locations (10%, 30%, 50%, 70%, and 90%) as shown in Figure

2.20. The thermocouple wires passed through sealed instrumentation pass-through fittings on the

test section outer casing. Outside the test section the thermocouples were connected to larger

diameter thermocouple extension wire (AWG 20), and were routed to the research data acquisition

system in the control room, as shown in Figure 2.20. The extension wire was composed of twisted

and shielded paired wires to minimize the influence of electrical noise.

The primary measurement reported in this dissertation is concentration effectiveness, and

Chapter 3 will discuss this measurement technique in great detail. The technique involved using

CO2 as a tracer gas in the secondary air supplies. Sampling the flow through pressure taps in the

turbine yielded a sealing effectiveness based on the gas concentration measurements. A gas

analyzer provided the CO2 concentration measurements. As given in Table 2.3 the accuracy of the

gas analyzer was ±1% of the full scale range. For the experiments presented in this dissertation the

Figure 2.20. Cross-section of 1.5 stage turbine, showing turbine instrumentation locations.

Total pressure probeStatic pressure tapTotal temperature

Instrumentation type

1V airfoil surface taps (through AM)Kiel temperature probesKiel pressure probesFront rim seal taps (through AM)Front rim cavity taps1V plenum TC’s and probesFront wheel-space TC’s and tapsTOBI supply TC’s and probes2V airfoil surface taps (through AM)Aft cavity taps

39

gas analyzer was calibrated with a calibration gas at 1,000 ppm and 10,000 ppm ranges. To reduce

instrument bias error in the concentration effectiveness low values of gas concentration were

measured with the 1,000 ppm range and higher values were measured with the 10,000 ppm range.

Mass flow controllers were also used to measure and control the mass flow rate at two

locations. One mass flow controller was used to maintain the CO2 supply flow rate constant, and

another was used to set and maintain the sampling flow rate for the gas concentration

measurements. The ranges and accuracies of these mass flow controllers is given in Table 2.3.

The volumetric flow rates of the secondary air flow supplies were measured using several

turbine flow meters. The flow meters were installed with the recommended 20 length-to-diameters

of straight pipe upstream of the flow meters and ten length-to-diameters of straight pipe

downstream. Pressure and temperature measurements located 20 length-to-diameters upstream of

the flow meter allowed for calculation of the air density in each supply, allowing the mass flow rate

to be calculated within a total uncertainty of ±1.2% of each measurement, as shown in Table 2.4.

As indicated in Table 2.3 a wide selection of turbine flow meters was available at several ranges.

The turbine flow meters had an accuracy of 1% of the reading and had a turndown ratio of at least

10:1, meaning the turbine flow meter with a range of 570 LPM (20 ACFM) maintained the rated

1% accuracy down to a flow of 57 LPM (2 ACFM). The higher range flow meters had turndown

ratios of 15:1 to 18:1 allowing for a wide range of flows to be measured with each flow meter. The

flow meters were often run in series to check their calibrations and the measurements agreed within

±1%.

40

Table 2.3. Summary of Turbine Instrumentation

Parameter Range Accuracy

Pressure scanner

(112 total channels)

350 kPa (50 psi gage)

200 kPa (30 psi gage)

100 kPa (15 psi diff)

35 kPa (5 psi diff)

±0.05% FS

Pressure calibrator 120 kPa (17.5 psi gage)

390 kPa (56 psi gage) ±0.01% FS

Type T thermocouples -250 to 350°C

(-328 to 662°F)

±0.5°C

(±0.9°F)

Gas analyzer Range 1: 0 to 1,000 ppm

Range 2: 0 to 10,000 ppm

±1.0% of reading

Mass flow controllers

CO2 supply: 0-100 SLPM

(3.5 SCFM)

Sampling: 0-5 SLPM

(0.2 SCFM)

±(0.2% FS + 0.8%

reading)

Turbine flow meters

570 LPM (20 ACFM)

1,700 LPM (60 ACFM)

3,700 LPM (130 ACFM)

6,400 LPM (225 ACFM)

13,000 LPM (450 ACFM)

±1.0% of reading

(turndown ratio of

>10:1)

The total uncertainty for the main facility and turbine measurements was computed

according to the method of Figliola and Beasley [54]. The total uncertainty, reported in Table 2.4,

was calculated to include both the instrument bias uncertainty and the precision uncertainty. Since

the measurements were time-averaged, large samples were used to minimize the precision

uncertainty.

41

Table 2.4. Uncertainty in Facility and Turbine Measurements

Parameter Total Uncertainty*

Main gas path flow rate, �̇�𝑖𝑛, �̇�𝑒𝑥

(venturi flow meters) ±0.4% to ±0.6%

Shaft rotational speed, Ω

(dynamometer) ±0.2%

Turbine inlet pressure, 𝑝𝑡,𝑖𝑛 ±0.1%

Facility temperatures, 𝑇 ±0.27°C (±0.48°F)

Turbine temperatures, 𝑇 ±0.55°C (±0.95°F)

1.5 stage pressure ratio, PR 𝑝𝑡,𝑖𝑛/𝑝𝑒𝑥 ±0.6%

Secondary flow rate, �̇�𝑗

(turbine flow meters) ±1.2%

Concentration effectiveness, 𝜀𝑐 ±0.015

*Including both bias and precision uncertainty per the method presented by Figliola and Beasley [54].

2.6 Control and Safety Precautions

The control of the facility was performed through a programmable logic controller (PLC)

as shown schematically in Figure 2.21. The programming of the facility control and safety logic

was executed by the Applied Research Laboratory at Penn State [56]. Several algorithms designed

to maintain safe operation of the facility were implemented in the PLC. Thus the PLC was a

standalone system, separate from the research data acquisition system, designed to operate

continuously at 100 Hz by checking the health status of the facility and its various components as

will be discussed in this section. The PLC received inputs, shown as green arrows in Figure 2.21,

and sent control commands, shown as orange arrows in Figure 2.21. The research data acquisition

system also received inputs, shown as purple arrows in Figure 2.21, and also sent control

commands. Some of the facility control commands could be sent by either the PLC or the research

data acquisition system, and these commands are shown by the blue arrows in Figure 2.21.

42

The main user interface with the PLC was a touchscreen panel in the control room that

allowed the user to see all the facility measurements, send commands to the control system, and

see feedback on the controls. The touch screen panel allowed the user to change the control valve

positions, the valve speed, the thrust piston control mode, and the thrust piston set point, among

other parameters. Most of the control logic was programmed into the PLC separately and was

locked down, and thus was not available to the user during testing to ensure safe and stable

operation of the facility.

The control of the facility could be switched from the touchscreen to a separate system

controlled by a LabVIEW Virtual Instrument (VI) on the research data acquisition computer.

Control loops could be implemented in the VI, but manual operation of the main facility control

valves within LabVIEW allowed for very steady and repeatable operation of the facility. Several

Boolean indicators (green = good, red = bad) were implemented in the VI allowing the user a quick

check on the health of the facility during operation. Select PLC input parameters (green items in

Figure 2.21) were checked against specified limits, such as the dynamometer pump status (on or

off), the dynamometer water exit temperature (<135°F), rotational speed (<11,000 RPM), axial

thrust load, and others. Some indicators were linked to warnings, which alerted the user to a

Figure 2.21. Schematic of the facility programmable logic controller (PLC).

Motor

Turbine Cooling System

Control System

Data AcquisitionSystem

Compressor Cooling System

PLC ControlCommand

Test Turbine

Compressor

(6)

(3)(2)

(5)

(7)

CO2

(1)

(4)

Dyno

PLC SystemPLC InputPLC Control CommandDAQ InputPLC or DAQ Control

Press/TempVibrationSeals/OilMotor CurrentHealth Status

Speed SensorsShaft PositionThrust LoadHealth Status

TorqueRot. SpeedWater FlowWater P/TPump Status

Flow Control

Flow Control

Flow Control

Emergency Relief

Mag BearingsThrust Piston

Flow Control

Emergency Shut-off

43

potential issue, while some were linked to hard alarms, which immediately initiated an emergency

shutdown procedure.

Alarms were linked to healthy signals from the magnetic bearing system and the

dynamometer system, as well as various overspeed alarms. If an alarm was activated the two fast-

acting valves actuated to prevent further air flow to the turbine: a fast-closing valve downstream of

the test section (shown as valve 7 in Figure 2.21) closed in approximately 0.5 seconds to prevent

further flow to the turbine, and a fast-opening valve on the bypass piping (shown as valve 5 in

Figure 2.21) opened in approximately 0.5 seconds to divert the high pressure facility air around the

test section. The magnetic bearings were set to maintain the shaft in a levitated state during the

emergency shutdown procedure to prevent the shaft from “dropping” on the auxiliary bearings and

causing premature wear. The dynamometer was set to maintain the braking torque to prevent a

freewheeling turbine rotor. The facility control valves (valves 3, 4, and 7 in Figure 2.21), although

much slower than the fast-acting valves, were also set to isolate the test section, and the facility

compressor was set to immediately unload and shutdown. The emergency shutdown logic and the

execution of the fast valves were thoroughly tested and were well-suited for safe operation of the

facility.

2.7 Summary

The facility was designed and built to extend the operating range of continuous flow turbine

research facilities to engine-relevant operating conditions for an engine-realistic turbine. The

facility equipment and infrastructure were designed, installed, and integrated to give a steady and

safe operation to perform experiments for the research presented here as well as future turbine

experiments. The instrumentation for both the facility and the turbine were selected to provide

accurate measurements. In the following chapters, first-of-their-kind results will be presented

regarding turbine secondary air systems, rim seal ingestion, and sealing effectiveness at engine-

relevant conditions.

44

Chapter 3

Using a Tracer Gas to Quantify Sealing Effectiveness

for Engine Realistic Rim Seals1

Abstract

As overall pressure ratios increase in gas turbine engines, both the main gas path and

cooling temperatures increase leading to component durability concerns. At the same time effective

use of the secondary air for both cooling and sealing becomes increasingly important in terms of

engine efficiency. To fully optimize these competing requirements, experiments at engine-relevant

conditions are required to validate new designs and computational tools. A test turbine has been

commissioned in the Steady Thermal Aero Research Turbine (START) lab. The test turbine was

designed to be a 1.5 stage turbine operating under continuous flow simulating engine-relevant

conditions including Reynolds and Mach numbers with hardware true to engine scale. The first

phase of research conducted using the test turbine, which was configured for a half-stage (vane

only), was to study hot gas ingestion through turbine rim seals.

This paper presents a series of facility benchmarks as well as validation experiments

conducted to study ingestion using a tracer gas to quantify the performance of rim seals and purge

flows. Sensitivity studies included concentration levels and sampling flow rates in flow regimes

that ranged from stagnant to compressible depending upon the area of interest. The sensitivity

studies included a range of purge and leakage flow conditions for several locations in the rim seal

and cavity areas. Results indicate reasonable sampling methods were used to achieve isokinetic

sampling conditions.

3.1 Introduction

Secondary air bled from the compressor is required to cool components in the hot section

of a gas turbine engine. These components are exposed to temperatures that can degrade the

components and lead to durability concerns. Frequent maintenance is costly and undesirable, so

cooling air must be provided to the hot section of the engine. Additionally, the cavities between the

1 Clark, K., Barringer, M., Thole, K., Clum, C., Hiester, P., Memory, C., and Robak, C., 2016, “Using a Tracer Gas to

Quantify Sealing Effectiveness for Engine Realistic Rim Seals,” Proc. ASME Turbo Expo, GT2016-58095.

45

rotating and stationary components typically do not feature the advanced cooling technologies seen

in main gas path hardware; however, hot gas can be ingested into these cavities leading to high

temperatures. Some of the secondary air is provided to the cavities to seal or purge the ingested hot

gas. Although some secondary air is required for cooling and sealing in a gas turbine, the excessive

use of secondary air negatively impacts the efficiency of the engine.

As the bypass ratios of aircraft engines increase the fans get larger accelerating more air

and the core mass flow rate decreases. To drive the fan the overall pressure ratio increases, leading

to higher turbine inlet and cooling air temperatures. Sealing the turbine rim cavities is increasingly

important as engine core diameter decreases. Absolute clearances remain consistent, but relative

rim seal clearances increase resulting in significant ingestion. To purge the rotor-stator cavities

more secondary flow is required. It is thus critical to quantify the sealing effectiveness of rim seal

geometries used in decreasing engine core sizes.

Quantifying the sealing effectiveness in actual engines is challenging given the high

temperatures and the complexity of the flow. The high temperatures in engines can cause sensors

to fail so detailed and reliable measurements in the engine are difficult. Conductive heat transfer in

the metal turbine components also confounds effectiveness measurements. Frictional heating in the

cavities can increase the air and disk temperatures further complicating effectiveness

measurements.

Sealing effectiveness is the quantification of how well a rim seal prevents main gas path

air from being ingested into the cavities. For the stated reasons most sealing effectiveness studies

use a tracer gas to quantify sealing effectiveness. The use of a tracer gas allows for the direct

measurement of the equivalent of the adiabatic wall temperature in the rim seals and cavities

thereby negating conductive effects. This paper presents a method for using CO2 as a tracer gas to

quantify the sealing effectiveness and for tracing the secondary flows in a realistic rim seal and rim

cavity. Although using CO2 as a tracer gas is a common method for quantifying sealing

effectiveness, this paper is unique in that it explicitly describes both the method as well as provides

several validation experiments in the rim cavity and rim seal of an engine realistic vane at engine

relevant Reynolds and Mach numbers.

3.2 Review of Literature

A variety of methods exist in the literature for studying turbine rim seals, particularly in

quantifying hot gas ingestion. Qualitative measurements of ingestion, such as pressure and flow

46

visualization, have been used to provide evidence of ingestion behaviors. Pressure measurements,

for example, were used by Phadke and Owen [57] to determine if ingress or egress occurred across

a turbine rim seal, which agreed well with companion flow visualization that was used to determine

the flow rate at which no ingestion occurred.

Quantitative measurements of sealing effectiveness have been made, similar to this study,

by seeding the secondary flow with a tracer gas such as CO2. When the CO2 fraction in the

secondary flow is low, the assumptions of the heat and mass transfer analogy hold, namely that the

turbulent Schmidt number is approximately equal to the turbulent Prandtl number, as described by

Graber, et al. [58]. Gentilhomme [59] highlighted the use of CO2 over other gases due to its high

light absorption coefficient, leading to accurate concentration measurements by gas analyzers.

Many studies have since used CO2 to quantify the rim seal performance, but the fraction of

CO2 in the secondary supply has varied between researchers. The molecular weight of CO2 is higher

than that of air, but as stated by Graber, et al. [58] “the overwhelming dominance of turbulent

mixing over molecular diffusion in the rim seal” justifies the use a tracer gas with a slightly higher

molecular weight. The authors have found that the fraction of CO2 used for rim seal studies has

varied from 1% [60–62] to as high as 30% [63], with various fractions in between [64–66].

Although some of the studies cited provided a schematic of the sampling system, no mention was

made with regards to the sampling flow rates used.

A few investigations explicitly mentioned the importance of sampling method. In addition

to the qualitative measurements described previously, Phadke and Owen provided quantitative

measurements of ingestion through the use of a tracer gas [57]. The mainstream flow was seeded

with 100 ppm of nitrous oxide and gas samples were extracted from within the cavity to determine

the flow rate required to seal the cavity. The sampling flow rate was adjusted to achieve adequate

sampling conditions as validated through experiments outside of the rim seal test rig. Their

experiments showed that increasing the sampling flow rate up to four times the isokinetic value

kept concentration measurement errors within ±3% [57].

Using a tracer gas is a powerful technique to quantify sealing effectiveness and the

literature contains several other studies that use this method. These studies provide many insights

into the performance of rim seals and the physics governing ingestion. Most of the tracer gas

studies, however, do not explicitly describe the sampling methods and the validation of those

methods. This paper presents a description and validation of using CO2 as a tracer gas to quantify

sealing effectiveness in an engine realistic turbine rim seal. The sampling method is explicitly

described with an emphasis on achieving isokinetic sampling conditions, where isokinetic sampling

is defined as drawing gas samples at the same kinetic conditions, or velocity, as the flow from

47

which it is drawn. This paper also presents a discussion of the overall facility, test turbine, and

instrumentation used for these studies.

3.3 Test Facility and Test Turbine

The experiments described in this paper were performed in a facility described in full by

Barringer, et al. [67] with a brief description provided in this paper. Both the test facility and test

turbine are described in this section.

Facility Description

The facility used for these experiments was a high pressure, steady state, open loop flow

path capable of simulating engine relevant conditions using engine hardware for the test turbine.

An overview of the facility is shown in Figure 3.1. The air was directed through the flow path with

a 1.1 MW (1500 hp) industrial compressor capable of providing the conditions given in Table 3.1.

The research presented in this paper was for a test turbine with half span airfoils, which required

only one of the two available compressors. A second identical compressor will be used in future

tests with full-span airfoils. Because the focus of this paper was on the platform, rim seal, and rim

cavity, half-span airfoils were justified. The success of test turbines with short span airfoils for

studying rim seal performance has been demonstrated in several previous studies [60–62,65,66,68].

Table 3.1. START Facility Operating Conditions

Parameter Value

Compressor discharge pressure 480 kPa

Compressor discharge temperature 395 K

Compressor mass flow rate (single) 5.7 kg/s

Vane exit Mach number 0.7

Vane exit Reynolds number* 6X105

* based on vane exit velocity magnitude

48

The compressor discharge air was sent to a large settling chamber located upstream of the

turbine test section as shown in Figure 3.1. The settling chamber contained a series of baffles and

screens, which were followed by a 20:1 contraction to ensure uniform flow and thermal profiles

enter the test turbine. Upon exiting the turbine, deswirler vanes straightened the flow and

minimized losses before entering the downstream settling chamber. The turbine exit flow was

eventually sent to an exhaust silencer on the roof of the laboratory. The flow control valves and by-

pass loop were designed to allow steady control of the turbine inlet pressure, turbine exit pressure,

and the main gas path flow rate. In the event of an emergency shut-down, fast-acting valves divert

the air through the bypass loop.

The facility instrumentation included flow meters, several resistance temperature devices

and pressure transducers to monitor facility operating conditions from the control room, as

indicated in Figure 3.1. The main gas path flow rate was redundantly measured by large venturi

flow meters both upstream and downstream of the test section. A programmable logic control

(PLC) system was used to set the operating conditions for the test facility from the control room.

A portion of the compressor discharge air was directed to the secondary air system to

supply the purge and leakage flows in the turbine test section. The secondary air passed through a

heat exchanger that cools the compressor discharge air from 395 K (250°F) to 283 K (50°F).

Figure 3.1. START facility layout.

49

Downstream of the heat exchanger, a moisture separator and filter removed condensate from the

flow.

Test Turbine Description

The test turbine design was a 1.5 stage (vane-blade-vane) at true engine scale with half-

span airfoils, as was previously mentioned. For the validation experiments presented in this paper,

only the first vane was present. The rotor was replaced with a static rim seal and rim cavity aft of

the vanes, and the second vanes were replaced with inner and outer casings. Figure 3.2 shows a

cross-section of the half stage turbine. The first vane doublets were additively manufactured by a

metal laser sintering process of an Inconel alloy. Quantified inspections of the first vane doublets

showed good agreement between the vane throat gaps with less than ±0.3% deviation from the

design gap width.

Geometric parameters of the rim seal and cavity are defined in Figure 3.3. The colored

areas clearly define the areas that we refer to as the trench, rim seal, and rim cavity. The two leakage

flow paths, including the purge flow from discrete holes and the mate face gap flow from gaps

between the vane doublets, are also clearly indicated in Figure 3.3. Note that there was no

downstream blade in these studies. Future studies will include the full 1.5 turbine stages.

The vane purge and leakage flows were designed such that each could be independently

flowed and measured. The secondary air supply entered the test turbine at the inner diameter

through five flexible hoses. Each hose connected to an internal manifold that contained three

successive baffle plates to ensure the secondary air uniformly entered the first vane plenum. The

first vane plenum supplied the secondary air purge and leakage flows associated with the vane. As

designed in the engine, the mate face gap leakage flow for each vane doublet was simulated by

flow passing from the vane plenum to the rim seal through the mate face gap. The purge flow

provided air to the rim cavity through discrete holes that acted to seal the cavity. For the studies

reported in this paper, 150 purge holes uniformly distributed around the circumference were used.

Figure 3.4 shows the instrumentation in the turbine test section with an inset image showing

the circumferential arrangement. The instrumentation is distinguished by different symbols, and

groups of similar instrumentation are distinguished by colors. The two large circles on the inset

image indicate the main gas path annulus, and the colored symbols are located at the approximate

radial and circumferential locations corresponding to that instrumentation group. The open symbols

are not shown on the inset image as they are spread circumferentially throughout the test section.

50

Figure 3.2. First vane only test turbine cross-section.

Main gas path(MGP)

First Vane

Secondary air supply

hoses Baffle plates

First vane plenum

(1VP)

Figure 3.3. Test turbine nomenclature and geometric parameter definitions.

h/b=0.03

sc/b=0.01

s/b=0.03

s/b=0.04

First Vane

Plenum(1VP)

Rim cavity(RC)

Trench

Rim seal (RS)

Mate face gap leakage

Purge

b

r/b=0.88

Rotor side

Stator side

51

Kiel pressure probes were used to measure the turbine inlet total pressure, the first vane

plenum pressure, and the total pressure in the rim cavity and trench regions as indicated in Figure

3.4. To reduce aerodynamic blockage the kiel pressure probes were small, with a 1.6 mm diameter

kiel head and a 0.5 mm diameter sensing tube. Kiel temperature probes were integrated into the

first vane leading edge on six vanes around the annulus. Three vanes had temperature probes at

10%, 50% and 90% spans, and three vanes had temperature probes at 30% and 70% span for a total

of 15 total temperature measurements at the turbine inlet.

Since additive manufacturing was used for the vanes, instrumentation was directly

integrated into the vanes. Pressure taps were used for two purposes: static pressure and

concentration measurements. Six vanes spaced around the annulus were designed with static

pressure taps on the airfoil surfaces as indicated in Figure 3.4. The pressure taps on the airfoil

surface had a diameter of 0.5 mm, and were located around the airfoil on both the pressure and

suction surfaces for airfoil loading measurements. Two vanes had taps at 10% span, two vanes had

taps at 50% span, and two vanes had taps at 90% span for a total of nearly 80 integrated static

pressure taps on six different vanes spaced around the annulus. Flow and leakage tests ensured that

no leakages or cross-talk between pressure taps or internal tubes existed in the hardware. Two vane

doublets each had pressure taps concentrated in the vane hub trailing edge, the platform trailing

edge, and on the stator side of the rim seal with a diameter of 0.5 mm, which were again made

possible through additive manufacturing.

There were additional pressure taps with a diameter of 0.5 mm on the stator side of the rim

cavity, as indicated in Figure 3.4, which were circumferentially located near the rim seal taps. The

pressure taps on the rotor side of the rim seal and cavity were spread circumferentially throughout

the test section and had a diameter of 0.9 mm. On the rotor side of the rim cavity there were taps at

the same circumferential locations as the stator side taps. Additionally there were pressure taps and

sampling probes spread throughout the rim seal at various axial locations. The sampling probes,

indicated by the triangles in Figure 3.4, were rigid tubes of 1 mm diameter that extended from the

rotor side into the flow in both the rim seal and rim cavity.

52

Purge and Leakage Flow Rates

In this paper several of the figures contain data corresponding to various purge or leakage

flow rates. It is important to note that the flow rates presented in these figures were normalized as

a percent of the full span turbine inlet mass flow rate rather than the half span turbine inlet mass

flow rate. Although half span airfoils were used in this study, the results were more directly

comparable to operating turbines by using the full span mass flow percentages. As such, the full

span mass flows were used as the reference.

3.4 Facility and First Vane Benchmarking

Extensive benchmarking experiments were performed for the facility and the test turbine.

The facility was designed to operate in a steady state mode and has been shown to hold pressure

and temperature for over 10 hours. Thermal steady state was reached in the turbine within two

hours.

The facility valves were designed to allow independent control of the vane inlet pressure,

vane exit pressure, and the main gas path flow rate thereby providing a range of Reynolds and

Mach number operating conditions. The operating range of the facility for this test turbine in terms

Figure 3.4. Test turbine instrumentation.

Kiel pressure probeStatic pressure tapSampling probeTotal temperature

MFG

Purge

Airfoil surface tapsKiel temperature probesKiel pressure probesRim seal tapsRim cavity tapsVane plenum TC’sVane plenum probes

Instrumentation type

Circumferential arrangement

Rotor sideStator

side

53

of the vane exit Mach number and the Reynolds number based on the vane exit conditions is shown

in Figure 3.5. Note that the vane exit Reynolds number in Figure 3.5 was based on the vane exit

velocity magnitude.

Circumferential uniformity was achieved in the test section as indicated by a variety of

measurements. Figure 3.6 shows the turbine inlet total temperature profile for a typical test. As

shown in the inset image in Figure 3.4 the measurements were obtained at five spanwise locations

across six vanes spaced around the annulus. The circumferential uniformity of the inlet total

temperature profile were all within the uncertainty at each spanwise location. Although not shown,

the turbine inlet total pressure also exhibited circumferential uniformity. Regarding the inlet total

pressure, the standard deviation between the four measurements was typically less than ±0.02% of

the measured value, which was less than half the measurement uncertainty of ±0.05%.

The aerodynamic loading on the first vane was measured on two vanes at 50% span as

shown in Figure 3.4. The agreement of the data between the different vanes exhibited excellent

circumferential uniformity. The data agreed well with each other, indicating a circumferentially

uniform flow through the first vanes. Similar circumferential uniformity was also observed at 10%

and 90% spans. Additionally the CFD pre-test predictions, which were generated before running

the experiments [69], and the data agreed well, instilling confidence in both the CFD and the data.

Figure 3.5. Range of operation for these measurements.

0.0

0.2

0.4

0.6

0.8

1.0

0 1 2 3 4 5 6 7 8

Vane exit

Mach number

Rex=ρ∞CxV

μ∞ vane exit

Vane exit Reynolds number (X105)

54

Figure 3.6. Turbine inlet total temperature.

0

10

20

30

40

50

60

70

80

90

100

0.99 1.00 1.01

% span

Normalized Temperature, Tt/Tt,avg

Vane AVane BVane CVane DVane EVane F

Figure 3.7. Circumferential uniformity of the first vane aerodynamic loading at 50% span.

0.5

0.6

0.7

0.8

0.9

1.0

0.0 0.2 0.4 0.6 0.8 1.0

p/pt,in

Percent wetted distance, S/Smax

CFDVane AVane B

55

3.5 CO2 Instrumentation and Data Acquisition

The objective of this paper is to validate the use of CO2 as a tracer gas while ensuring that

the flow field is not altered as a result of the sampling. This section describes the concentration

effectiveness definition used in this paper, the tracer gas injection system, the sampling system, and

the CO2 gas analyzer.

Gas concentration effectiveness, defined in Equation (3.1), is used in this paper to

characterize rim seal performance.

𝜀𝑐

𝑐 − 𝑐∞𝑐𝑠 − 𝑐∞

(3.1)

where c is the CO2 molar concentration, and the subscripts ∞ and s correspond respectively to the

main gas path and the secondary air supply. It is important to note that the main gas path CO2 is

subtracted such that a value of εc = 1 denotes no ingestion (fully sealed), and a value of εc = 0

denotes full ingestion (negligible sealing flow).

The molecular weight of CO2 is higher than that of air, but when used in small quantities

the resulting gas mixture is very similar to air allowing for the heat and mass transfer analogy to

hold. Highly accurate measurements of low CO2 concentrations are made possible by gas analyzers,

which require only small concentrations in the secondary air supply. CO2 is also used because it is

noncorrosive and nontoxic in small concentrations.

The CO2 injection and test turbine sampling system are shown in Figure 3.8. To ensure a

uniform supply concentration, 𝑐𝑠, the CO2 was injected into the secondary air far upstream of the

turbine test section. The mass flow rate of the secondary air supply was measured with a turbine

flow meter, and the CO2 flow rate was set using a mass flow controller such that the supply gas

concentration, 𝑐𝑠, is 10,000 parts per million (ppm), or 1%. The mass flow controller held the CO2

mass flow rate to within ±1% of the set point ensuring the secondary air supply concentration

remained constant over the duration of the test.

56

Measurements of CO2 concentration at four circumferential locations in the first vane

plenum are shown in Figure 3.9 for two different mate face gap leakage flows. The measurement

locations are shown in Figure 3.4 as the vane plenum probes. The concentration was also measured

in the secondary air supply plenum shown in Figure 3.8. Figure 3.9 shows the raw concentration

signal from the gas analyzer with time. The sampling system was used to cycle through the

upstream secondary air supply plenum, and then each of the vane plenum probes spaced around the

annulus. The vertical height of the boxes in Figure 3.9 indicates the bias uncertainty. The

measurements in the vane plenum all agreed with each other within 0.2% and with the secondary

air supply plenum within 0.2%. The data shown in Figure 3.9 indicate steady measurements were

achieved as well as uniform mixing of the CO2 with the secondary air before entering the turbine

and the rim cavity.

Continuous gas samples were extracted through the static taps and then routed to a gas

analyzer where the CO2 concentration was measured. The main gas path concentration, 𝑐∞, was

measured through a kiel pressure probe at the turbine inlet, and the secondary air supply

concentration, 𝑐𝑠, was measured through the first vane plenum probes shown in Figure 3.4. Gas

samples were extracted through pressure taps and probes, shown at various locations in the rim seal

Figure 3.8. CO2 injection and test turbine sampling system.

CO2

RegulatorMass Flow Controller

Secondary Air Supply

Sampling System

Mass Flow Controller

C8 ~0

Cs=1%

Gas Analyzer, C

Turbine Flow Meter

Plenum

57

and rim cavity in Figure 3.4, to characterize concentration effectiveness throughout the turbine. A

sampling system allowed a single gas sample to flow through the gas analyzer, and a mass flow

controller held the gas sampling flow rate to within ±1% of the set point. The sampling flow rate

was varied using the mass flow controller to determine the conditions at which isokinetic sampling

was achieved, which will be further discussed later in the paper.

The gas analyzer used infrared molecular absorption band sensors to measure the CO2

concentration. The infrared absorption bands of CO2 overlap with those of other molecules, but the

gas analyzer achieved highly accurate CO2 concentration measurements by using an infrared beam

divider and a double-layer detector, which minimized the effects of absorption wavelengths

overlapping with other species. Continuous flow was sent to the gas analyzer, which output an

analog voltage corresponding to the CO2 gas concentration.

Figure 3.9. Uniformity of seed concentration in first vane plenum for (a) mate face gap

leakage ṁmfg = 0.35% and (b) mate face gap leakage ṁmfg = 0.15%.

8000

8500

9000

9500

10000

0 2 4 6 8 10

Secondary air supply plenum

Vane B

Vane plenum probes

Vane A

Vane C

Vane D

8000

8500

9000

9500

10000

0 2 4 6 8 10Time [min]

Secondary air supply plenum

Vane B

Vane plenum probes

Vane A

Vane C

Vane D

(a)

(b)

CO2

[ppm]

CO2

[ppm]

58

3.6 Uncertainty and Repeatability

The overall uncertainty of the facility and test turbine measurements is given in Table 3.2,

as calculated from the instrument bias error and the precision error according to the method of

Figliola and Beasley [54]. Precision uncertainty was typically low for the measurements due to the

steady state capability of the facility.

The facility venturi flow meters and the adjoining pipes were calibrated by a commercial

laboratory to within ±0.34-0.47% of the measurement across the full range. The rated accuracy of

the electronic pressure scanner transducers was ±0.05% of full scale (FS), with ranges of 35, 100,

200 and 350 kPa. A pressure calibration system was used daily to perform calibration checks and,

if necessary, calibrations of the pressure scanner system.

The accuracy of the gas analyzer was ±1% of the full scale range, and the full scale range

could be set as low as 1,000 ppm or as high as 10,000 ppm through calibration. For the experiments

presented in this paper the gas analyzer was calibrated at two ranges with a calibration gas: 1,000

ppm and 10,000 ppm. To reduce uncertainty in the concentration effectiveness low values of gas

concentration, such as 𝑐, were measured with the 1,000 ppm range and higher values were measured

with the 10,000 ppm range. By using the two ranges on the gas analyzer the resulting bias

uncertainty in concentration effectiveness was ±0.013 for values of 0 < εc < 1.

To minimize the precision uncertainty a 60 second time average of at least 40,000 samples

was computed once the concentration signal was steady with time. It was this time-averaged

Table 3.2. Overall Measurement Uncertainty

Parameter Uncertainty

Turbine mass flow rate (% meas) ±0.34 to ±0.47

Secondary air flow rate (% meas) ±1.0

Turbine pressures (% FS) ±0.05

Vane aerodynamic loading, p/pt,in ±0.005

Facility temperatures (K) ±0.15

Turbine temperatures (K) ±0.5

CO2 concentration (% FS) ±1.0

Concentration effectiveness, εc ±0.015

59

concentration value that was used to compute concentration effectiveness in Equation (3.1). The

precision uncertainty for the concentration measurements was within ±0.2% of the measured value.

The overall uncertainty in concentration effectiveness was εc = ±0.015.

It was important that both the facility and the turbine test section exhibit repeatability even

as the ambient conditions change and after repeated disassembly and reassembly. The repeatability

of the vane aerodynamic loading was p/pt,in = ±0.002, including the effects of two different vanes,

disassembly and reassembly, and multiple test runs. Repeatability of the concentration

effectiveness measurements in the rim seal and in the rim cavity was typically within εc = ±0.015

and at most within εc = ±0.02.

3.7 Validating CO2 Sampling Methods

Achieving isokinetic sampling conditions is of paramount importance when using a tracer

gas to determine the representative flow physics and ultimately the success of sealing a turbine rim

cavity from the hot main gas path flow. The validation reported in this paper included evaluating

the supply concentration levels as well as the sampling methods in four different tests over a range

of flow conditions. The first test included sampling in the first vane plenum of the test turbine

where the flow was quasi-stagnant. The validation intent for this first test was to establish a baseline

sensitivity study to ensure the gas analyzer was working properly and that the CO2 was uniformly

spread throughout the turbine wheel. The second test included a benchtop experiment using a pipe

flow (outside of the test turbine) to validate sampling for high velocity flows similar to that

encountered in the rim cavity. The third and fourth tests included the rim cavity of the test turbine

and the rim seal (note both regions are defined in Figure 3.3), which were the primary areas of

interest for these secondary flow studies. Two distinct secondary flows were investigated for the

validation in the third and fourth tests: (1) the purge flow, which provided air to the rim cavity

through 150 discrete holes that acted to seal the cavity, and (2) the leakage flow between the mate

face gaps of adjoining vane doublets, which flowed from the first vane plenum to the rim seal.

First Vane Plenum Measurements

Prior to sampling the CO2 tracer gas over a range of leakage flow rates from the mate face

gap, an initial study was completed using a well-characterized calibration gas, which was a mixture

of 9400 ppm CO2 in pure nitrogen. The calibration gas bypassed the turbine and flowed directly

through the gas analyzer. The purpose of this first test was to isolate the gas analyzer and ensure it

60

was operating properly. Figure 3.10 shows that the measured concentration levels remained

constant within the measurement uncertainty for the full range of calibration gas flow rates. This

initial test gave confidence that the gas analyzer was operating as expected.

To acquire unbiased flow tracer measurements, an evaluation was needed to determine

whether there was a uniform concentration of the tracer gas in the secondary flow supply.

Validation of a uniform concentration was achieved through measured concentration levels in the

first vane plenum supply, which should not change with sampling flow rate. In this case, there was

no dominant flow within the stagnant plenum and, as such, there was no concern in altering the

stagnant flow even at high sampling flows. Rather, the objective of these tests was to validate a

constant concentration level would be measured no matter the sampling flow. During these

experiments, the main gas path flow in the test turbine was set to design conditions.

As previously described, Figure 3.9 shows the first vane plenum concentration was

uniformly mixed throughout all the locations. As is also shown in Figure 3.10, a number of CO2

sampling flow rates were conducted in the first vane plenum for two leakage flow rates through the

mate face gap. The measurement location inside the plenum is indicated by the “x” in the Figure

Figure 3.10. Concentration measurements in the first vane plenum for a range of

sampling flow rates at different leakage flow rates.

0.8

0.9

1

1.1

1.2

1E-07 1E-06 1E-05 1E-04

εc,s

Sampling flow rate [kg/s]

ṁmfg = 0.26%ṁmfg = 0.35%Calibration gas Out of turbine

In vane plenum

61

3.10 inset image. Note that the concentration level used for these experiments was approximately

9,000 ppm. The results in Figure 3.10 were normalized and presented in terms of the supply

concentration effectiveness, c,s, as defined in the nomenclature. Again, it was expected that these

tests would show a constant effectiveness of unity, which was indicative of uniformly distributed

CO2 tracer gas. Figure 3.10 shows that for each of the leakage flows and all sampling flow rates,

the measured concentration was invariant indicating that the CO2 was uniformly distributed for the

conditions of interest.

Pipe Flow Measurements

The half stage turbine used in these experiments included a static rim seal and rim cavity

with no rotor as shown in Figure 3.3. Because of the high turning of the vane there were significant

swirl velocities that occurred in the rim cavity, which were in the compressible flow regime [70].

As such achieving isokinetic CO2 sampling in the rim seal area presented yet another challenge. To

ensure that the sampling methodology held for representative compressible flow conditions as those

in the rim seal and cavity, a benchtop experiment was used to perform sampling flow rate sensitivity

studies with compressible flow, as would occur in the rim seal and rim cavity. Figure 3.11 shows a

diagram of the benchtop pipe experiment, outside of the test turbine, that was used for this test.

Similar to the turbine secondary air supply shown in Figure 3.8, CO2 was injected into a pipe with

flow supplied by a compressor. The pipe had a diameter designed for the available compressor flow

to provide a mean velocity equal to the rim cavity swirl velocity in the test turbine, which was at a

Mach number of approximately 0.3. In the pipe experiment, the CO2 was completely mixed with

the air supply due to the long length of both pipes as shown in Figure 3.11. Samples were measured

using two different diameter static pressure taps in the walls of the pipe to also determine the

sensitivity to the tap diameter. Figure 3.11 shows there was no change in the measured

concentration for varying the sampling flow rate or the pressure tap diameter to within the

measurement uncertainty. These results showed that even in a compressible flow, as long as

concentration was uniform, the measured concentration was constant with sampling flow rate.

62

Rim Cavity Measurements

The purpose of the first validation study performed in the cavity was to determine the

sensitivity of the concentration effectiveness measurements to the secondary air supply CO2

concentration. Figure 3.12 shows the concentration effectiveness measurements for three levels of

supply concentration for a low and a high purge flow rate at four locations as indicated in the inset

image. The concentration effectiveness measurements at all conditions in the figure were

insensitive to the supply concentration. The uncertainty of the measurements increased as supply

concentration decreased. A secondary air supply CO2 concentration of 10,000 ppm was used for all

experiments to reduce the uncertainty.

Figure 3.11. Benchtop experiment to validate gas sampling method with measurements in

a pipe with a similar velocity to the rim cavity swirl velocity, at a Mach number of

approximately 0.3.

Shop Air Supply

CO2 Supply

Turbine Flow Meter

Mass Flow Controller Gas Analyzer

L/D=80L/D=100

Mass Flow Controller

0.8

0.9

1.0

1.1

1.2

1E-07 1E-06 1E-05 1E-04

εc

Sampling flow rate [kg/s]

D=1.6 mmD=1.0 mm

Pressure tap diameter

63

Because the flow field in the rim cavity was quite complex, it was also important to verify

isokinetic sampling could be achieved using the CO2 tracer gas for both the mate face gap leakage

flow and the purge flow. The mate face gap leakage flow results are shown first and the purge flow

results follow. Figure 3.3 shows the injection location and the area defined as the rim cavity. For

all of the rim cavity experiments, the main gas path was present at the test turbine design conditions.

The inset image in Figure 3.13 shows the sampling location was on the stator side of the

cavity for the rim cavity studies of the mate face gap leakage flow. Recall the mate face gap leakage

flowed through a slot between adjoining vane doublets. Figure 3.13 shows the same sampling flow

rates and leakage flow rates as were used for the first vane plenum. The concentration levels were

much below unity indicating that there was significant main gas path flow entering the turbine rim

cavity. Even though significant ingestion occurred in the rim seal, the flow ingested from the rim

seal into the rim cavity was at a uniform concentration for a given mate face gap leakage flow rate

as indicated by the constant concentration effectiveness levels at all sampling flow rates. As would

be expected, the concentration effectiveness levels increased with increased mate face gap flows

because better sealing occurred with increased leakage flows. The data in Figure 3.13 showed that

Figure 3.12. Comparison of concentration effectiveness in the rim cavity with varying

secondary air supply CO2 concentrations for two purge flow rates.

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

0 2000 4000 6000 8000 10000

εc

Secondary Air Supply CO2 Concentration [ppm]

ṁp = 2.7%

ṁp = 0.9%

64

there was no appreciable concentration gradient in the rim cavity for the mate face gap leakage

flow. As will be presented later in the paper, there was a concentration gradient in the rim seal

where the ingested air and the mate face gap leakage flow were both introduced.

Figure 3.14 shows a sampling flow rate sensitivity performed at several locations in the

rim cavity for the purge flow to determine the effects of sampling flow rates for the CO2 tracer gas.

The purge flow entered the rim cavity uniformly through 150 discrete purge holes as shown in

Figure 3.3. A purge flow rate of 0.9% of the main gas path was used for these studies. Note that the

test turbine was operated at design conditions and there was no mate face gap flow. The inset image

of the turbine in Figure 3.14 contains a legend showing where each measurement was located in

the rim cavity.

Figure 3.13. Concentration effectiveness measurements in the rim cavity for a range of

sampling flow rates and mate face gap leakage flow rates.

0.0

0.1

0.2

0.3

0.4

0.5

1E-07 1E-06 1E-05 1E-04

εc

Sampling flow rate [kg/s]

ṁmfg = 0.35%

ṁmfg = 0.26%

ṁmfg = 0.15%

65

The main finding in Figure 3.14 was that at low sampling flows, the concentration

measurements were invariant. As the sampling flow rate increased, however, the measured

concentration effectiveness increased with sampling flow rate. The data shown in Figure 3.14

suggested that at higher sampling flow rates the sampling method influenced the flow field thereby

resulting in non-isokinetic sampling conditions. For this test, a concentration gradient existed in

the wall normal direction in the rim cavity, due to the shear layers between the purge flow with the

ingested main gas path flow which carried only background CO2 levels. The purge flow entered

the rim cavity with a concentration effectiveness of unity, but significant ingestion occurred that

acted to dilute the purge flow and reduce the mean concentration in the rim cavity.

At high sampling flow rates as shown in Figure 3.14, the data suggested the gas sample

was not withdrawn from the flow right at the sampling probe or tap. It is interesting to note that the

concentration levels on both the vane side of the rim cavity and on the downstream rotor side of

the cavity were both relatively low values at isokinetic sampling conditions because the high

concentration measured at high sampling flow rates suggested that the sampling method affected

the flow field by redirecting the purge flow toward the sampling taps.

Figure 3.14. Comparison of measured concentration effectiveness for varying sampling flow

rates in the rim cavity for purge ṁp = 0.9%.

0.0

0.1

0.2

0.3

0.4

0.5

1E-07 1E-06 1E-05 1E-04

εc

Sampling flow rate [kg/s]

66

Below a sampling flow rate of 2E-6 kg/s the measured concentration effectiveness was

constant with sampling flow rate, indicating that the flow field was not affected by the sampling

method. For sampling taps at the walls where the no slip condition dictates zero velocity, a sampling

flow rate of zero would correspond to perfect isokinetic sampling conditions. In practice, there was

a range over which sampling flow rates yielded the same measured concentration as is shown in

Figure 3.14 up to 2E-6 kg/s, which were considered for these studies to be isokinetic sampling

conditions.

Rim Seal Measurements

Similar to the rim cavity the flow field in the rim seal was complex with regions of

ingestion and flow egress. Significant ingestion occurred in the rim seal, so a strong concentration

gradient existed, likely in the axial and radial directions, requiring a sampling sensitivity study to

verify isokinetic sampling could be achieved in the rim seal. This section describes the effects of

sampling flow rate on concentration measurements in the rim seal for both the mate face gap

leakage flow and the purge flow. The mate face gap leakage flow is shown first, and the purge flow

follows. The injection of both the mate face gap leakage and the purge flow from the 150 uniformly

distributed holes remained the same as was discussed in the previous two sections. The test turbine

was also operated at design conditions for these studies.

It was expected that a concentration gradient in all three spatial directions would be present

in the rim seal for the mate face gap leakage flow resulting from the high degree of ingestion. The

sampling flow rate sensitivity of concentration effectiveness at two locations in the rim seal for

three mate face gap leakage flow rates is shown in Figure 3.15. The symbol colors represent

different mate face gap leakage flow rates, and the symbols represent different pitchwise locations

in the rim seal at the same radius relative to the mate face gap leakage slots as indicated in the inset

image in Figure 3.15. The triangles are located at 15% pitch, and the diamonds are located at 40%

pitch. Note that the mate face gap leakage slot was located at 35% vane pitch in the rim seal. Similar

to the rim cavity, as shown in Figure 3.14, the concentration effectiveness increased with sampling

flow rate and was constant for a sampling flow rate less than 2E-6 kg/s. Figure 3.15 also shows that

there was a concentration gradient that existed in the rim seal for the mate face gap leakage flow,

although it was not clear from the measurements in which direction the gradient existed.

67

Concentration gradients existed in locations where flow was not directly injected as

evidenced in Figure 3.16. Figure 3.16a presents the sensitivity of the concentration effectiveness

across the axial gap of the rim seal in the rim cavity for a purge flow of ṁp = 1.6%. Figure 3.16a

shows the sampling flow rate sensitivity for three axial locations, as indicated in the inset image.

The concentration effectiveness was again shown to be invariant for sampling flow rates less than

2E-6 kg/s, indicating isokinetic sampling conditions were achieved in the rim seal.

The concentration effectiveness was measured at isokinetic sampling conditions at five

axial locations at the same radius in the rim seal and the variation is shown in Figure 3.16b as a

function of the axial location across the rim seal at the same purge flow of ṁp = 1.6%. The

concentration effectiveness increased with distance from the stator side of the rim seal suggesting

three major conclusions: (1) the purge flow exited the rim cavity and mainly stayed on the rotor

side of the rim seal, (2) the ingested main gas path flow primarily stayed on the stator side of the

rim seal, and (3) an axial concentration gradient existed between the stator and rotor sides of the

rim seal where the purge and ingested flows mixed. As stated, at high sampling flow rates the

sampled flow entered the probe from all directions, which pulled purge air from the rim cavity. The

Figure 3.15. Comparison of measured concentration effectiveness for varying sampling flow

rates in the rim seal for three mate face gap leakage flows.

0.0

0.1

0.2

0.3

0.4

0.5

1E-07 1E-06 1E-05 1E-04

εc

Sampling flow rate [kg/s]

ṁmfg= 0.15% 0.26% 0.35%

15% pitch

40% pitch

Vane

0% pitch100% pitch

Vane

MFG leakage slot

68

concentration effectiveness levels in the rim cavity were also noted in the inset image in Figure

3.16b. The increase in effectiveness observed at higher sampling flow rates was likely due to the

superposed radially outward flow of the purge air as it entered the rim seal. Higher effectiveness

levels were observed in the rim cavity than in the rim seal, indicating that there were significant

shear layers between the ingested flow and the purge flow in the rim seal inner clearance, further

emphasizing the need for careful sampling methods.

The trend of higher effectiveness on the rotor side of the rim seal compared to the stator

side existed at a range of purge flow rates as shown in Figure 3.17. Even at higher purge flows of

ṁp = 2.3% and 2.7% the concentration effectiveness on the rotor side was higher than on the stator

side indicating the purge flow mainly stayed on the rotor side of the rim seal across a wide range

of purge flows. Repeatability of the effectiveness measurements was also shown in the figure, and

the measurements all agreed within the uncertainty of ±0.015.

Figure 3.16. (a) Sampling flow rate sensitivity at three axial positions in the rim seal, and (b)

concentration effectiveness across rim seal axial gap; both at ṁp = 1.6%.

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

1E-07 1E-06 1E-05 1E-04

εc

Sampling flow rate [kg/s]

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

0 25 50 75 100

εc

% axial gap in rim seal

(a) (b)

εc=0.66

0.87

0.850.72

69

3.8 Mate Face Gap Leakage Effects in Rim Seal

Using a tracer gas, such as CO2, for flow tracing is a powerful diagnostic tool in assessing

the performance of rim seals and identifying the representative local flow phenomena. Given the

validation studies previously described, a sampling flow rate of 2E-6 kg/s was used to investigate

the flow physics in the rim seal region in order to achieve isokinetic sampling conditions. CO2

tracing was used to spatially track the leakage flow in the rim seal region for the case with a range

of mate face gap leakage flows. Concentration effectiveness levels are given in Figure 3.18 at a

constant radius in the middle of the rim seal on the stator side. The inset image in Figure 3.18 shows

an isometric view of the vane doublet and the rim seal. The mate face gap leakage slots are also

shown, located at 0% and 100% pitch in the circumferential direction. Note that the swirl direction

was clockwise. Six sampling taps were spread circumferentially through the stator side of the rim

seal as shown in the inset image by the different colors.

Figure 3.17. Concentration effectiveness measurements on the stator and rotor sides of the

rim seal for multiple purge flow rates.

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

0.9% 1.6% 2.3% 2.7%

εc

ṁp [%]

Rotor side, test 1Rotor side, repeat testStator side, test 1Stator side, repeat test

70

The data in Figure 3.18 indicated that there was a significant variation in the concentration

effectiveness with mate face gap leakage flow rates and circumferential location. The concentration

effectiveness increased with the mate face gap leakage flow rate throughout the rim seal. At higher

leakage flows the concentration effectiveness approached a constant value with increasing mate

face gap leakage flow rates indicating that the leakage may have been choking. The mate face gap

leakage flow alone could not provide enough flow to fully seal the rim seal from ingestion, but

contributed some effectiveness in the rim seal as indicated in Figure 3.18. Local effectiveness

values as high as 0.5 on the stator side of the rim seal were shown to exist just downstream of the

leakage slot.

Regarding the circumferential variation, the measured data was consistent with the mate

face gap leakage flow being similar to a planar jet in a strong crossflow that resulted from the swirl

velocity. Similar to a jet-in-crossflow, high concentration effectiveness levels (representing the

leakage flow) occurred just downstream of the leakage slot followed by a decay as the distance

from the leakage slot increased. At higher mate face gap leakage flow rates, the effectiveness

continued to show a variation in the circumferential direction, but the effectiveness began to level

off with increasing leakage flow rate. The data suggested strong concentration gradients existed in

the axial direction in the rim seal. Curiously the effectiveness was zero for ṁmfg = 0.1%, suggesting

that the ingested hot main gas path flow overpowered the leakage flow at low flows. The physical

mechanism for why the effectiveness was so low is not well understood at this time, but further

experiments are expected to reveal the flow physics.

71

3.9 Conclusions

This paper presented benchmarking and validation of the use of a tracer gas for quantifying

sealing effectiveness in an engine realistic turbine rim seal. The commissioning and benchmarking

of the facility, test turbine, and instrumentation were described. The focus of this paper was on the

measurement technique used to characterize rim seal performance, which was the use of CO2 as a

tracer gas in the secondary air supply.

A high pressure, steady state, open loop turbine research facility has been successfully

commissioned, and was shown to be capable of simulating engine relevant conditions using engine

realistic hardware. Benchmarking experiments showed that the facility exhibited long duration

steady state capability, turbine inlet uniformity, and repeatability. The facility was also shown to

reliably simulate a range of Reynolds numbers and Mach numbers in the turbine. The

measurements described in this paper were for a heavily instrumented half-stage (vane only) turbine

design with a realistic rim seal and rim cavity. The successful commissioning of the test turbine

Figure 3.18. Concentration effectiveness on the stator side of the rim seal for multiple

circumferential locations.

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

0.0 0.1 0.2 0.3 0.4

εc

ṁmfg [%]

Rim seal, 15% pitchRim seal, 29% pitchRim seal, 40% pitchRim seal, 53% pitchRim seal, 65% pitchRim seal, 78% pitch

Vane

0% pitch

100% pitch

Vane

MFG leakage slot

72

was also shown, with an emphasis on the test section uniformity, repeatability, and agreement with

CFD pre-test predictions.

The use of CO2 as a tracer gas was validated in this paper for quantifying rim seal and

purge flow performance for engine realistic hardware at engine relevant Mach and Reynold

numbers. Careful sampling was required to ensure that accurate concentration effectiveness

measurements were obtained. In regions where purge and leakage flows mixed with ingested flow,

the measured concentration was sensitive to the sampling flow rate. At high sampling flow rates,

the concentration measurements suggested that the sampling method could affect the flow field. At

low sampling flow rates, the concentration measurements were constant with sampling flow rate,

indicating that the flow field was not affected by the sampling method and, as such, isokinetic

sampling conditions were achieved. For the results presented in this paper a sampling flow rate of

2E-6 kg/s was found to correspond to isokinetic sampling conditions in the rim seal and the rim

cavity.

CO2 tracing was proven to be a powerful method not only for quantifying rim seal and

purge flow performance, but also for deducing flow patterns in the rim seal and rim cavity region.

Deducing flow patterns is especially important for engine realistic hardware as the various leakage

flows contribute to the complexity of the flows in the rim seal and rim cavity. Many rim seal

performance studies consider the effects of a uniform purge flow originating from deep with the

cavity, but the leakage flows in the rim seal and rim cavity contribute significantly to the local

concentration effectiveness where most of the hot gas ingestion occurs. Although it is not a

designed purge flow the mate face gap leakage flow contributes to the concentration effectiveness

in a turbine rim seal. Although the mate face gap leakage flow alone was not sufficient to fully seal

the rim seal from ingestion, the contribution of the mate face gap leakage flow to the concentration

effectiveness in the rim seal is significant.

The data presented is also of value for benchmarking numerical simulations since a tracer

gas can also be modelled in such simulations. The verification of numerical models for a designer

is important because purge and leakage flows need to be accurately predicted. If insufficient flow

is provided to seal the cavity, it means there will be hardware durability issues. Alternatively, an

excess of secondary air would result in unnecessary penalties in engine efficiency. Results from

these initial studies show the complexity of the flow pattern in the rim seal and cavity and the

importance of understanding these flows for determining better sealing methods.

73

Chapter 4

Effects of Purge Jet Momentum on Sealing Effectiveness2

Abstract

Driven by the need for higher cycle efficiencies, overall pressure ratios for gas turbine

engines continue to be pushed higher thereby resulting in increasing gas temperatures. Secondary

air, bled from the compressor, is used to cool turbine components and seal the cavities between

stages from the hot main gas path. This paper compares a range of purge flows and two different

purge hole configurations for introducing the purge flow into the rim cavities. In addition, the mate

face gap leakage between vanes is investigated. For this particular study, stationary vanes at engine

relevant Mach and Reynolds numbers were used with a static rim seal and rim cavity to remove

rotational effects and isolate gas path effects. Sealing effectiveness measurements, deduced from

the use of CO2 as a flow tracer, indicate that the effectiveness levels on the stator and rotor side of

the cavity depend on the mass and momentum flux ratios of the purge jets relative to the swirl

velocity. For a given purge flow rate, fewer purge holes resulted in better sealing than the case with

a larger number of holes.

4.1 Introduction

Gas turbines are used extensively in both aviation propulsion and power generation

applications, so increases in efficiency are desirable. To increase cycle efficiency, overall pressure

ratios for gas turbine engines continue to rise causing increased turbine inlet temperatures, which

can lead to component durability concerns and increased maintenance costs. Secondary air, which

is bled from the compressor, is required to both cool components and seal cavities against hot gas

2 Clark, K., Barringer, M., Thole, K., Clum, C., Hiester, P., Memory, C., and Robak, C., 2016, “Effects of Purge Jet

Momentum on Sealing Effectiveness,” Proc. ASME Turbo Expo, GT2016-58099. Accepted for publication in the ASME

Journal of Engineering for Gas Turbines and Power.

74

ingestion. Efficient use of the secondary air is necessary as excessive use causes a parasitic loss in

engine efficiency.

The cooling air is directed to the cavity regions inboard of the airfoil platform to counter

the effects of hot gas ingestion, which is driven by pressure fields in the main gas path, disk

pumping in the cavities, and turbulent transport between the gas path and cavities. Rim seals located

at the airfoil endwall platform between rotating and stationary components are used to isolate the

main gas path from the cavities, minimize ingestion, and maintain component durability. Rim seals

use combinations of radial and axial overlapping geometries to minimize hot gas ingestion. Sealing

air is still required to purge ingested hot gas from the cavity. Additionally, gaps between segmented

hardware of a single stage and secondary air leakages through those gaps contribute to the already

complex flow patterns in the rim seal and cavity regions. Because of this geometric and flow field

complexity, there is a need for high fidelity predictive methods to accurately model the cavity flow

physics. Experiments at engine-relevant conditions are required to validate new designs and

computational tools.

This paper provides a unique study as the seal and cavity geometry is engine-relevant

unlike many past studies that have used simplified geometries, and the purge flow delivery methods

are investigated. The combination of these two factors has not been previously reported. This paper

describes the sealing effectiveness for a range of different flow conditions and sealing

configurations for an engine-realistic rim seal for engine-relevant Mach and Reynolds numbers

without rotational effects. The use of CO2 as a tracer gas in the secondary air was used to quantify

sealing effectiveness throughout the rim seal and rim cavity. For this study, sealing effectiveness

measurements were made only for the first vane with a static rim seal and rim cavity in place of the

1.5 stage test turbine.

4.2 Review of Literature

Many rim seal and hot gas ingestion studies exist in the open literature, but there are few

with engine relevant geometries found in modern turbines. Several studies have identified the

effectiveness of radial overlap seals and double seals [60,71,72]. Others have identified the

importance of rotational effects such as disk pumping [58,61,72]. Additional factors such as three-

dimensional, unsteady interactions of the vane-blade pressure fields have been shown to affect

ingestion into the rim seal region [9]. Rim seal geometries in turbine engines are designed to

75

minimize ingestion of hot main gas path flow into regions under the airfoil platforms. The majority

of the rim seal studies in the literature, however, have used simplified geometries but have provided

important fundamental knowledge of cavity flows, rim seals, and hot gas ingestion. Although some

engine-realistic rim seals have been published [62,68,73,74], there is a need for more studies in the

open literature that include the complex overlaps, buffer cavities, and flow physics of engine-

realistic rim seals.

In addition to complex rim seal geometries, the effects of the methods used to deliver the

purge flow are not well understood. The different methods for delivering purge flow to the rim

cavity have not been studied extensively, but have been shown to affect the level of ingestion in a

few papers. Purge flow angling and purge flow momentum have been found to affect cavity sealing

effectiveness. Measurements by Coren et al. [75] indicated that the cavity sealing effectiveness

varied greatly with the number of purge flow jets entering a blade-vane cavity. As the number of

purge holes decreased, jet momentum increased and the purge flow spread throughout the rim

cavity leading to more effective rotor disk cooling. A companion CFD study by Andreini et al. [76]

showed that the cavity flow dynamics differed with purge air delivery angles and purge hole

locations. Both the axial and tangential purge flow momentum was shown to affect the cavity

sealing effectiveness. As the purge holes angled toward the rotor the disk experienced more

effective cooling for both low and high purge flow rates. For tangentially angled purge holes the

flow provided significant stator side cooling. Measurements by Coren et al. [75] also supported

these findings.

To accurately represent the flow physics driving hot gas ingestion, it is important that

experiments are performed at engine-relevant conditions. In addition to rotational Reynolds number

[58,61,72], the Mach number at the upstream airfoil exit has been shown to be important in

accurately predicting ingestion. Using simplified but relevant axial and radial overlap rim seals

Teuber et al. [77] showed that the minimum flow rate required to fully seal the rim cavity increased

with airfoil exit Mach number. Experiments were only performed up to Ma = 0.44, but CFD

solutions were generated up to Ma = 0.86 showing an increase in the vane exit non-dimensional

pressure difference, which has been shown to significantly affect ingestion [61,78]. Additional

experimental work at engine-relevant Mach numbers is required to fully understand sealing

effectiveness at engine-relevant conditions and further validate computational models.

Gibson et al. [73] showed that the trends obtained in a low speed linear cascade without

rotational effects were applicable to rotating rim seals as long as a proper crossflow was introduced

at the rim seal and trench. A sector-based annular cascade study by Bunker et al. [79] at engine-

76

relevant Mach numbers showed that cascade measurements aid in understanding and validating

key ingestion flow conditions, although rotation is required to fully represent the instantaneous and

time-averaged physics.

Although the effects of rotation are important to fully represent rim cavity flows, this paper

presents a fundamental study varying the purge air flow rates and delivery methods in an engine-

realistic rim seal at engine-relevant Mach numbers without the effects of rotation to isolate gas path

effects. In the configuration presented in this paper, the hardware permitted investigation of

ingestion dynamics where the gas path contained only an upstream vane row. No rotor or rotational

component was present in the rim cavity. Thus, the results are most applicable to the externally-

induced ingestion regime as discussed in [2, 14], where ingestion is modeled as being driven by the

pressure field in the main gas path. By isolating non-rotating effects in this study we can determine

the degree to which they contribute to hot gas ingestion. The work presented in this paper is unique

given that realistic geometries are operating at or near engine relevant conditions. Measurements

included the use of a tracer gas to deduce the flow phenomena in the cavity region.

4.3 Description of Facility and Turbine

The experiments presented in this paper were performed in a steady state, turbine research

facility, the full design of which was described by Barringer et al. [67]. A detailed description of

the test turbine and instrumentation used in these experiments was given by Clark et al. [80]. A

brief review of the test facility, test turbine, and instrumentation will be provided in this section.

The test facility was an open loop, steady state, flow path capable of simulating engine

relevant conditions for realistic turbine hardware as shown in Figure 4.1. The facility operating

conditions are given in Table 4.1. A large industrial compressor provided high pressure air through

the turbine test section. A second compressor installed in parallel to the first will be used in future

tests to double the mass flow rate capability. For the experiments presented in this paper, one

compressor provided enough high pressure air to the facility for half-span airfoils. Given the focus

of this paper was on the sealing performance of the turbine rim seal and purge flows, it was deemed

appropriate to use only half-span airfoils similar to previous researchers [2, 7, 18]. The turbine was

designed to produce the same pressure field near the rim seal as a turbine with full span airfoils.

A portion of the compressor discharge air was also used to supply the secondary air inboard

of the platform of the turbine test section. The secondary air was cooled in a heat exchanger, and

77

flowed into the test section through hoses routed through the inner diameter of the test turbine. The

secondary air flows were independently controlled with the mass flow rate being measured by a

turbine flow meter.

The facility included instrumentation, as shown in Figure 4.1, to monitor the operating

conditions. Facility temperatures, pressures, and flow rates were respectively measured with

resistance temperature devices, pressure transducers, and calibrated venturi flow meters. Additional

details regarding the facility instrumentation were given in [80]. Flow control valves and a

programmable logic controller (PLC) were used to control the facility operating conditions.

Emergency situations were mitigated by the PLC and fast-acting safety valves that diverted flow

through the by-pass.

Figure 4.1. START facility layout.

78

The design of the test turbine was a 1.5 stage (vane-blade-vane) turbine. For this paper the

test turbine included a half stage (vane only) turbine with an engine realistic rim seal and rim cavity

design. A cross-section of the test turbine is provided in Figure 4.2 with a detailed cross-section of

the rim seal and rim cavity geometry given in Figure 4.3. Figure 4.3 also provides the nomenclature

that will be used to describe the turbine throughout this paper. The colored regions delineate the

areas that will be referred to as the trench, rim seal, and rim cavity.

The secondary air supply entered the test turbine and passed through successive baffle

plates to ensure uniform flow entered the first vane plenum, which in turn supplied the controlled

purge and leakage flows associated with the first vane. Two different secondary flows are presented

in this paper as shown in Figure 4.3. The purge flow entered the rim cavity through uniformly

spaced circular holes oriented in the axial direction. The leakage flow, hereafter referred to as the

mate face gap leakage, was directed through slots at the interfaces between adjoining vane doublets

to the rim seal. Four secondary flow configurations will be discussed in this paper: (1) the purge

flow with 150 holes, (2) the purge flow with 16 holes, (3) the mate face gap leakage flow, and (4)

the purge flow with 16 holes and the mate face gap leakage flow. The ranges of flow rates for each

configuration are provided in Table 4.1. It is important to note that the flow rates presented in this

paper are normalized as a percent of the full span turbine inlet mass flow rate rather than the half

Table 4.1. START Facility Operating Conditions

Parameter Value

Compressor discharge pressure 480 kPa

Compressor discharge temperature 395 K

Compressor mass flow rate (single) 5.7 kg/s

Vane exit Mach number 0.7

Vane exit Reynolds number* 6X105

Purge flow rate – 150 holes up to 2.7%†

Purge flow rate – 16 holes up to 0.5%†

Mate face gap leakage flow rate up to 0.4%†

Purge flow rate – 16 holes with

mate face gap leakage flow rate up to 1.0%†

* based on vane exit velocity magnitude † based on full span main gas path flow rate

79

span turbine inlet mass flow rate. Although half span airfoils were used in this study, the results

were more directly comparable to operating turbines by using the full span mass flow percentages.

The test turbine was instrumented with several static pressure taps that were used for

pressure and gas sampling measurements as reported by Clark et al. [80]. Figure 4.4 shows a cross-

section of the test turbine with the instrumentation. The inset image shows the circumferential

arrangement. The first vanes were additively manufactured through a metal laser sintering process

in pairs, or doublets, and as such, integrated static pressure taps were designed into the vanes. The

static pressure taps were spread throughout the rim seal as indicated in Figure 4.4. Each of these

radial locations had a series of pressure taps at several circumferential locations allowing for

detailed spatially resolved measurements with true scale engine hardware.

80

Figure 4.2. Test Turbine cross-section.

Figure 4.3. Test turbine nomenclature and geometric parameter definitions.

81

To characterize the rim seal performance and determine the flow patterns in the rim seal

and rim cavity, a CO2 tracer gas was used to measure concentration effectiveness. The definition

of concentration effectiveness used in this paper is given in Equation (4.1),

εc 𝑐 − 𝑐∞𝑐𝑠 − 𝑐∞

(4.1)

where c is the measured CO2 molar concentration, and the subscripts ∞ and s correspond

respectively to the main gas path and the secondary air supply. Details regarding the CO2 injection

system, the sampling system, the gas analyzer, as well as a full validation of the technique were

provided in [80].

The secondary air supply was seeded with 1% CO2 by volume. Gas samples were extracted

through the sampling taps shown in Figure 4.4 at a constant sampling flow rate of 2X10-6 kg/s,

corresponding to isokinetic sampling conditions. Continuous flow was sent to a gas analyzer that

measured the CO2 molar concentration with an accuracy of ±1% of the full scale range, and a 60

second time-average was computed after the analyzer signal steadied with time. The overall

uncertainty in the concentration effectiveness measurements was εc = ±0.015, and the repeatability

was typically within εc = ±0.015.

4.4 Facility and First Vane Benchmarking

The test turbine and facility were shown to be successfully benchmarked by Clark et al.

[80]. Additional benchmarking will be provided in this section, particularly in terms of the

circumferential uniformity and a comparison with CFD pre-test predictions performed by the

industry sponsor [69,82]. The measurements presented in this section were obtained with the

turbine vane operating at design conditions.

82

The vane aerodynamic loading is shown in Figure 4.5 for two vanes at 50% span. The

measurements on the different vanes agree within p/pt = ±0.005 on average, which is equal to the

overall uncertainty of p/pt = ±0.005, showing that circumferential uniformity was achieved in the

main gas path. Repeatability for each measurement location was previously shown to be within

p/pt = ±0.002 [80]. The vanes were fore-loaded, with a sharp acceleration evident on the first 20%

wetted distance on the suction surface for all spans. From 65-100% wetted distance on the suction

surface the pressure was relatively constant until downstream where the pressure and suctions side

pressures converged at the vane trailing edge.

Although computational simulations were not the focus of this paper, blind pre-test

numerical predictions were performed. The CFD loading predictions are also given in Figure 4.5

and showed good agreement with the measurements in the main gas path. Good agreement was

also observed in steep gradient regions such as from 10-40% wetted distance. Although not shown

in Figure 4.5 similar circumferential uniformity and agreement between CFD and the

measurements were observed at 10% and 90% spans.

Figure 4.4. Test turbine instrumentation.

83

Many ingestion models rely on the nondimensional pressure difference, Δp/pt,in = (pmax –

pmin)/pt,in, at the rim seal as a boundary condition [2, 14]. To determine the pressure boundary

conditions on the rim seal, static pressures were measured on the endwall at the vane trailing edge,

at the platform trailing edge (22% chord downstream of the vane trailing edge), and in the rim seal

(8% chord upstream of the vane trailing edge) as shown in Figure 4.6. The inset image shows the

axial and radial locations of the measurements where the different symbols correspond to

measurements on vanes located 180° apart. Each of the indicated locations comprised several taps

in the circumferential direction. The vane trailing edge taps were equally-spaced in the passage

between the vanes of the doublet and the platform trailing edge taps were equally-spaced across

1.5 vane pitches. It should be noted that the measurements shown in Figure 4.6 were obtained with

no purge or leakage flows. Although not shown here, no changes were observed in the vane trailing

edge and platform trailing edge pressure profiles even at high purge and leakage flows; however,

as expected, the pressure in the rim seal was observed to increase slightly with purge flow.

It is first noted that the measurements given in Figure 4.6 showed good agreement between

the measurements on different vanes indicating circumferential uniformity was achieved in the

Figure 4.5. First vane aerodynamic loading at 50% span compared to CFD pre-test

predictions.

84

main gas path and trench. The average difference between the measurements was p/pt = ±0.009 at

both the vane trailing edge and at the platform trailing edge. The measurements and the pre-test

predictions agreed well with each other, including the magnitudes, widths, and locations of the

pressure peaks, also shown in Figure 4.6. The measurements and CFD agreed within an average

difference of p/pt = ±0.004, ±0.001, and ±0.0005 at the vane trailing edge, platform trailing edge,

and the rim seal respectively. The pressure peaks due to the vane potential field can be seen at the

trailing edge at θ = 6°, 18°, and 29°.

At the platform trailing edge shown in Figure 4.6, located at 22% axial chord downstream

of the vane trailing edge, there was a strong mixing in the trench region and flow separation over

the step that resulted in a relatively flat static pressure across the trailing edge. The non-dimensional

pressure difference, Δp/pt,in = (pmax – pmin)/pt,in, at the platform trailing edge decayed to 17% of the

vane trailing edge value. Full attenuation of vane exit potential field was observed in the rim seal

where there was essentially no circumferential variation.

Figure 4.6. Pressures on vane trailing edge, platform trailing edge (22% Cx downstream of

vane trailing edge), and in rim seal (8% Cx upstream of vane trailing edge) for the no leakage

case compared to CFD pre-test predictions.

85

Although the rim seal and cavity were static for these particular experiments, significant

tangential velocity was observed in the trench and rim cavity regions as would be expected because

of the large turning of the flow in the main gas path by the vane. The isentropic exit Mach number

for the vane in the main gas path is shown in Figure 4.7 as a reference.

Figure 4.7 shows the Mach numbers in the trench and rim cavity, which were calculated

according to the isentropic flow equations from the ratio of the measured total to static pressures.

The total and static pressures were measured using kiel pressure probes and static pressure taps at

discrete locations in the rim cavity and trench. The legend in Figure 4.7 shows where the total

pressures were measured for each location with kiel pressure probes aligned with the

circumferential direction. The measurements were made in the trench (33% axial chord

downstream of the vane trailing edge and r/b = 0.97), at the rim cavity outer radius (80% of the rim

cavity axial width and r/b = 0.92), and at the rim cavity inner radius (65% of the rim cavity axial

width and r/b = 0.90). The kiel pressure probes were insensitive to flow angle deviations of ±30°.

It is important to note that the flow in the trench and in the rim cavity was almost exclusively in the

tangential, or swirl, direction, so Figure 4.7 shows the Mach number for the flow in the swirl

direction.

The measured Mach number variation in the trench and rim cavity is shown in Figure 4.7

for a range of purge flow rates for the previously mentioned flow geometries: 16 and 150 purge

holes. Figure 4.7 shows that significant swirl velocities were measured in the trench area and,

although lower in magnitude but comparable to the vane exit Mach number, there were significant

swirl velocities in the rim cavity. Since there was no rotor present in these experiments the swirl in

the trench was induced by shear from the flow exiting the vane in the main gas path. The high swirl

velocities in Figure 4.7 are important to keep in mind as one considers the sealing effectiveness of

the purge flow since the purge flow entered as discrete jets in a relatively strong crossflow. Within

the rim cavity, the swirl Mach number appeared to be insensitive to the number of holes indicating

that for a given purge flow the swirl flow in the rim cavity was found to be consistent no matter

whether the purge flow was issued from 16 holes or 150 holes, further suggesting that the swirl

flow was due to the shear from the vane exit flow.

86

As expected the swirl decreased with decreasing radius: the trench swirl Mach number was

lower than that of the vane exit, and the rim cavity swirl Mach number was lower than that of the

trench. The swirl Mach number decreased with increasing purge flow in the trench and in the rim

cavity. As the purge flow increased ingestion decreased, which resulted in lower swirl velocities

propagating into the rim cavity from the main gas path. This behavior was slightly exaggerated

compared to a rim cavity with a rotor, since the swirl in a rotating rim cavity would not go to zero

as shown in Figure 4.7.

4.5 Sealing Effectiveness with Purge Flow

As was stated previously, the focus of this paper was to investigate sealing effectiveness

as a function of the purge flow delivery method into the rim cavity. Two configurations of purge

flow delivery were used: Case 1: 150 purge holes, and Case 2: 16 purge holes. In both cases, the

Figure 4.7. Swirl Mach number in the trench region and the rim cavity for a range of purge

flow rates.

87

mate face gaps were sealed and the turbine was operated at the design flow conditions as shown in

Figure 4.5.

Case 1: 150 Purge Holes

The first flow configuration presented in this paper is for 150 purge holes equally spaced

around the circumference. The intent of using 150 holes, relative to the case with only 16 purge

holes, was to uniformly distribute the purge flow in the rim cavity. The concentration effectiveness,

as previously defined, is presented in Figure 4.8 for both the stator and rotor sides of the rim cavity

at the purge hole radius for the 150 purge hole configuration. The inset image shows the locations

of the purge holes as well as the measurement locations at the same radius as the purge holes.

The flow was uniform with circumferential position as shown in Figure 4.8 for several

purge flow rates. The standard deviation of all the measurements at a given purge flow rate and

radial location was calculated, and circumferential uniformity was within an average standard

Figure 4.8. Circumferential uniformity of concentration effectiveness for 150 purge holes for

multiple purge flows.

88

deviation of εc = ±0.011, and a maximum deviation of εc = ±0.018, indicating the purge flow did

indeed enter the rim cavity uniformly around the annulus. The data in Figure 4.8 indicated that the

effectiveness on the stator side was higher than the rotor side for ṁp ≤ 1.26% at all circumferential

locations. For ṁp > 1.7%, however, the rotor side effectiveness was higher than the stator side as

will be discussed further later in the text.

The concentration effectiveness is presented in Figure 4.9 for a range of purge flow rates

for the 150 purge hole configuration for different locations in the rim seal and cavity. The inset

image shows the locations of the purge jets as well as the measurement locations. Wall static taps

were used for the measurements as well as a sampling probe placed in the cavity. Figure 4.9 also

shows schematics indicating the likely flow patterns in the rim seal and rim cavity for high and low

purge flows as determined by the concentration effectiveness measurements.

As shown in Figure 4.9, the concentration effectiveness increased with purge flow rate at

all locations as would be expected. Also, as expected, the highest concentration effectiveness values

were at the locations deepest inside the cavity. At the low purge flows, the data in Figure 4.9

indicated that the ingested flow was present throughout the rim seal and deep into the rim cavity.

As the purge flow rate increased, the concentration effectiveness levels began to increase more

prominently in the rim cavity relative to the rim seal. It was not until ṁp = 1% that the effectiveness

on the stator and rotor sides of the rim seal began to increase indicating that at this purge flow rate

the walls of the rim seal were positively affected by the purge air. Although not shown here,

effectiveness measurements were obtained in the rim seal axial gap between the stator and rotor

sides. At low purge flows effectiveness levels of 0.05 < εc < 0.15 were measured in the center of

the axial gap identifying the path of the purge flow through the rim seal.

89

The concentration effectiveness measurements in Figure 4.9 also indicated the highest

values for the sampling location were in the center of the cavity as compared to the stator and rotor

walls for all flow rates. These maximum concentration effectiveness levels indicated most of the

purge flow resided in the core of the rim cavity crossflow.

The concentration effectiveness at the same radial location as the purge holes on the stator

and rotor sides of the rim cavity exhibited different trends as shown in Figure 4.9. A cross-over of

the maximum effectiveness occured at ṁp = 1.35%. For ṁp < 1.35% the stator side of the rim cavity

exhibited higher effectiveness than the rotor side. In contrast for ṁp > 1.35%, the rotor side of the

rim cavity exhibited higher effectiveness than the stator side. Figure 4.7 shows there were high

levels of swirl velocity at purge flow rates below 1.5%, which convected the purge flow in the swirl

direction. As illustrated in Figure 4.9, at low purge flows the jets entered the cavity with low

momentum and did not penetrate far into the cavity and thereby did not reach the rotor cavity wall.

Conversely at high purge flows, the jets entered the rim cavity with high momentum, which carried

purge air to the rotor side of the cavity. The difference in purge jet trajectory is illustrated in the

flow schematic images in Figure 4.9 for the high purge flow and low purge flow cases. Since the

Figure 4.9. Concentration effectiveness for 150 purge holes.

90

swirl velocity in the cavity was a confined crossflow, the effectiveness on the stator side increased

monotonically with purge flow rate; however, as the purge flow entered into the cavity crossflow,

the rate of increase in effectiveness with flow rate was lower for the stator side of the rim cavity

than the other locations between 1.5% < ṁpurge < 2.5%.

On the stator side of the rim cavity outboard of the purge holes, the effectiveness trend was

similar to that of the stator side at the purge hole radius, but at lower effectiveness values as shown

in Figure 4.9. These results were consistent with the fact that some of the purge flow mixed with

the rim cavity core flow before exiting the cavity. The flow schematic images show that as the

purge flow entered the rim cavity the flow likely induced a counter-clockwise rotating circulation

in the rim cavity outboard of the purge holes. The flow recirculation region likely existed at this

location inboard of the rim seal inner clearance, which caused the ingested flow to recirculate as

shown in the flow schematic images in Figure 4.9, similar to behavior shown by previous rim seal

studies [2, 3, 8].

The concentration effectiveness was higher in the cavity than in the rim seal as shown in

Figure 4.9. For example, at ṁp = 1.5%, the average effectiveness in the rim cavity was 0.75 while

effectiveness was only 0.3 in the rim seal. In the rim seal, the effectiveness was higher on the rotor

side than on the stator side, indicating the purge flow entered the rim seal from the rim cavity and

convected along the rotor side of the rim seal. In contrast to the purge flow, the ingested flow

mainly stayed on the stator side of the rim seal. This type of flow pattern is typical of a rim cavity

with a rotor due to disk pumping and the thermal buffering effect [83,84], however, it is interesting

to note a consistent trend in this static rim seal where rotational effects have been removed.

Case 2: 16 Purge Holes

The second configuration for introducing the purge flow was the case with only 16

uniformly distributed purge holes around the circumference. The intent of using 16 holes was to

understand the importance of the number of holes where discrete jets were used to locally provide

purge flow to the rim cavity. The purge holes used in this configuration were the same diameter as

the 150 holes and were located at the same radius. As such, the momentum of each jet for the 16

holes was significantly higher than for the case with 150 holes for a given purge flow rate.

With 16 purge holes around the circumference of the rim cavity, concentration

effectiveness was expected to vary with circumferential position. The circumferential variation of

the concentration effectiveness on the stator side of the rim cavity at the purge hole radius is shown

in Figure 4.10 for three purge flow rates. The inset images in Figure 4.10 show the measurement

91

locations in the rim cavity as well as the purge hole location at θ = 15°. Note the swirl direction as

induced from the vanes was clockwise in the inset image. Concentration effectiveness is only

shown for the stator side of the cavity as limited data were acquired for the rotor side.

Significant circumferential variation in concentration effectiveness can be seen in Figure

4.10, especially at low purge flows. The purge jets behaved like a jet-in-crossflow with a decay in

effectiveness with increasing circumferential distance from the purge hole. At low purge flows, the

jets entered the cavity with low momentum and convected mainly along the stator side. At ṁp =

0.17% the effectiveness increased sharply downstream of the purge hole. At ṁp = 0.27% the same

trend was observed as with ṁp = 0.17% with increased effectiveness downstream of the jet

injection, but lower concentration effectiveness than at the ṁp = 0.17% indicating the purge jet

separated from the stator wall. At the highest purge flow rate of ṁp = 0.39% the concentration

effectiveness was mostly uniform with circumferential direction on the stator side, indicating the

purge jet was completely separated from the stator wall.

Figure 4.10. Circumferential variation of concentration effectiveness for 16 purge holes on

the stator side of the rim cavity at the purge hole radius for multiple purge flows.

92

Figure 4.11 shows the variation in concentration effectiveness with purge flow rate for the

four sampling locations on the rim cavity stator side. The inset images show the measurement

locations in the rim cavity, where the symbol colors correspond to different circumferential

locations. Similar to Figure 4.10, Figure 4.11 shows there was significant circumferential variation

in concentration effectiveness, especially from 0.15% < ṁp < 0.4%. The effectiveness increased

downstream of the purge hole, then decayed as the distance from the purge hole increased,

particularly for ṁp < 0.3% where the effectiveness at θ = 18° was significantly higher than at other

circumferential locations.

The effectiveness at θ = 18° behaved similar to that of a jet-in-crossflow. As was shown in

Figure 4.7 significant swirl velocities are measured in the cavity for 16 purge holes. The

concentration effectiveness shown in Figure 4.11 increased sharply with increasing purge flow rate

where the jet appeared to remain attached to the stator wall up to ṁp = 0.2%. As the purge flow

increased the jet momentum increased such that the purge flow separated from the stator wall, and

Figure 4.11. Variation of concentration effectiveness with purge flow rate on the stator side

of the rim cavity at the purge hole radius for 16 purge holes.

93

the effectiveness gradually decayed with flow rate up to ṁp = 0.4%. The effectiveness at the

remaining sampling locations increased monotonically with purge flow rate, although the rate of

increase in effectiveness decreased for ṁp > 0.2%, where the jet appeared to separate from the stator

side of the cavity. The jet did not appear to reattach to the stator wall for ṁp > 0.35%, but the

effectiveness for θ = 23° exhibited higher effectiveness for 0.35% < ṁp < 0.5%, indicating the purge

flow jet mixed with the cavity cross flow as it swirled around the rim cavity.

The variation in effectiveness with purge flow rate at the four circumferential locations was

averaged in Figure 4.12 for each location indicated in the inset image. As observed in Figure 4.11,

effectiveness increased with purge flow rate for ṁp < 0.2%, beyond which the rate of increase in

effectiveness decreased up to ṁp = 0.4%. The momentum of the purge flow caused the jets to

penetrate farther into the rim cavity where the flow mixed with the ingested swirl flow. Limited

concentration effectiveness measurements were averaged for the rotor side and are shown in Figure

4.12. The concentration effectiveness in Figure 4.12 indicated that the purge flow penetrated farther

into the rim cavity at high purge flow rates, which is consistent with high momentum jets. The

cross-over point where the rotor side effectiveness was higher than the stator side effectiveness

occurred at ṁp = 0.4% for 16 purge holes. Recall that for 150 purge holes a similar cross-over

occurred at ṁp = 1.35% as shown in Figure 4.9.

The stator side of the rim seal showed zero effectiveness over the range of flow rates shown

in Figure 4.12. This trend was similar to Figure 4.9, which showed zero effectiveness for ṁp < 1%

in the rim seal. The data showed that significant ingestion occurred in the rim seal for 16 purge

holes. At the outer radius on the stator side of the cavity the effectiveness was again seen in Figure

4.12 to be lower than at the purge hole radius, which was similar to Figure 4.9 for 150 purge holes.

The reduced effectiveness inboard of the rim seal on the stator side was evidence of the flow

recirculation for 16 purge holes similar to 150 purge holes, as shown in Figure 4.9.

94

4.6 Sealing Effectiveness with Mate Face Gap Leakage Flow

The focus of this section is on the sealing effectiveness for different methods of introducing

purge and leakage flows into the rim seal and rim cavity, specifically the mate face gap leakage

that flowed through the gaps between adjoining vane doublets into the rim seal. For the

effectiveness measurements presented in this section, as was previously mentioned, two

configurations of mate face gap leakage flow were used: Case 3: mate face gap leakage only, and

Case 4: mate face gap leakage with 16 purge holes. Again the vane was operated at aerodynamic

design conditions as previously discussed.

Case 3: MFG Leakage Only

For the mate face gap studies reported for Case 3, there was no purge flow present. The

purpose of this configuration was to identify any contributions of the mate face gap leakage flow

Figure 4.12. Averaged concentration effectiveness for 16 purge holes.

95

on the concentration effectiveness in the rim seal and the rim cavity. Concentration effectiveness

for the mate face gap leakage case is given in Figure 4.13. Note that the maximum flow was lower

than the purge flows, which is consistent with what would occur in an operating turbine. The inset

images show the vane doublet, the rim seal, and the mate face gap leakage slots located at 0% and

100% pitch in the circumferential direction. Note that the swirl direction was clockwise. Three

sampling taps were distributed circumferentially through the stator side of the rim seal as shown in

the inset image by the different colored symbols, and three sampling taps were located on the stator

side of the rim cavity.

There was a significant variation in the concentration effectiveness in the rim seal with

mate face gap leakage flow rate and circumferential location as shown in Figure 4.13. The

concentration effectiveness increased with the mate face gap leakage flow at all three sampling

locations in the rim seal. The effectiveness was highest just downstream of the mate face gap

leakage slot, as evidenced by the measurements at 15% pitch, followed by a decay in effectiveness

with increasing distance from the slot, shown by the measurements at 40% and 65% pitch. The

mate face gap leakage flow was similar to an angled planar jet in a strong cross flow, with the

highest effectiveness just downstream of the flow injection location, followed by a decrease in

effectiveness as the distance from the leakage slot increased.

The concentration effectiveness on the stator side of the rim cavity is also shown for the

mate face gap leakage flow in Figure 4.13. The concentration effectiveness increased with leakage

flow rate, but never exceeded εc = 0.2. The effectiveness was observed to be uniform in the cavity

for a given mate face gap leakage flow rate, so Figure 4.13 only shows one symbol at each flow

rate. The uniform effectiveness for a given flow rate suggested the mate face gap leakage mixed

with the ingested main gas path flow and entered the rim cavity with uniform concentration

regardless of circumferential location. Thus, the mate face gap leakage contributed to the sealing

effectiveness in the rim cavity and was shown to be circumferentially uniform.

96

Case 4: MFG with 16 Purge Holes

The fourth case is for the mate face gap leakage with 16 purge holes. The purpose of this

configuration was to determine the sensitivity of the 16 purge holes to the mate face gap leakage

flow. This configuration was a combination of Case 2: 16 purge holes and Case 3: mate face gap

leakage only.

The circumferential variation of concentration effectiveness is presented in Figure 4.14 for

both the stator and rotor sides of the rim cavity (solid and dashed lines respectively) at the purge

hole radius for three flow rates. As in previous figures, the inset image shows the measurement

locations, and the purge hole location is shown in the graph at θ = 15°. It should be noted that the

flow rates presented in the figure are the sum of the purge and mate face gap leakage flow rates.

Figure 4.13. Concentration effectiveness for mate face gap leakage flow only (no purge).

97

Concentration effectiveness increased with purge and leakage flow rate in Figure 4.14. The

results were similar to those for 16 purge holes as shown in Figure 4.10 with significant

circumferential variation in effectiveness. At ṁp + ṁmfg = 0.48%, the concentration effectiveness

on the stator side of the rim cavity showed an increase at θ = 18° similar to that seen at low purge

flow rates in Figure 4.10. The effectiveness on the rotor side was essentially constant for 0.48%

indicating the purge flow stayed attached to the stator side of the cavity and convected to the rotor

side of the cavity only through mixing with the rim cavity swirl flow. At a slightly higher flow rate

of ṁp + ṁmfg = 0.77% the stator side effectiveness was mostly uniform, and the rotor side of the

cavity exhibited a local maximum in concentration effectiveness at θ = 18°, indicating the higher

momentum of the purge flow jet carried high concentration purge air to the rotor side of the cavity.

The effectiveness on the rotor side decreased with distance downstream of the purge hole location.

At an even higher flow rate of ṁp + ṁmfg = 0.95% the trend was even more pronounced, with the

maximum concentration effectiveness again at θ = 18° on the rotor side. At this highest flow

condition, the high momentum of the purge jets resulted in a local concentration effectiveness

Figure 4.14. Circumferential variation in concentration effectiveness on the stator and rotor

sides of the rim cavity for 16 purge holes and the mate face gap leakage.

98

maximum on the rotor side of the cavity across from the purge holes, again followed by a decrease

with distance from the purge jets.

4.7 Scaling of Sealing Effectiveness for Purge

This section presents a comparison of the different purge flow configurations tested in the

rim cavity for Case 1: 150 purge holes and Case 2: 16 purge holes. The mate face gap leakage was

shown to affect concentration effectiveness locally in the rim seal, but showed little effect in the

rim cavity so the mate face gap leakage will not be used in this comparison.

The concentration effectiveness for both 150 purge holes and 16 purge holes is presented

in Figure 4.15 for the range of purge flow rates shown previously. The two inset images contain

corresponding legends for each configuration indicating each measurement location in the rim

cavity, as was shown in Figure 4.9 and Figure 4.12. Note that since the diameter of the holes was

the same for both cases the range of tested flow rates was smaller for 16 purge holes than 150 holes.

Of primary interest in Figure 4.15 was that concentration effectiveness was higher for 16 purge

holes than for 150 purge holes at a given flow rate. The trend of higher concentration effectiveness

for 16 purge holes compared to 150 purge holes was observed at all sampling locations in the rim

cavity.

99

It is clear from the measurements shown in this paper that the dynamics of the purge jet

affect the concentration effectiveness throughout the rim cavity. Similar to a film cooling jet issued

into a crossflow, the purge jets can also be evaluated in terms of two physical parameters that

govern jet dynamics, namely the mass flux (blowing) ratio, M, and the momentum flux ratio, I. It

is worth noting that the density ratio, DR = 𝜌𝑝/𝜌𝑟𝑐, was approximately 1.1 for the experiments

presented in this paper. The crossflow velocity used in the definitions of the mass and momentum

flux ratios was calculated from the swirl Mach number presented in Figure 4.7.

The two purge flow configurations had notably different mass flux and momentum flux

ratios for a given purge flow. Figure 4.16 shows the mass and momentum flux ratios for a range of

purge flow rates for both 150 purge holes and 16 purge holes. The mass and momentum flux ratios

of the 16 purge holes was approximately an order of magnitude higher than that of the 150 purge

holes for a given flow rate.

Figure 4.15. Concentration effectiveness for 150 purge holes and 16 purge holes with varying

purge flow rates.

100

Recall that there was a cross-over in concentration effectiveness between the rotor side of

the rim cavity and the stator side as shown in Figure 4.9 and Figure 4.12. At low purge flow rates,

the stator side exhibited higher effectiveness, and at higher purge flow rates the rotor side was

higher. The cross-over flow rate for 150 purge holes was shown in Figure 4.9 to be ṁp = 1.35%.

This corresponded to a mass flux ratio of M = 2.7 and a momentum flux ratio of 6.5 in Figure 4.16.

The cross-over flow rate for 16 purge holes was shown in Figure 4.12 to be approximately ṁp =

0.4%, which corresponded to M = 2.6 and I = 6 in Figure 4.16. The consistency between the cross-

over point in both purge flow configurations indicated that similar jet dynamics resulted in similar

effectiveness trends in the rim cavity. The agreement also suggested that it was reasonable to scale

the effectiveness measurements with M and I.

The concentration effectiveness in the rim cavity for both 150 purge holes and 16 purge

hole is shown as a function of the mass flux ratio in Figure 4.17 and as a function of momentum

flux ratio in Figure 4.18. It should be noted that since the rim cavity swirl velocity approached zero

as the purge flow increased, M and I lose meaning, so the data are only shown up to M = 10 and I

= 25.

Figure 4.16. Mass flux ratio, M, and momentum flux ratio, I, for 150 purge holes and 16

purge holes.

101

Figure 4.17 shows that the concentration effectiveness measurements nearly collapsed for

150 purge holes and 16 purge holes for M < 1.2. The concentration effectiveness levels at the purge

hole radius on the stator side agreed up to the point where the effectiveness of the 16 purge holes

began to level off, suggesting that insufficient flow was provided to fully purge the cavity with only

16 purge holes. Overall mass flow rate is important to fully seal cavities and an intermediate number

of purge holes may provide sufficient mass flow rate and jet momentum to fully purge the cavity.

The effectiveness on the rotor side of the rim cavity was much lower for 16 holes than for 150

holes, but the cross-over point discussed previously was seen to be consistent for both cases at M

= 2.6.

In Figure 4.18 the concentration effectiveness measurements also collapsed for 150 purge

holes and 16 purge holes for I < 1.3. Again, the concentration effectiveness measurements on the

stator side of the rim cavity exhibited similar values for both 150 and 16 purge holes up to I = 1.3,

where the concentration effectiveness leveled off for 16 purge holes. The cross-over point where

Figure 4.17. Concentration effectiveness for 150 purge holes and 16 purge holes plotted

against blowing ratio.

102

the rotor side effectiveness was higher than the stator side effectiveness was shown for both cases

at I = 6. Additional purge hole configurations between 16 and 150 may provide scaling across a

wider range of mass flux and momentum flux ratios.

4.8 Conclusions

The sealing of rim cavities is very important for gas turbine engines in order to avoid

catastrophic failures resulting from hot gas ingestion between stages and other gaps. The results

presented in this paper for a static vane with engine relevant hardware operated at design conditions

indicated the importance of understanding the usage of secondary air supplies. Sealing

effectiveness levels for different methods of introducing purge and leakage flow into the rim seal

and rim cavity were presented. The sealing effectiveness results were presented for four different

configurations to investigate the number of purge holes as well as the influence of the mate face

gap leakage.

Figure 4.18. Concentration effectiveness for 150 purge holes and 16 purge holes plotted

against momentum flux ratio.

103

High swirl velocities were measured in the trench region that were important to

understanding how the jets from the purge flow reacted in the rim cavity. The purge jets were found

to be similar to that of a jet-in-crossflow while the mate face gap leakage was similar to that of a

planar jet-in-crossflow.

The sealing effectiveness was highly dependent upon how the purge flow was introduced

into the cavity. The results indicated better sealing with fewer purge holes as opposed to a large

number of holes. Fewer purge holes resulted in higher momentum flux jets for a given purge flow

rate as compared to a larger number of holes. No matter the number of purge holes, the

concentration effectiveness exhibited a cross-over in the maximum sealing effectiveness at the

same mass flux ratio and momentum flux ratio. For low jet momentum flux ratios, the sealing

effectiveness on the stator side was higher than on the rotor side of the rim cavity, while for high

jet momentum flux ratios, the sealing effectiveness on the rotor side was higher than the stator side

of the cavity. It should be noted that when a rotor is included in future studies, the disk pumping

may decrease the value of M and I corresponding to the cross-over point.

The mate face gap leakage flow affected sealing effectiveness in the circumferential

direction in the rim seal. Uniform sealing effectiveness was measured in the rim cavity as the mate

face gap leakage mixed with the ingested flow before entering the rim cavity. As such, the mate

face gap leakage flow contributed a small amount to the sealing effectiveness in the rim cavity.

Results from these initial studies have indicated the complexity of the flow field in the rim

seal and cavity and the importance of understanding the flows for determining better sealing

methods. Because of the flow field complexity there is a need for high fidelity predictive methods

to accurately model the seal air and cavity flow physics to predict sealing effectiveness. These

experiments at engine-relevant conditions will help validate new designs and computational tools.

Future studies will include rotational effects in the full 1.5 stage turbine, which may further

complicate the flow field in the rim seal and cavity.

104

Chapter 5

Effects of Purge Flow Configuration on Sealing Effectiveness

in a Rotor-Stator Cavity3

Abstract

Secondary air is bled from the compressor in a gas turbine engine to cool turbine

components and seal the cavities between stages. Unsealed cavities can lead to hot gas ingestion,

which can degrade critical components or, in extreme cases, can be catastrophic to engines.

Understanding the physics of hot gas ingestion at engine-relevant conditions is crucial for engine

designers. For this study, a 1.5 stage turbine with an engine-realistic rim seal was operated at an

engine-relevant axial Reynolds number, rotational Reynolds number, and Mach number. Purge

flow was introduced into the inter-stage cavity through distinct purge holes for two different

configurations. This paper compares the two configurations over a range of purge flow rates.

Sealing effectiveness measurements, deduced from the use of CO2 as a flow tracer, indicated that

the sealing characteristics were improved by increasing the number of uniformly distributed purge

holes. Effectiveness levels at some locations within the cavity were well-predicted by an orifice

model, but due to the complexity of the realistic rim seal and the purge flow delivery, the

effectiveness levels at other locations were not well-predicted.

5.1 Introduction

High efficiencies and power density in gas turbines are increasingly important for both

power generation and aviation propulsion. To increase thermodynamic efficiencies, manufacturers

are continually increasing overall pressure ratios, which also increases temperatures in the engine.

3 Clark, K., Barringer, M., Johnson, D., Thole, K., Grover, E., and Robak, C., “Effects of Purge Flow Configuration on

Sealing Effectiveness in a Rotor-Stator Cavity.” To be submitted to the 2017 ASME Turbine Technical Conference and

also for the ASME Journal of Turbomachinery.

105

Increasing temperatures can lead to durability concerns for hot section components so secondary

air, which is bled from the compressor, is used to provide the cooling and sealing flows necessary

to maintain durability of the hot section components. Excessive use of the secondary air results in

a parasitic loss to the thermodynamic cycle of the engine so efficient use of the air is necessary to

maintain efficiency.

Sealing technologies, in particular, are needed to protect the cavity regions inboard of the

airfoil platform. Rim seals are located at the platform, where rotating blades and stationary vanes

meet. Rim seals are purposefully complex, using a combination of axial and radial clearances to

minimize the hot gas ingestion into the cavities inboard of the airfoil platform. Seal clearances,

however, must be large enough to allow for thermal expansion and engine transients for all gas

turbines, and also allow for high acceleration forces for engines in aircraft. Propulsive efficiencies

of aircraft engines are improving through increasing bypass ratios, which results in reduced engine

core sizes. In these engines the seal clearances do not always shrink as much as the engine cores so

these compact engine cores may exhibit increased relative clearances resulting in more ingestion

and thus a more difficult task for the secondary air system to minimize the flow while maintaining

sealing performance. To supplement rim seals, sealing flow, also referred to as purge flow, is used

to purge the cavity of hot gas. Johnson et al. [85] reported that a 50% reduction in the sealing flow

in a two stage turbine would increase the turbine efficiency by 0.5% and decrease fuel consumption

by 0.9%. There is a need to develop more advanced sealing technologies that minimize flow

requirements while maintaining sealing performance.

This paper describes the sealing effectiveness in an engine-realistic rim seal and rim cavity

at engine-relevant conditions for two distinct purge flow configurations. A review of previous

literature will be given that illustrates the uniqueness of the work presented in this paper, followed

by a description of the facility, turbine, and benchmarking. The sealing effectiveness was

determined through concentration measurements whereby CO2 was used as a tracer gas in the

secondary air supply. For the data presented in this paper, sealing effectiveness measurements were

made in the front vane-blade cavity for a 1.5 stage test turbine for two engine-realistic purge flow

configurations. The effectiveness data will also be compared to the orifice model reported by Owen

et al. [86].

106

5.2 Review of Literature

Our understanding of the complex topic of hot gas ingestion has increased over the past

several decades due to several fundamental studies. This section will briefly review some of these

fundamental studies that correspond to the influencing parameters identified by Johnson et al. [85]

including rotational effects, such as disk pumping; external effects, such as periodic vane and blade

pressure fields; and geometry effects, such as the rim seal design. Additionally, the influence of the

purge flow delivery method and the influence of operating conditions on sealing effectiveness will

be briefly discussed.

Early ingestion experiments used simplified rotor-stator cavities, where one side was a

rotating plane disk and the other side was a stationary plane disk, to study the effects of rotation.

Bayley and Owen [87] developed a simple correlation to predict the minimum flow rate required

to seal a plane rotor-stator cavity from ingress. Phadke and Owen [88] studied a similar plane rotor-

stator cavity with several rim seal geometries. Through flow visualization, they found the purge

flow was entrained in the rotor boundary layer and moved radially outward to conserve angular

momentum (disk pumping), which caused a radial inward flow on the stator. To satisfy continuity,

this flow field induced an axial flow across the cavity. They found that this disk pumping affected

the cavity flow field for the different rim seal geometries. They provided an improved correlation,

over that of Bayley and Owen [87], to predict the minimum sealing flow rate for a variety of

simplified rim seal geometries. Their data showed that the minimum flow rate required to seal the

cavity increased with increasing rotational speed, or rotational Reynolds number [88].

Ingestion has not only been shown to be affected by rotational effects, but has also been

shown to be affected by the main gas path flow, especially the vane exit pressure field. Phadke and

Owen [89] performed experiments for several rim seal geometries at a variety of main gas path

conditions. Although the study did not include airfoils, they showed that at low main gas path

velocities (related to low airfoil Reynolds numbers), ingestion was driven by rotational effects,

such as disk pumping, but at higher main gas path velocities (related to high airfoil Reynolds

numbers) ingestion was driven by the non-axisymmetric pressure difference in the main gas path.

Two main regimes were identified: rotationally-induced and externally-induced ingestion, both

with different governing physics. Hamabe and Ishida [90] showed that for externally-induced

ingestion the rim seal could be modeled as an orifice and there was good agreement for the sealing

effectiveness data using a simplified seal geometry. Chew, Green, and Turner [91] also showed the

107

presence of two regimes at different purge flow rates. Their results showed that the ingress and

egress discharge coefficients varied with the swirl flow angle in the main gas path.

The boundary conditions on the rim seal in the main gas path are not only affected by the

upstream vanes, but also by the interaction of the vane and blade pressure fields. Unsteady flow

field measurements, performed by Bohn et al. [92] in the outer portion of the rim cavity for a simple

axial seal, showed the strong influence of the passing rotor on ingestion. The potential field of the

passing rotor interacted with the vane wake flow and caused more ingestion than when the passing

rotor interacted with the vane core flow. Unsteady CFD simulations performed by Wang et al. [93]

showed that the interaction of the vane and blade pressure fields caused rotating cells in the main

gas path and in the rim cavity. These cells resulted in unsteady boundary conditions on the rim seal

that caused alternating pockets of ingestion and egress through the rim seal. Their results also

illustrated the importance of simulating a full annulus rather than just a sector.

The geometry of the rim seal has also been shown to have a substantial effect on sealing

effectiveness. The effectiveness of multiple rim seal geometries was compared by Graber et al.

[94], who showed that reducing the radial clearance of the rim seal was more effective at reducing

ingestion than increasing the axial overlap of the rim seal. Sangan et al. [62] also showed that

reducing the radial clearance of the rim seal reduced ingestion for a single overlap rim seal. Their

data from more complex rim seals, such as double rim seals, exhibited a further reduction in

ingestion for a given sealing flow thereby indicating that complex rim seal geometries are needed.

Although the literature clearly shows the importance of the rim seal geometry, most studies have

used simplified geometries. To generate data and models applicable to engines it is important to

study engine-realistic rim seal geometries.

There have only been a few studies in the literature that have investigated the influence of

the delivery method for the purge flow, which is the secondary flow devoted to sealing cavities

against hot gas ingestion. Most ingestion experiments have introduced the purge flow at the center

of rotation, but the purge flow is generally not introduced to turbine cavities in this manner. Engine

designs are complex, and often require that the purge flow be delivered in a variety of ways. Coren

et al. [75] reported that the incoming angle and the jet momentum of a radially-injected purge flow

affected the sealing effectiveness and the cavity flow physics in a blade-vane cavity. When the

purge flow was directed upstream toward the rotor, the flow was more likely to be entrained into

the disk pumping flow. This entrainment phenomenon increased the effectiveness by as much as

50% as a direct result of the purge flow delivery system. Companion numerical studies by Andreini

108

et al. [76] showed that increasing the momentum of the purge flow affected the robustness of the

purge flow delivery.

Hot gas ingestion is a complex topic, and fundamental studies have provided useful

information that has enhanced our understanding of the cavity flow physics as well as provided the

basis for several ingestion models. Many of these studies, however, have been performed at

simplified conditions with simplified geometries. Teuber et al. [95] showed that higher external

Mach numbers resulted in more ingestion. In regards to the topic of hot gas ingestion, Green and

Turner [96] stated that “oversimplified experimental rigs operated far from engine conditions may

often only serve to confuse the issue…. Before design rules can be established with confidence, all

the influencing parameters should be examined together at conditions as close to modern engine

operating levels as possible.”

There is a need to study all the influencing parameters at engine-relevant conditions to

allow for the development and validation of ingestion models. This paper is unique in that the

effectiveness for two engine-realistic purge flow configurations is presented for an engine-realistic

rim seal operated at engine-relevant Mach and Reynolds numbers. The effectiveness data are then

compared to an orifice model.

5.3 Description of Facility and Turbine

The experiments presented in this paper were performed in the Steady Thermal Aero

Research Turbine (START) facility, the design of which was previously described by Barringer et

al. [67]. Previous papers by Clark et al. [70,80] also described the facility commissioning, the half-

stage turbine configuration, and instrumentation. In this section the facility will be briefly reviewed,

and the 1.5 stage turbine used for these experiments will be described in more detail.

Facility

The START facility, which houses a 1.5 stage test turbine, is an open, continuous flow

loop, as shown in Figure 5.1. Ambient air enters a compressor requiring 1.1 MW (1500 hp) of

power that provides up to 5.7 kg/s (12.5 lbm/s) of air flow at 480 kPa (70 psia) to the turbine. A

portion of that flow is used to provide the turbine secondary air. A heat exchanger and chiller

outside the lab cools the compressor and turbine secondary air respectively. Air from the

109

compressor flows through a supply pipe to an upstream settling chamber, where it is then directed

to the turbine test section. The turbine exit air enters a downstream settling chamber after which it

exits the facility. Venturi flow meters that are fully calibrated by a NIST certified laboratory [55],

provide the turbine inlet and exit flow rates. Operating conditions in the test section are controlled

by three flow control valves that independently control the turbine inlet pressure and the pressure

ratio to within ±1.4 kPa (±0.2 psi) and ±0.002 respectively.

The turbine rotor speed, torque, and power are controlled by a water brake dynamometer.

The water brake provides braking torque by flowing pressurized water through rotating perforated

disks connected to the turbine shaft. A simplified diagram of the water system is shown in the top

right of Figure 5.1. The dynamometer inlet and exit valves are hydraulically actuated and the set

point is maintained by a controller that is provided and tuned by the dynamometer vendor. At a

typical operating speed the water brake dynamometer is shown to hold the turbine rotor speed

within a standard deviation of ±0.2% of the mean speed.

The turbine rotor is supported by two radial magnetic bearings through an electromagnetic

field. Auxiliary bearings are also available to support the rotor when the magnetic bearings are not

active or to catch the rotor in the case of a magnetic bearing failure. Sensors track the radial orbits

of the shaft, and a controller maintained the magnetic bearing parameters to ensure safe and stable

operation. The shaft centerline is held within ±2.5 µm (±0.0001 in) of the true centerline by the

magnetic bearings during normal operation.

An electromagnetic thrust bearing is used to maintain the nominal axial position of the

rotor, but the thrust from the turbine could exceed the 6.7 kN (1500 lbf) capacity of the thrust

bearing. A two-stage pneumatic thrust piston system is incorporated into the rotor to provide an

additional 8.9 kN (2000 lbf) of counter thrust. The thrust piston operates as pressurized air on the

aft side (high pressure side) of the pistons flows to the exhaust cavities (low pressure side) across

commercial brush seals. The pressure difference across the two thrust pistons provides sufficient

counterthrust in addition to the magnetic thrust bearing for a wide operating range of the turbine.

110

Figure 5.1. START facility layout, which houses the 1.5 stage turbine.

Test Turbine

The test section included a 1.5 stage turbine with a vane-blade-vane configuration as shown

in Figure 5.2. Similar to the research of Clark et al. [70,80], partial span airfoils were used for this

study as well to reduce the mass flow requirements while maintaining an engine-relevant axial

Reynolds number, rotational Reynolds number, and Mach number. Other researchers have also

proven the use of partial span airfoils for studying rim seals and ingestion [60,68,81].

The airfoils, disk, rim seals and cavities, and secondary flow supplies in this test turbine

used modern gas turbine hardware. The first and second vanes were additively manufactured from

a nickel alloy in doublets, or pairs. The blades were solid single crystal castings attached to the disk

through individual fir tree slots. There were also inter-blade gaps and seals as in an operational gas

turbine. Two cover plates with labyrinth seals, shown in Figure 5.2, were installed on the front and

Settling Chamber

Venturi

Turbine Cooling:Heat Exchanger

Coolant Pipes

TURBINE

Compressor Cooling SystemOutdoor Heat Exchanger

(800 kW)

Turbine Cooling SystemOutdoor Chiller

(200 kW)

Mo

tor

Star

ter

CoolingTower

Overhead Crane

Pump System

Flow Meters

Building Back Wall

Roof Exhaust

Turbine Cooling Air

RoofIntake

COMPRESSORS

Motor

Building Exterior Wall

Motor

By-Pass

PLC

HotWell

Dynamometer Water System

Pump Chamber

ColdWell

Mo

tor

Star

ter

Controls+ DAQ System

COMPRESSOR ROOM (14m x 11m)

CONTROL ROOM(7.5m x 3m)

TEST BAY ROOM(14m x 12m)

Tank

HydraulicPump for

Dyno ValvesOil

H2O Treatment

H2O Pre-Treatment

DynoH2O FillMakeup

TP air standOil cooler

Intercooler

111

aft sides of the disk to axially secure the blades and provide an engine representative front wheel-

space, front rim cavity, and aft rim cavity.

Two main secondary air supplies were available on the front side of the disk: (1) purge

flow from the vane plenum and (2) tangential on-board injection (TOBI) flow as shown in Figure

5.2. The purge flow was provided directly to the front rim cavity from the vane plenum in two

different configurations: 150 axially oriented holes or 32 axially oriented holes. In an operational

gas turbine, a portion of the TOBI flow, which is a pre-swirled air flow injected inboard of the front

wheel-space, would pass through the front cover plate and provide the blade cooling flow. The

purpose of the test program was to investigate cause and effect relationships for engine-relevant

purge configurations, and this paper separates the effects. As the focus of this paper is on the effects

of the purge flow in the rim cavity, the TOBI flow was not used for these experiments. Results will

be presented in the future for TOBI flow.

Figure 5.2. 1.5 stage turbine cross-section: (a) first vane plenum, (b) front rim seal, (c) front

rim cavity, (d) front wheel-space, (e) purge flow, (f) TOBI flow, and (g) aft rim cavity.

The individual blades, gaps, and slots allowed for leakage through the fir tree gaps of the

blades as well as from the front rim cavity to the aft rim cavity through the gaps between the blades

Main gas path(MGP)

First Vane(1V)

Second Vane(2V)

Blade(B)

ae

d

c

b

g

f

112

as in an operational gas turbine. Since the TOBI flow was not used for these experiments a small

portion of the flow (approximately 5% of the minimum flow required to seal the front cavity) in

the front rim cavity passed across the front labyrinth seal into the front wheel-space, through holes

in the front cover plate, and across the disk through the blade fir tree gaps. A larger portion of the

flow, nominally around 15-20% of the minimum flow required to seal the front rim cavity, passed

from the front rim cavity to the aft rim cavity through the gaps between the blades. Additional

leakage flow was introduced into the aft rim cavity across the labyrinth seal from the thrust piston

for the magnetic bearing cooling. The leakage flow from the thrust piston into the aft cavity

accounted for approximately 35-40% of the minimum flow required to seal the front rim cavity.

Instrumentation and Uncertainty

The turbine test section was heavily instrumented as shown in Figure 5.3 and previously

described by Clark et al. [70,80]. Several static pressure taps, pressure probes, and thermocouples

composed the instrumentation in the turbine. The first and second vanes were additively

manufactured, which allowed for integrated static pressure taps on the airfoil surfaces and near the

rim seal of the first vane. These taps are indicated in Figure 5.3 as “through AM”.

In addition to pressure measurements, the static pressure taps were used for concentration

effectiveness measurements. According to the method described in detail by Clark et al. [70,80],

the secondary air supply was seeded with 1% CO2 to be used as a tracer gas in the turbine. Flow

was sampled at specific locations in the front rim seal, rim cavity, and wheel space through the

static pressure taps. The CO2 concentration of the sampled flow was measured by a gas analyzer.

By measuring the background concentration, 𝑐∞, the supply concentration, 𝑐𝑠, and the

concentration at the location of interest, 𝑐, an effectiveness based on concentration was determined

according to the following definition

𝜀𝑐

(𝑐 − 𝑐∞)

(𝑐𝑠 − 𝑐∞) (5.1)

When making gas sampling measurements it is important to perform sampling sensitivity

studies to ensure that the sampling method does not affect the flow field. The study by Clark et al.

[80] showed that concentration gradients were present where the purge flow interacted with the

ingested flow in both the rim seal and cavity, and that achieving isokinetic sampling was crucial to

113

obtaining reliable effectiveness measurements. Although data is not shown in this paper for the

sake of brevity, sampling sensitivity studies at various purge flow rates for the 1.5 stage turbine

were performed according to the method described by Clark et al. [80] to ensure accurate

concentration effectiveness measurements were obtained. Concentration effectiveness

measurements will be presented in this paper for the locations shown in Figure 5.3, namely (A) the

front rim seal, (B) the outer radius of the front rim cavity, (C) the purge hole radius of the front rim

cavity, and (D) the front wheel-space.

Figure 5.3. Turbine cross-section with instrumentation locations. Effectiveness data will be

presented for the following locations: (A) front rim seal, (B) outer radius of front rim cavity,

(C) purge hole radius of front rim cavity, and (D) front wheel-space.

An uncertainty analysis was performed for the facility and turbine measurements reported

in this paper according to the method of Figliola and Beasley [54]. The total uncertainties for these

measurements are reported in Table 5.1, which included both the bias and precision uncertainties

for each measurement. The main measurement reported in this paper is concentration effectiveness.

The bias uncertainty was minimized by using two concentration ranges and calibration gases on

Kiel pressure probeStatic pressure tapTotal temperature

Instrumentation type

1V airfoil surface taps (through AM)Kiel temperature probesKiel pressure probesFront rim seal taps (through AM)Front rim cavity taps1V plenum TC’s and probesFront wheel-space TC’s and tapsTOBI supply TC’s and probes2V airfoil surface taps (through AM)Aft cavity taps

A

B

D

C

114

the gas analyzer. The precision uncertainty was minimized by taking a 60 second average of the

signal from the gas analyzer (approximately 40k samples). This method resulted in a total

uncertainty in concentration effectiveness of ±0.015 to ±0.02 over the entire range.

Operating Conditions

Effectiveness measurements will be presented in this paper for two purge hole

configurations: (1) 150 purge holes and (2) 32 purge holes. The 150 purge hole configuration was

tested up to a flow rate of 𝛷/𝛷𝑟𝑒𝑓 ≤ 1.5, where 𝛷𝑟𝑒𝑓 was the reference flow rate defined as the

flow rate at which the front rim cavity (locations B and C in Figure 5.3) was fully sealed (𝜀𝑐 ≥

0.99). The 32 purge hole configuration was tested up to 𝛷/𝛷𝑟𝑒𝑓 ≤ 0.6 with the same reference

flow rate as in the case with 150 purge holes so relative comparisons between the two cases can be

made given only one scaling number. The experiments presented in this paper were operated at a

blade inlet relative Mach number of 0.2, a blade inlet axial Reynolds number of 1.4x105, and a

rotational Reynolds number of 3.8x106 as defined in the nomenclature.

5.4 Facility and Turbine Benchmarking

This section briefly describes the benchmarking that was performed for the 1.5 stage

turbine, including the turbine operating conditions, the first and second vane aerodynamic loadings,

Table 5.1. Uncertainty in Facility and Turbine Measurements

Parameter Total Uncertainty

Main gas path flow rate, �̇� �̇�𝑟𝑒𝑓⁄ ±0.004 to ±0.006

Shaft rotational speed, Ω Ω𝑟𝑒𝑓⁄ ±0.002

Pressures, p p𝑟𝑒𝑓⁄ ±0.005

Temperatures, T ±0.5 K

1.5 stage pressure ratio, PR PR𝑟𝑒𝑓⁄ ±0.006

Purge flow rate, 𝛷/𝛷𝑟𝑒𝑓 ±0.012

Concentration effectiveness, 𝜀𝑐 ±0.015 to ±0.02

115

and the front cavity boundary conditions. Clark et al. [70,80] previously described benchmarking

for the facility for the half stage turbine.

Facility Operation

The facility exhibited stable and repeatable steady state operation. During a typical test

day, the facility was operated for 8-12 hours at various conditions. Figure 5.4 shows the facility

flow conditions for a 2.5 hour window of a typical test. The turbine inlet mass flow rate, shaft

rotational speed, turbine inlet total pressure, turbine inlet total temperature, and the 1.5 stage

pressure ratio shown in Figure 5.4 were all scaled by reference values. All of these parameters were

held constant for this test point. Figure 5.4 shows that the facility, including the compressor, flow

control valves, and dynamometer, exhibited long term steady state operation. For the experiments

presented in this paper the main gas path conditions, including 1.5 stage pressure ratio and corrected

speed, were all held constant.

Figure 5.4. Conditions in the turbine for a typical test.

0.95

0.96

0.97

0.98

0.99

1.00

1.01

1.02

1.03

1.04

1.05

12:50 13:20 13:50 14:20 14:50 15:20

ṁ/ṁref, Ω/Ωref, p/pref, T/Tref,

PR/PRref

Time

ṁin/ṁref

Ω/Ωref

pt,in/pref

Tt,in/Tref

PR/PRref

116

Although effectiveness measurements are not presented in this paper for the aft cavity, it

was important to maintain the aft cavity flow conditions constant for all of the experiments to ensure

consistent operating conditions. The front cavity purge flow rate was set to various conditions

throughout testing, and the turbine inlet flow rate was then set to match the desired 1.5 stage

pressure ratio. The aft cavity flow rate was determined from the difference between the turbine exit

mass flow rate and the turbine inlet plus the front cavity purge flow. Several leakage flows, which

will be discussed in more detail later, entered the aft cavity as in an operational engine. As discussed

previously, the magnetic bearing cooling air also entered the aft cavity. These flows were accounted

for, and the aft cavity egress flow rate was held constant at 0.55 < 𝛷/𝛷𝑟𝑒𝑓 < 0.65 for all test

conditions presented in this paper.

Turbine Operation

Turbine measurements indicated both repeatability and uniformity in the test section. The

first vane aerodynamic loading at 50% span is shown in Figure 5.5a, and the second vane

aerodynamic loading at 50% span is shown in Figure 5.5b. The static pressures on the first vane

airfoil surface in Figure 5.5a were normalized by the turbine inlet total pressure, and the static

pressures on the second vanes in Figure 5.5b were normalized by the second vane inlet total

pressure as measured at the stagnation point on the second vane leading edge. The solid line

represents pre-test prediction CFD simulations. The different colored symbols represent

measurements from different test days and the different symbols represent measurements obtained

on separate vanes.

The agreement between the measurements obtained on different vanes for a given test day

showed circumferential uniformity for both the first and second vanes. The measurements from

different test days indicated good repeatability. Although not shown here, measurements at 10%

and 90% span on both vanes also indicated uniformity and repeatability. The agreement between

the CFD and the measurements on the first vanes indicated the flow through the first vanes operated

as intended, and that the additively manufactured first vanes were built to specifications. The

agreement of the measurements for the different test days, different vanes, and with the CFD shown

in Figure 5.5a for the first vanes was similar to that shown for the half-stage (vane only) turbine as

described by Clark et al. [70]. The pre-test prediction CFD and data did not agree as well on the

second vanes, but the agreement was sufficient to instill confidence that the second vanes were

built to specifications, thus allowing the researchers to proceed with the effectiveness experiments.

117

Figure 5.5. Static pressure normalized by the vane inlet total pressure at 50% span for (a)

first vane and (b) second vane.

The boundary conditions measured at the rim seal also indicated periodicity in the turbine.

Figure 5.6 shows the static pressures measured on the vane trailing edge at the endwall, on the

platform trailing edge, and in the rim seal normalized by the turbine inlet total pressure. The

different colored symbols correspond to the measurement locations shown in the inset image. The

pressure taps that were used to obtain these measurements were located on two vanes at different

circumferential locations as shown by the different symbols types. The static pressure

measurements on different vanes agreed with each other indicating periodicity. Similar to the same

measurements found for the vane only configuration [70], the vane exit pressure distortion was

observed. The local peaks in static pressure highlighted the vane wake region, and the minima

corresponded to the vane core flow. The difference in the static pressure in the vane wake and the

vane core flow was 𝑝/𝑝𝑡,𝑖𝑛 0.07 for the measurements at the vane trailing edge. At the trailing

edge of the platform the pressure difference was strongly attenuated to 𝑝/𝑝𝑡,𝑖𝑛 0.025 and in the

rim seal the difference was completely attenuated.

0.5

0.6

0.7

0.8

0.9

1.0

0 0.5 1

ppt

Percent wetted distance, S/Smax

0 0.5 1

Percent wetted distance, S/Smax

Key: CFD Test 1Vane AVane B

Test 2Vane AVane B

(a) (b)

118

The measurements shown in Figure 5.6 show that the same time-averaged boundary

conditions that existed on the rim seal for the vane only configuration described by Clark et al. [70]

were also present for the 1.5 stage configuration presented in this paper. To study ingestion for this

1.5 stage configuration several additional effects were included in addition to the vane exit pressure

field included in the vane only study: rotational effects such as disk pumping, the potential field

upstream of the blade, and the unsteady interactions between the vane exit pressure and blade

potential field.

Figure 5.6. Static pressure normalized by the inlet total pressure on the vane trailing edge,

the platform trailing edge (22% 1V axial chord downstream of vane trailing edge), and in the

rim seal (8% 1V axial chord upstream of vane trailing edge).

5.5 Sealing Effectiveness for Purge Flow

Concentration effectiveness measurements in the front wheel-space, front rim cavity, and

front rim seal are presented in this section for two purge flow configurations: (1) 150 purge holes

and (2) 32 purge holes. The effectiveness data are presented versus the scaled purge flow rate

𝛷/𝛷𝑟𝑒𝑓, where 𝛷 is the purge flow rate and 𝛷𝑟𝑒𝑓 is the reference sealing flow rate defined as the

0.60

0.65

0.70

0.75

0.80

5 10 15 20 25 30

ppt,in

θ [ ]

Vane 7 Vane 23Vane TE:Platform TE:Rim Seal:

119

purge flow rate at which the rim cavity was fully sealed for the 150 purge hole configuration (see

locations B and C in Figure 5.3). Thus by definition, the rim cavity effectiveness of unity

corresponds to a purge flow rate of 𝛷/𝛷𝑟𝑒𝑓 1. The variation in concentration effectiveness with

purge flow rate and location is shown for both 150 purge holes and 32 purge holes in Figure 5.7.

The data shown for both configurations in Figure 5.7 were measured at several circumferential

positions and averaged circumferentially for each radial location in Figure 5.7. Although the data

for 150 purge holes exhibited negligible circumferential variation, the 32 purge holes showed a

periodic variation in effectiveness, which will be discussed at the end of this section. Figures 5.8

and 5.9 are also presented for both configurations to provide a physical understanding of the data

being presented.

Configuration 1: 150 Purge Holes

In this section we will discuss the measurements in Figure 5.7 for 150 purge holes. The

measurements for 32 purge holes will be discussed in the next section. The effectiveness

measurements for 150 purge holes in Figure 5.7 showed that effectiveness increased with

decreasing radius and with increasing purge flow. The highest effectiveness was observed in the

front wheel-space (location D), which was fully sealed for 𝛷/𝛷𝑟𝑒𝑓 0.9, or 10% less than the

reference flow rate. Farther outboard in the rim cavity (locations B and C), the effectiveness was

slightly lower as more ingestion occurred for a given flow rate than for the front wheel-space. Both

locations in the rim cavity were fully sealed for a purge flow rate of 𝛷/𝛷𝑟𝑒𝑓 1 according to the

definition of 𝛷/𝛷𝑟𝑒𝑓 . The lowest effectiveness was observed near the main gas path in the rim seal

(location A), which was fully sealed for 𝛷/𝛷𝑟𝑒𝑓 1.5. Due to the high ingestion rates from the

main gas path, 50% more flow was required to fully seal the rim seal than the rim cavity. Each of

these locations will be discussed in more detail in this section starting with location D and moving

radially outward to location A.

A major result was that there was appreciable ingestion in the front wheel-space (location

D in Figure 5.7) for the 150 hole configuration at low purge flows. The front wheel-space was

isolated from the rim cavity by a two-stage labyrinth seal. Despite the small clearances on the

labyrinth seal, there was still appreciable ingestion that occurred over a wide range of purge flow

rates. For example, at 𝛷/𝛷𝑟𝑒𝑓 0.1 the effectiveness was 𝜀𝑐 0.7, and at 𝛷/𝛷𝑟𝑒𝑓 0.5 the

effectiveness was 𝜀𝑐 0.9. Although these effectiveness values were relatively high values

compared to the other data in Figure 5.7, it was expected that the front wheel-space should be fully

120

purged. For location D, it was not until a purge flow of 𝛷/𝛷𝑟𝑒𝑓 0.9 that the front wheel-space

was fully sealed. If ingestion occurred deep within the turbine wheel-spaces in an engine, then the

disk would heat up with potentially catastrophic effects; hence it is of the utmost importance to

fully seal the front wheel-space.

To explain why there was appreciable ingestion in the front wheel-space (location D in

Figure 5.7) we shall examine the flow schematic in Figures 5.8 and 5.9. The test turbine was an

engine-realistic design, so a leakage across the disk through the blade fir trees, as indicated in Figure

5.8, was present with a nominal flow rate of 𝛷/𝛷𝑟𝑒𝑓 0.05. The fir tree leakage was fed from the

front wheel-space (location D), which was in turn fed from the rim cavity (locations B and C)

through the labyrinth seal as indicated by the arrow in Figure 5.8. This flow pattern allowed the

flow ingested into the rim cavity (locations B and C) to be ingested farther inboard to the front

wheel-space (location D).

The deep penetration of the hot gas into the cavity (location D) has major implications for

the engine. These effectiveness measurements in Figure 5.7 for the front wheel-space highlight the

need to provide flow directly to the front wheel-space. The tangential on-board injection (TOBI)

flow, as shown in Figure 5.2, is designed to pressurize the front wheel-space and cause the leakage

across the labyrinth seal to flow in the opposite direction indicated in Figure 5.8 into the rim cavity,

thereby minimizing the ingestion inboard of the seal. Results will be presented in the future that

include the TOBI flow. If the seal clearances in an engine were too large due to excessive wear or

a poor design, then the pressure in the wheel-space may not build sufficiently even with the TOBI

flow and ingestion past the labyrinth seal could occur. These measurements show the importance

of (1) accurately setting the labyrinth seal clearances, (2) maintaining the engines, and (3) providing

sufficient TOBI flow, otherwise ingestion could occur deep within the turbine wheel-spaces with

potentially catastrophic effects.

The effectiveness in the front wheel-space (location D in Figure 5.7) and at the purge hole

radius in the rim cavity (location C) exhibited behavior that deviated from previously measured

effectiveness curves [31,60,62], like that shown for the outer radius in the rim cavity (location B).

The data shown for location B were similar to previous effectiveness curves for which the orifice

model was derived, and it will be shown later that the model fit the data well. The effectiveness

data for locations C and D were quite different from location B, and it will be shown that the orifice

model did not fit the data for locations C and D. Compared to location B, the effectiveness for

locations C and D increased sharply at low flow rates for 𝛷/𝛷𝑟𝑒𝑓 < 0.2. At 𝛷/𝛷𝑟𝑒𝑓 0.2 the

121

effectiveness leveled off, then began to gradually increase again to fully sealed conditions.

Although both locations C and D exhibited similar qualitative behavior with a sharp increase in

effectiveness for low flow rates, the effectiveness for location C was slightly lower than location

D. For example, at a low flow rate of 𝛷/𝛷𝑟𝑒𝑓 0.2, effectiveness was 𝜀𝑐 0.76 at location C and

𝜀𝑐 0.83 at location D. As the purge flow rate increased, effectiveness for location C gradually

increased to unity.

Figure 5.7. Sealing effectiveness measurements for Configuration 1: 150 purge holes and

Configuration 2: 32 purge holes.

The sharp bend in the effectiveness curve for locations C and D at 𝛷/𝛷𝑟𝑒𝑓 0.2 was

similar to that observed by Clark et al. [70] for the case with no rotation, and was attributed to the

behavior of a jet-in-crossflow. Recall that the purge flow entered the rim cavity through axially-

oriented holes normal to the cavity swirling flow. At low purge flow rates, the jet momentum was

low, which caused more of the purge flow to diffuse near the stator side causing the sharp increase

in effectiveness at the low flow rates, as shown in the flow schematic in Figure 5.9a at low flow

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6

εc

Φ/Φref

150 purge holes

32 purge holes

A

B

D

C

122

rates. At higher flow rates the jet momentum was higher, which caused the purge flow to be

entrained in the axial flow across the cavity to the rotor side, as shown in Figure 5.9b at high flow

rates. Note that as the purge flow rate increased, the effectiveness at the purge hole radius (location

C) approached that at the outer radius (location B). As the purge flow rate increased beyond

𝛷/𝛷𝑟𝑒𝑓 > 0.55 the purge jets were likely separated from the stator side of the rim cavity. The

purge flow was entrained axially across the rim cavity, as shown in Figure 5.9b. The purge flow

was then pumped radially outward on the rotor, and then fed the recirculating flow back to the

stator side resulting in similar effectiveness at both locations B and C.

The effectiveness at the outer radius in the rim cavity (location B in Figure 5.7) was lower

than both the purge hole radius (location C) and the wheel-space (location D). The effectiveness

measurements at location B displayed a characteristic curve that was similar to those measured by

previous researchers [31,60,62], with a monotonic increase in effectiveness as flow rate increased

and an exponential asymptote to an effectiveness of unity. As will be shown in the next section, the

effectiveness data at the outer radius in the rim cavity was well suited to empirical modeling.

In the rim seal (location A in Figure 5.7) the effectiveness was zero for 𝛷/𝛷𝑟𝑒𝑓 ≤ 0.2.

Despite providing 20% of the flow required to seal the rim cavity, there was no appreciable effect

in the rim seal. The flow schematic of the secondary flows, given in Figure 5.8, will be used to

discuss this phenomenon. Figure 5.8 shows a leak across the disk from the front rim cavity to the

aft rim cavity. This particular flow passed through the gaps between the blades. The reason for the

zero effectiveness in the rim seal was because a significant amount of purge flow, approximately

20% of 𝛷𝑟𝑒𝑓, was lost from the rim cavity to feed the blade gap leakage and did not reach the rim

seal. Once the blade gap leakage was satisfied by the purge flow in the front rim cavity for

𝛷/𝛷𝑟𝑒𝑓 ≈ 0.2, then the effectiveness in the rim seal increased with purge flow rate. As the purge

flow rate increased, the effectiveness in the rim seal increased in a nearly linear fashion up to the

fully sealed condition. Even though the rim cavity (locations B and C) was fully sealed at

𝛷/𝛷𝑟𝑒𝑓 1.0, the effectiveness in the rim seal (location A) at the same flow rate was only 𝜀𝑐

0.8, and for fully sealed conditions the rim seal required 50% more flow than the rim cavity.

123

Figure 5.8. Flow schematic of secondary flows in the 1.5 stage test turbine.

A

B

D

C

Fir tree leakage

Thrust piston

leakage

Blade gap leakage

124

Figure 5.9. Schematic of flows in the rim cavity for the following configurations: (a) 150 purge

holes at low flow rates, (b) 150 purge holes at high flow rates, (c) 32 purge holes at low flow

rates, and (d) 32 purge holes at high flow rates.

A

B

C

(a) 150 purge holes at low flow rates

A

B

C

(c) 32 purge holes at low flow rates

A

B

C

(b) 150 purge holes at high flow rates

A

B

C

(d) 32 purge holes at high flow rates

125

Configuration 2: 32 Purge Holes

In this section we will discuss the effectiveness measurements for 32 purge holes that are

also presented in Figure 5.7. The same 𝛷𝑟𝑒𝑓 was used to scale the data for 32 purge holes as 150

purge holes to provide a direct comparison between both configurations. As can be seen from the

data, the 32 purge holes did not provide enough flow to fully purge the wheel-space, rim cavity, or

rim seal for the flow rate range over which the tests were conducted. Effectiveness measurements

for 32 purge holes were obtained for a purge flow rate of 𝛷/𝛷𝑟𝑒𝑓 < 0.6, and beyond that flow rate

the pressure ratio across the purge holes was higher than could be expected in an engine.

The effectiveness measurements for 32 purge holes in Figure 5.7 again showed that

effectiveness increased with decreasing radius and with increasing purge flow. In the front wheel-

space (location D) the effectiveness was higher than the other locations for all flow rates. The

effectiveness also showed a sharp increase in effectiveness at the low flow rates, followed by a

more gradual increase in effectiveness as the purge flow rate increased. At the purge hole radius

(location C) the effectiveness also exhibited an increase in effectiveness at lower flow rates due to

the jet-in-crossflow behavior. The effectiveness at the purge flow radius (location C) approached

the effectiveness at the outer radius (location B) as purge flow rate increased. At the outer radius

in the rim cavity (location B) the effectiveness characteristic was again similar to those measured

by previous researchers [31,60,62]. In the rim seal (location A), the effectiveness showed zero

effectiveness for 𝛷/𝛷𝑟𝑒𝑓 ≤ 0.16, but for 𝛷/𝛷𝑟𝑒𝑓 > 0.16 the effectiveness increased with flow

rate.

Circumferential Variation in Effectiveness

The effectiveness in the rim cavity was found to be circumferentially uniform for 150 purge

holes, so effectiveness with circumferential position for 150 purge holes is not shown here for the

sake of brevity. The circumferential uniformity was observed at both the purge hole radius (location

C) and at the outer radius (location B) of the rim cavity. For 150 purge holes, the distance between

the holes was circumferentially spaced less than 4D, so the purge flow entered the rim cavity in a

very uniform manner. The circumferentially uniform purge flow in the rim cavity was similar to

that previously observed [70].

The circumferential spacing between 32 purge holes was higher than for 150 purge holes,

at approximately 16D, which resulted in a circumferential variation in effectiveness in the rim

126

cavity at the purge hole radius. The effectiveness in the front rim cavity as a function of

circumferential position is shown in Figure 5.10 for four purge flow rates for the 32 purge hole

configuration. As indicated in the image, the purge holes were located at θ = 4° and 15°, and the

swirl flow produced by the vane was from left to right. At the lowest flow rate of 𝛷/𝛷𝑟𝑒𝑓 0.05

the effectiveness at the purge hole radius (location C) varied greatly with circumferential location.

At θ = 13° the effectiveness was at 𝜀𝑐 0.25, but just downstream of the purge hole at θ = 18° the

effectiveness increased by 200% to 𝜀𝑐 0.75 This increase was followed by a decay in

effectiveness to 𝜀𝑐 0.3 at θ = 23°. For the same flow rate of 𝛷/𝛷𝑟𝑒𝑓 0.05 the effectiveness at

the outer radius (location B) was mostly constant, with a slight increase in effectiveness observed

from 13 to 23°. For a slightly higher purge flow rate of 𝛷/𝛷𝑟𝑒𝑓 0.1 the effectiveness at location

C exhibited a similar trend as at the lowest flow rate, but at a reduced level. Effectiveness at θ =

13° was 𝜀𝑐 0.4, followed by a 50% increase to 𝜀𝑐 0.6 at θ = 18° just downstream of the purge

holes. Farther downstream at θ = 23° the effectiveness again decayed to 𝜀𝑐 0.45, which was

again slightly higher than what was measured at θ = 13°. At location B the effectiveness remained

constant near 𝜀𝑐 0.3 for 𝛷/𝛷𝑟𝑒𝑓 0.1. For a higher purge flow rate of 𝛷/𝛷𝑟𝑒𝑓 0.3 the

effectiveness at locations B and C were a constant 𝜀𝑐 0.6 for all circumferential locations.

Similarly at 𝛷/𝛷𝑟𝑒𝑓 0.5 a constant effectiveness of approximately 𝜀𝑐 0.75 was observed for

all circumferential locations.

The effectiveness trends versus circumferential position observed for 32 purge holes at the

purge hole radius (location C) were consistent with the behavior of a jet-in-crossflow. At low flow

rates, or low jet momentum, a jet-in-crossflow would exhibit higher effectiveness just downstream

of the hole with a decay in effectiveness with increasing distance from the hole. This same behavior

was observed by Clark et al. [70] at the purge hole radius for 16 purge holes in a stationary rim

cavity study. At low purge flow momentum, the effectiveness data was shown to increase

dramatically downstream of the purge hole, but at high momentum the effectiveness was mostly

uniform.

The jet-in-crossflow behavior in the data shown in Figure 5.10 did not affect the outer

radius in the front rim cavity (location B). The effectiveness was mostly circumferentially uniform,

which indicated that the flow was pumping radially inward on the stator side as indicated in Figures

5.9c and 5.9d. If the flow on the stator side were pumping radially outward, then the same trends

observed at the purge hole radius (location C) could be expected at the outer radius (location B),

but this was not the case here.

127

Figure 5.10. Circumferential variation in concentration effectiveness for 32 purge holes.

Comparison of Configurations 1 and 2

At the purge hole radius (location C) and in the front wheel-space (location D) the

effectiveness was considerably affected by the number of purge holes. The effectiveness was higher

for 150 purge holes than 32 purge holes for locations C and D over the entire flow range, but

exhibited the largest difference for 0.1 ≤ 𝛷/𝛷𝑟𝑒𝑓 ≤ 0.4. At location C the effectiveness was

between 𝜀𝑐 0.1 and 0.25 greater for 150 purge holes than for 32 purge holes, and at location D

the effectiveness increase was slightly less at 𝜀𝑐 0.1 and 0.18 greater. Clearly the number of

purge holes had an effect on the sealing effectiveness at and inboard of the purge holes, with 150

holes producing higher effectiveness than 32 holes for a given purge flow rate.

As previously discussed, Figures 5.9a and 5.9b show a flow schematic in the front rim seal

and cavity for 150 purge holes at low and high flow rates respectively, and Figures 5.9c and 5.9d

show the schematic for 32 purge holes at the same flow rates. These schematics help to explain

why 150 purge holes displayed higher effectiveness at and inboard of the purge hole radius than 32

purge holes. The arrows indicate the bulk flow directions, and the colors of the arrows represent

Pu

rge

ho

le lo

cati

on

Pu

rge

ho

le lo

cati

on

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

0 5 10 15 20 25

εc

θ [°]

/ 𝒆 = 0.05

/ 𝒆 = 0.1

/ 𝒆 = 0.3

/ 𝒆 = 0.5

Swirl

BC

BC

128

the relative effectiveness of that flow with red indicating ingested flow and blue indicating purge

flow. The radial inward pumping on the stator side pulled ingested flow into the rim cavity from

the rim seal at low flow rates for both cases, as shown in Figures 5.9a and 5.9c. For 32 purge holes

there was significant spacing between the purge holes (~16D) for the pumped flow to penetrate

between the holes farther into the rim cavity, thus lowering the concentration effectiveness, as

shown in Figure 5.9c. This lower concentration air also fed the leak across the labyrinth seal to the

front wheel space, as shown in Figure 5.8, leading to lower effectiveness there. The holes were

much more closely spaced for the 150 purge holes (less than 4D), which entrained the radially

inward pumped flow from the stator into the axial flow across the cavity to the rotor, as shown in

Figures 5.9a and 5.9b. The closely spaced purge holes prevented more ingested flow from being

pumped inward past the purge holes, which allowed more of the purge flow to be pumped radially

inward, as shown in Figures 5.9a and 5.9b. The purge flow fed the region inboard of the 150 purge

holes, as shown in Figure 5.8, which also fed the labyrinth seal leakage, thus increasing

effectiveness in the wheel-space (location D). For 32 purge holes the region inboard of the purge

holes was fed by less purge flow and more ingested flow at low flow rates, as shown in Figure 5.9c,

which resulted in lower effectiveness at location D. At high flow rates the stator side flow

penetrated inward past the purge holes, but there was higher effectiveness than at low flow rates

because the purge flow that moved radially outward on the rotor was recirculated back onto the

stator side, as shown in Figure 5.9d.

At the outer radius in the rim cavity (location B in Figure 5.7) the effectiveness

measurements for both configurations were very similar, indicating that the number of purge holes

did not affect the concentration effectiveness measurements at this location. Likewise, in the rim

seal (location A) the number of purge holes had a negligible effect on the concentration

effectiveness measurements. It can thus be concluded that the number of purge holes had a minimal

effect on the sealing effectiveness on the stator side of the cavity outboard of the purge holes.

The flow field at the outer radius of the rim cavity is also shown schematically in Figure

5.9 for both low and high flow rates for both configurations. The purge flow entered the rim cavity

through axially oriented holes and was entrained axially across the cavity to the rotor, which fed

the rotor boundary layer. Part of this rotor boundary layer flow was lost through the blade gap

leakage before entering the rim seal as shown in Figure 5.9. At low flow rates, most of the flow on

the rotor was lost through the blade gap leakage, as shown in Figures 5.9a and 5.9c, and no purge

flow ended up in the rim seal. At high flow rates, a portion of the flow on the rotor passed through

129

the blade gap, but most of the flow recirculated back into the cavity or egressed into the rim seal,

as shown in Figures 5.9b and 5.9d. The flow rate at which the effectiveness in the rim seal crossed

zero was slightly different between both configurations, with 150 purge holes crossing at

𝛷/𝛷𝑟𝑒𝑓 0.2 and 32 purge holes crossing at 𝛷/𝛷𝑟𝑒𝑓 0.16. Since the uncertainty in 𝛷/𝛷𝑟𝑒𝑓

was ±0.012 the difference was very small between both configurations.

5.6 Empirical Modeling

Theoretical models for hot gas ingestion have traditionally been based on an orifice

assumption, where the rim seal was assumed to be an orifice with a discharge coefficient for ingress,

𝐶𝑑,𝑖, and a discharge coefficient for egress, 𝐶𝑑,𝑒. Several orifice models have been developed with

varied success [86,90,91,97,98]. This section compares the experimental data to such an orifice

model presented by Owen et al. [86] for externally-induced ingress. The model given in Equation

(5.2) is a simple, yet powerful method for modeling sealing effectiveness as it requires no inputs

beyond the minimum sealing flow rate and the ratio of discharge coefficients, which can be

empirically determined. Specifically, the relationship between sealing flow rate and effectiveness

is given by

𝛷∗

𝛷𝑚𝑖𝑛

𝜀

[1 + Γ𝑐−2/3(1 − 𝜀)2/3]

3/2 (5.2)

where Γ𝑐 is the ratio of the ingress and egress discharge coefficients, 𝐶𝑑,𝑖 𝐶𝑑,𝑒⁄ , 𝜀 is the

sealing effectiveness, and 𝛷𝑚𝑖𝑛 is the minimum flow required to fully seal that location. There is a

slight modification here to the original equation presented by Owen et al. [86], and that is the

definition of 𝛷∗, which is the net sealing flow rate given by

𝛷∗ 𝛷 −𝛷0 (5.3)

where 𝛷0 is defined here as the value of 𝛷 at which effectiveness crosses zero. Note that for both

of the rim cavity locations 𝛷0/𝛷𝑟𝑒𝑓 0 for the data presented in Figure 5.7. For the rim seal

location 𝛷0/𝛷𝑟𝑒𝑓 0.2 for 150 purge holes and 𝛷0/𝛷𝑟𝑒𝑓 0.16 for 32 purge holes as shown in

Figure 5.7. Accounting for the zero-crossing is an important part of modeling the secondary air

130

system for a gas turbine, especially for an engine-realistic geometry where several sources, sinks,

and leakages may be present. Converting the gross flow rate, 𝛷, into a net flow rate, 𝛷∗, allowed

the model to account for the blade gap leakage and the rim seal source flow in this 1.5 stage turbine.

Since the front and aft rim cavity pressures remained constant for these experiments even as purge

flow rate varied, the blade gap leakage was assumed to be constant.

Figure 5.11 shows the effectiveness data at the purge hole radius (location C), at the outer

radius (location B), and in the rim seal (location A) plotted against 𝛷∗/𝛷𝑚𝑖𝑛 for 150 purge holes,

and Figure 5.12 shows the data for 32 purge holes. The lines represent the models for each data set

as indicated in Figures 5.11 and 5.12. The effectiveness data were used to determine the best fit for

the ratio of discharge coefficients, Γ𝑐, using Equation (5.2). Figures 5.11 and 5.12 also show the

empirically determined values of Γ𝑐 for each data set.

Figure 5.11. Empirical models for concentration effectiveness for 150 purge holes in terms of

the net and minimum sealing flow rates, ∗ and 𝒎𝒊𝒏.

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

0 0.2 0.4 0.6 0.8 1

εc

Φ*/Φmin

Data Model Γc

150 holes

0.380.985.20.1

A

B

C

131

Figure 5.12. Empirical models for concentration effectiveness for 32 purge holes in terms of

the net and minimum sealing flow rates, ∗ and 𝒎𝒊𝒏.

In the rim seal (location A shown in Figures 5.11 and 5.12) the model fit the data well over

most of the flow rate range. At higher flow rates, from 0.7 < 𝛷∗/𝛷𝑚𝑖𝑛 < 0.9, the effectiveness

data for 150 purge holes was ~0.1 higher than the model, as shown in Figure 5.11. The data and

empirical model were nearly linear for both configurations, which resulted in a large value for Γ𝑐.

For 150 purge holes Γ𝑐 5.2, as noted in Figure 5.11, and for 32 purge holes Γ𝑐 8.0, as noted in

Figure 5.12. The effectiveness for 32 purge holes never fully reached unity so the 𝛷𝑚𝑖𝑛 for the 150

holes configuration was used in the model for both configurations. The high values of Γ𝑐 indicated

the ingress discharge coefficient was much higher than the egress discharge coefficient, thus

promoting significantly more ingress than egress flow.

As Owen et al. [86] explained, their model uncoupled the pressure difference in the main

gas path from the effectiveness, which had the effect of changing the characterization of Γ𝑐

compared to previous models. The parameter Γ𝑐 presented by Owen et al. [86] empirically included

the effects of the pressure difference in the main gas path, while previous models required the

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

0 0.2 0.4 0.6 0.8 1

εc

Φ*/Φmin

A

B

CData Model Γc

32 holes

0.911.258.0

0.25

132

pressure difference as an input. Very high values of Γ𝑐 𝐶𝑑,𝑖 𝐶𝑑,𝑒⁄ , like those presented here for

the rim seal, are not represented in the literature at the present time. This model was presented in

recent years, so there are relatively few studies that report values of Γ𝑐 according to the definition

of Owen et al. [86]. It was not clear at the time of writing why the ratio of the discharge coefficients

was so high for the effectiveness data in the rim seal, but these values indicated that a better

understanding of the physics of ingestion in the rim seal is required to more accurately model the

effectiveness in the rim seal at engine-relevant conditions.

At the outer radius in the rim cavity (location B in Figures 5.11 and 5.12) the models fit

the data well for both 150 and 32 purge holes with values of Γ𝑐 0.98 and 1.25 respectively. The

ingress and egress discharge coefficients were thus of similar magnitude for the effectiveness

measurements at location B since Γ𝑐 was close to one. Previous measurements by other researchers

have revealed a variety of values for Γ𝑐 for different seal geometries. Some representative rim seals

with their associated values of Γ𝑐 are shown in Figure 5.13. Sangan et al. [60] reported a value of

0.48 for an axial seal and 1.32 for a single radial overlap seal, and Sangan et al. [62] reported 0.22

and 0.74 for the inner wheel space of two double seal designs. More recently Patinios et al. [31]

reported 0.68 for the same double radial overlap rim seal. The double seals shown by Sangan et al.

[62] also had a “buffer cavity”, which was similar to the rim seal in this paper and had higher values

of Γ𝑐 of 0.86 and 1.54. A review of ingress by Scobie et al. [99] used the data of Johnson et al. [98]

and Balasubramanian et al. [100] to derive a value of Γ𝑐 of 0.66 and 0.14 respectively for the same

radial overlap rim seal. Although the seal studied in this paper was a hybrid between a single radial

overlap and a double radial overlap seal, the ratio of discharge coefficients for location B compared

most closely to the single radial overlap seal (Γ𝑐 1.32) or the data in the buffer cavity in the

double seal (Γ𝑐 0.86) [60,62].

At the purge hole radius (location C in Figures 5.11 and 5.12) the empirical model did not

fit the data well due to the jet-in-crossflow behavior. Despite the poor match, a best-fit value of Γ𝑐

was still determined for the data for comparison purposes. The values were Γ𝑐 0.38 and 0.67 for

150 and 32 purge holes respectively, both of which indicated better sealing performance than at

location B at lower flow rates.

133

Figure 5.13. Comparison of several rim seals in terms of the empirically determined ratio of

discharge coefficients, 𝚪𝒄.

Due to the jet-in-crossflow behavior, the effectiveness data at the purge hole radius

(location C) exhibited better sealing at the lower flow rates, which could be modeled as a different

characteristic. Figure 5.11 also shows a characteristic for Γ𝑐 0.1, and the effectiveness data for

150 purge holes at location C agreed well with this characteristic up to a flow rate of 𝛷∗/𝛷𝑚𝑖𝑛

0.2. As the flow rate increased further the data approached the effectiveness characteristic at the

outer radius in the rim cavity (location B), and for 𝛷∗/𝛷𝑚𝑖𝑛 > 0.55 the data fit the characteristic

of Γ𝑐 0.98 shown at location B very well. An order of magnitude variation in the characteristic

Γ𝑐 over such a small range has not been observed in the literature previously, and it is believed that

the manner through which the purge flow was introduced and the momentum of the purge flow

were the causes. The importance of the purge flow momentum was shown previously by Clark et

al. [70] for a half-stage turbine with a stationary rim cavity. The momentum of the purge jet,

𝜌𝑗𝑒𝑡𝑉𝑗𝑒𝑡2 , at the higher value of Γ𝑐 at 𝛷∗/𝛷𝑚𝑖𝑛 0.55 was 4.4 to 6.2 times greater than the purge

jet momentum at the lower value of Γ𝑐 at 𝛷∗/𝛷𝑚𝑖𝑛 0.2.

The data for 32 purge holes at location C in Figure 5.12 showed similar behavior as 150

purge holes, although to a reduced degree. At low flow rates, for 𝛷∗/𝛷𝑚𝑖𝑛 < 0.15, the model

seemed to fit the data again with a lower characteristic of Γ𝑐 0.25 as shown in Figure 5.12. As

the flow rate increased the data approached the model at location B, and for 𝛷∗/𝛷𝑚𝑖𝑛 > 0.3 the

characteristic changed to Γ𝑐 1.25. The characteristic value of Γ𝑐 changed by a factor of five due

to the momentum of the purge jets. Again, the momentum of the purge jets was approximately 3.6

Γc=0.48 [60] Γc=1.32 [60] Γc=0.22 [62]Γc=0.74 [62], Γc=0.68 [31]

Γc=1.54 [62]

Γc=0.98, Γc=1.25

Γc=5.2-8.0Γc=0.86 [62]st

ato

r

roto

r

stat

or

roto

r

stat

or

roto

r

stat

or

roto

r

stat

or

roto

r

Γc=0.66 [98,99]Γc=0.14 [99,100]

134

to 5.3 times greater at the higher value of Γ𝑐 at 𝛷∗/𝛷𝑚𝑖𝑛 0.3 than at the lower value of Γ𝑐at

𝛷∗/𝛷𝑚𝑖𝑛 0.15.

5.7 Conclusions

This paper has presented sealing effectiveness measurements in the front cavity of a 1.5

stage turbine, with engine-realistic airfoils and cavity geometries, operated at engine-relevant

Reynolds and Mach numbers. Sealing effectiveness measurements were acquired through the use

of CO2 as a tracer gas. Benchmarking of the facility indicated steady state operation, as well as

periodic and repeatable conditions in the 1.5 stage turbine.

Sealing effectiveness increased with purge flow rate and with radial distance from the main

gas path. Less purge flow was required to produce fully sealed conditions in the rim cavity than in

the rim seal. Despite showing higher effectiveness in the wheel-space than in the rim cavity,

appreciable ingestion was shown to occur in the wheel-space for low flow rates. This data indicates

that ingestion deep within turbine cavities in an operating gas turbine would result in reduced

component lifetimes, or possibly catastrophic failures, highlighting the need to provide sufficient

TOBI flow to purge the front wheel-space in an operating engine. Of the two purge flow

configurations tested, the configuration including more purge holes resulted in higher sealing

effectiveness than fewer purge holes inboard of the purge flow injection location.

An orifice model was compared to the sealing effectiveness data with mixed results. The

model matched the effectiveness data at the outer radius in the rim cavity and in the rim seal, but

did not match the data at the purge flow injection location. The model’s inability to predict the

effectiveness is most likely due to the complexity of the geometry and purge flow delivery method.

The results suggest that orifice models may work well for matching sealing effectiveness data for

some cases, but the models break down for engine-realistic geometries and purge flow delivery as

shown in this paper.

135

Chapter 6

Conclusions and Recommendations

An experimental turbine research facility was designed, built, and commissioned during

the course of this dissertation. The facility was designed to simulate engine-relevant axial Reynolds

numbers, rotational Reynolds numbers, and Mach numbers at continuous flow conditions for

engine-realistic turbine hardware. Substantial infrastructure was required to produce the needed

operating conditions in the test section. A half stage turbine and a 1.5 stage turbine were designed,

built, installed, and commissioned to complete the research reported in this dissertation. The turbine

components—including the airfoils, rim seals, and rim cavities—were representative of a modern

operating gas turbine.

Extensive benchmarking experiments for the facility and turbine were also performed for

this dissertation work. The facility demonstrated steady and repeatable operation at engine-relevant

conditions, and the facility control system was shown to be safe and reliable. The test section

demonstrated circumferentially uniform inlet and exit flow conditions. The turbine demonstrated

periodic conditions while using additively manufactured first and second vanes and solid-cast

blades.

A major portion of the dissertation work also included designing, specifying, installing,

and commissioning the turbine instrumentation to provide accurate measurements. The facility

instrumentation was selected to provide accurate and robust measurements to ensure safe and

reliable operation of the facility. The turbine instrumentation was selected to provide high quality

pressure, temperature, gas concentration, and flow rate measurements. The measurement locations

in the turbine were selected to provide spatial resolution that is not typically obtained for engine-

realistic turbine hardware.

The primary measurement reported in this dissertation was sealing effectiveness between

turbine stages, which was shown to be a powerful means for deducing flow patterns in turbine rim

seals and rim cavities. Sealing effectiveness measurements were obtained by using CO2 as a tracer

gas in the secondary air supply and withdrawing gas samples at discrete locations in the turbine rim

seals and rim cavities. An experimental method was developed to determine the sensitivity of the

136

concentration measurements to the sampling flow rate. For both the stationary rim cavity in the half

stage turbine and the rotor-stator rim cavity in the 1.5 stage turbine, sampling sensitivity studies

allowed for the determination of the appropriate sampling flow rate at each location to ensure high

quality concentration measurements were acquired.

6.1 Conclusions

A unique feature of this research was the use of engine-realistic hardware. The rim seal,

rim cavity, and purge flow delivery were representative of a modern turbine design, including

engine-realistic leakage paths, such as leakages across the disk and through gaps between the

mating faces of adjoining airfoils. The sealing effectiveness data presented in this dissertation were

more complex than data for simplified geometries previously shown in the literature. Complex

interactions existed between the front and aft rim cavities, the front rim seal, and the front wheel-

space that have not been observed in the literature previously. Additionally, the presence of the

engine-realistic leakages further complicated the flow patterns in the rim seal and cavity. Important

information regarding the cavity flow physics has been obtained from past fundamental studies

using simplified geometries, which has led to the development and validation of ingestion models,

but as was shown in this dissertation, the geometries and associated flow fields in operating engines

are sufficiently complex to warrant further examination of those models with sealing effectiveness

data for engine-realistic hardware.

The sealing effectiveness in both the stationary and in the rotor-stator rim cavities increased

with purge flow rate. The sealing effectiveness was also shown to increase with radial distance

from the main gas path into the cavity for both the stationary and rotating cases. The highest

effectiveness for a given flow rate was observed deepest within the cavities, and the lowest

effectiveness was observed closest to the main gas path in the rim seal. A major result of this

research was that appreciable ingestion was observed in the front wheel-space inboard of the

labyrinth seal for low flow rates. The effectiveness measurements showed that the labyrinth seal

was not effective at preventing ingestion without providing supplementary flow inboard of the seal.

For the turbine designer, these results show the importance of using supplementary flow to

pressurize the front wheel-space to minimize ingestion past the labyrinth seals, because ingestion

137

deep within the turbine cavities can lead to reduced component lifetimes, or, in extreme cases,

potentially catastrophic effects.

In the rotor-stator cavity, the effectiveness depended on the number of purge holes at and

inboard of the purge injection location, with more purge holes exhibiting higher effectiveness than

fewer purge holes. Outboard of the purge location, the effectiveness was not affected by the number

of purge holes, indicating that the number of holes only affected the ingestion at and inboard of the

purge injection location. This finding was consistent with the limited previous literature for an

engine-realistic purge flow delivery, which showed that changing the purge flow delivery changed

the cavity flow field and the sealing effectiveness. The results of this dissertation show that, in

addition to studying engine-realistic rim seal geometries, it is also important to study engine-

realistic purge flow delivery.

This dissertation is unique in that the same rim seal and rim cavity geometry was tested

with and without the effects of rotation. The results indicated that different trends in effectiveness

observed in the stationary cavity and the rotor-stator cavity were attributed to flow field differences,

specifically the boundary layers in the rim cavities. For the rotating case, the flow in the rotor

boundary layer moved radially outward, which caused an axial flow across the cavity from the

stator to the rotor side and a radial inflow on the stator side. The axial flow entrained the purge flow

and allowed the purge flow to distribute throughout the cavity, thereby reducing ingestion. More

purge holes entrained more purge flow and more effectively prevented the ingested flow from

penetrating deeper within the cavity. For the stationary case, the lack of a rotor boundary layer and

its associated disk pumping resulted in significantly more ingestion as the purge flow did not

distribute throughout the cavity as much as in the rotor-stator cavity. As has been shown by the

literature, the results indicated the importance of including rotational effects in ingestion research.

An empirical orifice model was compared to the sealing effectiveness data in the rotor-

stator cavity. The model matched the effectiveness data at the outer radius in the rim cavity and in

the rim seal. The model did not match the data, however, at the purge holes or in the wheel-space.

At the purge holes, the model parameters exhibited a wide variation over a narrow range of flow

rates, highlighting the deficiency of the model in predicting effectiveness for realistic engine

geometries. The results of this dissertation indicate that orifice models may match sealing

effectiveness data for some cases, however, the models break down for engine-realistic seal

geometries and purge flow delivery methods.

138

6.2 Recommendations for Future Work

The sealing effectiveness measurements presented in this dissertation were very thorough,

but a higher resolution of effectiveness measurements will lead to an enhanced understanding of

cavity flow physics. Sealing effectiveness measurements were shown to be a powerful technique

in deducing rim cavity flow patterns, and effectiveness measurements in the aft cavity should also

be obtained to understand the sealing characteristics for the aft rim seal and rim cavity. There are

few studies in the open literature providing effectiveness measurements in the aft cavity. Obtaining

effectiveness measurements in the aft cavity of an engine-realistic 1.5 stage turbine would be

especially applicable to engines as they would allow for an enhanced understanding of the leakages

across the disk, as well as an understanding of re-ingested purge flow from the front cavity.

Geometry effects are important to sealing effectiveness. The rim seal geometry, in

particular, has been shown in the literature to have the most dramatic effect on sealing effectiveness.

It is important to continue to quantify the effectiveness of engine-realistic rim seal geometries at

engine-relevant conditions. Additional experiments in the START Lab, and in other turbine test

facilities that can operate near engine-relevant conditions, should test multiple engine-realistic rim

seal and rim cavity geometries to allow for a more complete effectiveness data set at engine-

relevant conditions.

Hot gas ingestion is affected by the pressure field in the main gas path, which is in turn

affected by the aerodynamic design of the turbine. The pressure field at the vane exit, the potential

field upstream of the passing rotor, and the unsteady interaction of the two fields affect the

boundary conditions on the rim seal. Endwall contouring is often used to reduce the effects of

aerodynamic secondary flows in turbine vane and blade passages. Efficient endwall contouring is

designed to manage the secondary flows such that the aerodynamic losses are minimized, which

would have an impact on the pressure field at the rim seal and would thus affect ingestion.

Effectiveness measurements with various airfoil and endwall contouring designs could greatly

enhance the understanding of how main gas path aerodynamic effects could be used to minimize

hot gas ingestion.

Computational fluids dynamics (CFD) simulations have had mixed success at predicting

hot gas ingestion. High-fidelity CFD simulations should be performed to more fully understand the

flow fields and the effectiveness trends observed in the measurements presented in this dissertation.

The measurements are time-averaged by nature, but the time-accurate nature of hot gas ingestion

139

can be studied through computational simulations. Gas turbine manufacturers require quick design

tools, so there is also a need to perform accurate, low-cost CFD simulations that accurately predict

hot gas ingestion for a wide range of geometries and conditions. CFD design tools will need to be

developed based on effectiveness data obtained at engine-relevant conditions. These tools will

likely be a mix of steady-state and time-accurate simulations. Performing accurate CFD simulations

for the data presented in this dissertation will be a step toward developing better design tools.

The experiments presented in this dissertation were performed at engine-relevant Mach

numbers, axial Reynolds numbers, and rotational Reynolds numbers. At the time of this writing,

very few experiments have been operated near these conditions and presented in the open literature,

so there is a need to continue operating experiments at engine-relevant conditions to understand

ingestion at these conditions. A more expansive effectiveness data set at engine-relevant conditions

will lead to the development of improved ingestion models.

6.3 Concluding Remarks

It is expected that the results presented in this dissertation will be used to develop better

tools for secondary air system designers. Better design tools will allow for more accurate prediction

of ingestion, leading to more durable engines, and will also lead to more efficient use of the

secondary air, which will lead to a reduction in fuel burn and emissions for gas turbines. It is

expected that this dissertation work will contribute to the goal set by the U.S. Department of Energy

– National Energy Technology Laboratory of reducing secondary air usage in land-based gas

turbines to increase the efficiency of land-based, combined-cycle, power plants by 3-5%. It was

noted that in an aircraft engine a savings in the secondary air system of 10% of the core inlet flow

would reduce the overall fuel burn by 5%. This research alone will not be sufficient to reduce those

numbers, as a reduction in secondary air usage due to cooling technologies will also be needed to

achieve that goal. A combination of research in cooling technologies and sealing technologies is

needed, and it is expected that the START Lab, with its unique capabilities, will be at the forefront

of that research for years to come.

140

References

[1] Glahn, A., and Schmitz, J., 2012, “Internal Report.”

[2] U.S. Energy Information Administration, 2016, “Monthly Energy Review, June 2016” [Online].

Available: www.eia.gov/totalenergy/data/monthly/archive/00351606.pdf. [Accessed: 23-Jul-2016].

[3] International Air Transport Association (IATA), 2016, “Fuel Price Analysis” [Online]. Available:

http://www.iata.org/publications/economics/fuel-monitor/Pages/price-analysis.aspx. [Accessed: 22-

Jul-2016].

[4] U.S. Energy Information Administration, 2016, “Carbon Dioxide Emissions Coefficients” [Online].

Available: www.eia.gov/environment/emissions/co2_vol_mass.cfm. [Accessed: 23-Jul-2016].

[5] U.S. Environmental Protection Agency, 2016, “Greenhouse Gas Emissions from a Typical Passenger

Vehicle” [Online]. Available: www.epa.gov/sites/production/files/2016-

02/documents/420f14040a.pdf. [Accessed: 23-Jul-2016].

[6] Siemens AG, 2016, “Press Release: Siemens sets new performance and efficiency world record at

Düsseldorf power plant.”

[7] United Technologies Corporation, Pratt & Whitney Division, 2014, “PW4000-112 ENGINE” [Online].

Available: http://pw.utc.com/PW4000112_Engine. [Accessed: 24-Nov-2014].

[8] Rolls-Royce, The Jet Engine.

[9] Johnson, B., Mack, G., Paolillo, R., and Daniels, W., 1994, “Turbine Rim Seal Gas Path Flow Ingestion

Mechanisms,” 30th AIAA/ASME/SAE/ASEE Joint Propulsion Conference.

[10] Barringer, M., Coward, A., Clark, K., Thole, K., Schmitz, J., Wagner, J., Alvin, M. A., Burke, P., and

Dennis, R., 2014, “Development of a Steady Aero Thermal Research Turbine (START) for Studying

Secondary Flow Leakages and Airfoil Heat Transfer,” GT2014-25570.

[11] Owen, J. M., Zhou, K., Pountney, O., Wilson, M., and Lock, G., 2012, “Prediction of Ingress Through

Turbine Rim Seals--Part I: Externally Induced Ingress,” J. Turbomach., 134(3), pp. 031012–031012.

[12] Barringer, M., Coward, A., Clark, K., Thole, K., Schmitz, J., Wagner, J., Alvin, M. A., Burke, P., and

Dennis, R., 2014, “Development of a Steady Aero Thermal Research Turbine (START) for Studying

Secondary Flow Leakages and Airfoil Heat Transfer,” ASME Turbo Expo 2014, Paper No. GT2014-

25570, Dusseldorf, Germany.

[13] Dunn, M. G., and Stoddard, F. J., “Application of Shock-Tube Technology to the Measurement of Heat-

Transfer Rate to Gas Turbine Components,” 11th International Symposium on Shock Tubes and Waves.

[14] Ainsworth, R. W., Schultz, D. L., Davies, M. R. D., Forth, C. J. P., Hilditch, M. A., and Oldfield, M.

L. G., 1988, “A Transient Flow Facility for the Study of the Thermofluid-Dynamics of a Full State

Turbine Under Engine Representative Conditions,” ASME Paper No. 88-GT-144.

[15] Hilditch, M. A., Fowler, A., Hones, T. V., Chana, K. S., Oldfield, M. L. G., Ainsworth, R. W., Hogg, S.

I., Anderson, S. J., and Smith, G. C., 1994, “Installation of a Turbine Stage in the Pyestock Isentropic

Light Piston Facility,” ASME Paper No. 94-GT-277.

[16] Consigny, H., and Richards, B. E., 1982, “Short Duration Measurements of Heat-Transfer Rate to a

Gas Turbine Rotor Blade,” J. Eng. Gas Turbines Power, 104(3), pp. 542–550.

[17] Haldemann, C. W., Clark, J. P., MacArthur, C. D., and Murawski, C. G., 1992, The USAF Advanced

Turbine Aerothermal Research Rig (ATARR).

[18] Anthony, R. J., Clark, J. P., Finnegan, J. M., and Johnson, D., 2012, “Modifications and Upgrades to

the AFRL Turbine Research Facility,” ASME Paper No. GT2012-70084, pp. 793–804.

[19] Dunn, M., and Mathison, R., 2013, “History of Short-Duration Measurement Programs Related to Gas

Turbine Heat Transfer, Aerodynamics, and Aeroperformance at Calspan and The Ohio State

University,” J. Turbomach., 136(4), p. 41004.

[20] Chana, K. S., Cardwell, D., and Jones, T., 2013, “A Review of the Oxford Turbine Research Facility,”

ASME Paper No. GT2013-95687.

141

[21] Anthony, R. J., and Clark, J. P., 2013, “A Review of the AFRL Turbine Research Facility,” ASME

Paper No. GT2013-94741.

[22] Paniagua, G., Sieverding, C. H., and Arts, T., 2013, “Review of the Von Karman Institute Compression

Tube Facility for Turbine Research,” ASME Paper No. GT2013-95984.

[23] Epstein, A., Guenette, G., and Norton, R. J. G., 1984, “The MIT Blowdown Turbine Facility,” ASME

Paper No. 84-GT-116.

[24] Thole, K., Barringer, M., and Alvin, M. A., 2011, “Development of a Rotating Rig to Study Secondary

Flow Leakages and Aerothermal Cooling.”

[25] Childs, P. R. N., 2011, Rotating Flow, Butterworth-Heinemann, Elsevier, Oxford.

[26] Roy, R. P., Xu, G., Feng, J., and Kang, S., 2001, “Pressure Field and Main-Stream Gas Ingestion in a

Rotor-Stator Disk Cavity,” ASME Paper No. 2001-GT-0564.

[27] R. P. Roy, J. F., 2005, “Experiment on Gas Ingestion Through Axial-Flow Turbine Rim Seals,” J. Eng.

Gas Turbines Power, 127(3).

[28] Sangan, C. M., Pountney, O. J., Zhou, K., Wilson, M., Michael Owen, J., and Lock, G. D., 2012,

“Experimental Measurements of Ingestion Through Turbine Rim Seals—Part I: Externally Induced

Ingress,” J. Turbomach., 135(2), pp. 021012–021012.

[29] Sangan, C. M., Pountney, O. J., Scobie, J. A., Wilson, M., Owen, J. M., and Lock, G. D., 2013,

“Experimental Measurements of Ingestion Through Turbine Rim Seals—Part III: Single and Double

Seals,” J. Turbomach., 135(5), pp. 051011–051011.

[30] Pountney, O. J., Sangan, C. M., Lock, G. D., and Owen, J. M., 2013, “Effect of Ingestion on

Temperature of Turbine Disks,” J. Turbomach., 135(5), pp. 051010–051010.

[31] Patinios, M., Scobie, J. A., Sangan, C. M., Owen, J. M., and Lock, G. D., 2016, “Measurements and

Modelling of Ingress in a New 1.5-Stage Turbine Research Facility,” ASME Paper No. GT2016-57163.

[32] Lakshminarayana, B., Camci, C., Hallie, I., and Zaccaria, M., 1996, “Design, Development, and

Performance of a Turbine Research Facility to Study Rotor Stator Interaction Effects,” Int. J. Turbo Jet

Engines, 13(3), pp. 155–172.

[33] Gallier, K., Lawless, P., and Fleeter, S., 2014, “Investigation of seal purge flow effects on the hub flow

field in a turbine stage using particle image velocimetry,” 36th AIAA/ASME/SAE/ASEE Joint

Propulsion Conference and Exhibit, American Institute of Aeronautics and Astronautics.

[34] Sell, M., Schlienger, J., Pfau, A., Treiber, M., and Abhari, R. S., 2001, “The 2-Stage Axial Turbine Test

Facility ‘LISA,’” ASME Paper No. 2001-GT-0492.

[35] Schmitz, J., 2010, “Experimental Measurements in a Highly Loaded Low Pressure Turbine Stage,”

University of Notre Dame.

[36] Bohn, D. E., Decker, A., Ma, H., and Wolff, M., 2003, “Influence of Sealing Air Mass Flow on the

Velocity Distribution In and Inside the Rim Seal of the Upstream Cavity of a 1.5-Stage Turbine,”

ASME Paper No. GT2003-38459.

[37] Bohn, D. E., Decker, A., Ohlendorf, N., and Jakoby, R., 2006, “Influence of an Axial and Radial Rim

Seal Geometry on Hot Gas Ingestion Into the Upstream Cavity of a 1.5-Stage Turbine,” pp. 1413–

1422.

[38] Coren, D. D., Atkins, N. R., Turner, J. R., Eastwood, D. E., Davies, S., Child, P. R. N., Dixon, J. A.,

and Scanlon, T. J., 2012, “An Advanced Multiconfiguration Stator Well Cooling Test Facility,” J.

Turbomach., 135(1), p. 11003.

[39] Coren, D. D., Atkins, N. R., Long, C. A., Eastwood, D., Childs, P. R. N., Guijarro-Valencia, A., and

Dixon, J. A., 2011, “The Influence of Turbine Stator Well Coolant Flow Rate and Passage

Configuration on Cooling Effectiveness,” ASME Paper No. GT2011-46448.

[40] Dixon, J. A., Valencia, A. G., Coren, D., Eastwood, D., and Long, C., 2013, “Main Annulus Gas Path

Interactions--Turbine Stator Well Heat Transfer,” J. Turbomach., 136(2), p. 21010.

[41] Andreini, A., Da Soghe, R., and Facchini, B., 2010, “Turbine Stator Well CFD Studies: Effects of

Coolant Supply Geometry on Cavity Sealing Performance,” J. Turbomach., 133(2), p. 21008.

[42] Palafox, P., Ding, Z., Bailey, J., Vanduser, T., Kirtley, K., Moore, K., and Chupp, R., 2013, “A New

1.5-Stage Turbine Wheelspace Hot Gas Ingestion Rig (HGIR): Part I - Experimental Test Vehicle,

Measurement Capability and Baseline Results,” ASME Turbo Expo 2013, Paper No. GT2013-96020.

142

[43] Ding, Z., Palafox, P., Moore, K., Chupp, R., and Kirtley, K., 2013, “A New 1.5-Stage Turbine

Wheelspace Hot Gas Ingestion Rig (HGIR): Part II — CFD Modeling and Validation,” p.

V03AT15A022.

[44] Buckingham, E., 1914, “On Physically Similar Systems: Illustrations of the Use of Dimensional

Equations,” Phys. Rev., 4, pp. 345–376.

[45] Gazda, M., 2016, Dynamometer Commissioning, Kahn Industries, Personal Communication.

[46] Longnecker, K., Wiley, D., and Mardis, J., 2013, Design of Turbine, Pratt & Whitney and Belcan,

Personal communication.

[47] Feiereisen, J. M., Paolillo, R. E., and Wagner, J., 2000, “UTRC Turbine Rim Seal Ingestion and

Platform Cooling Experiments,” 36th AIAA/ASME/SAE/ASEE Joint Propulsion Conference.

[48] Murphy, B. T., 2012, “XLRotor Version 3.91” [Online]. Available: www.xlrotor.com.

[49] Coward, A., 2012, Rotor Dynamics Analysis, Personal Communication.

[50] Pratt & Whitney, 2012, Rotor Dynamics Analysis, Personal Communication.

[51] Synchrony, 2013, Rotor Dynamics Analysis, Personal Communication.

[52] Campbell, W., 1924, Protection of Steam Turbine Disk Wheels from Axial Vibration.

[53] Meher-Homji, C. B., and Prisell, E., 2005, “Dr. Max Bentele--Pioneer of the Jet Age,” J. Eng. Gas

Turbines Power, 127, pp. 231–239.

[54] Figliola, R. S., and Beasley, D. E., 2006, Theory and Design for Mechanical Measurements, John Wiley

& Sons, Inc.

[55] Colorado Experiment Engineering Station Inc. (CEESI), 2012, Venturi Flow Meter Calibration,

Personal Communication.

[56] Mickey, J., 2012, Programmable Logic Controller, Penn State Applied Research Lab, Personal

Communication.

[57] Phadke, U. P., and Owen, J. M., 1988, “Aerodynamic aspects of the sealing of gas-turbine rotor-stator

systems Part 2: The performance of simple seals in a quasi-axisymmetric external flow,” Int. J. Heat

Fluid Flow, 9(2), pp. 106–112.

[58] Graber, D. J., Daniels, W. A., and Johnson, B. V., 1987, Disk Pumping Test., Aero Propulsion

Laboratory, Wright-Patterson Air Force Base, Ohio 45433.

[59] Gentilhomme, O. J. P., 2004, “Turbine Rim Seal Ingestion,” University of Sussex.

[60] Sangan, C. M., Pountney, O. J., Zhou, K., Wilson, M., Michael Owen, J., and Lock, G. D., 2012,

“Experimental Measurements of Ingestion Through Turbine Rim Seals—Part I: Externally Induced

Ingress,” J. Turbomach., 135(2), pp. 021012–021012.

[61] Sangan, C. M., Pountney, O. J., Zhou, K., Owen, J. M., Wilson, M., and Lock, G. D., 2012,

“Experimental Measurements of Ingestion Through Turbine Rim Seals—Part II: Rotationally Induced

Ingress,” J. Turbomach., 135(2), pp. 021013–021013.

[62] Sangan, C. M., Pountney, O. J., Scobie, J. A., Wilson, M., Owen, J. M., and Lock, G. D., 2013,

“Experimental Measurements of Ingestion Through Turbine Rim Seals—Part III: Single and Double

Seals,” J. Turbomach., 135(5), pp. 051011–051011.

[63] Palafox, P., Ding, Z., Bailey, J., Vanduser, T., Kirtley, K., Moore, K., and Chupp, R., 2013, “A New

1.5-Stage Turbine Wheelspace Hot Gas Ingestion Rig (HGIR): Part I - Experimental Test Vehicle,

Measurement Capability and Baseline Results,” ASME Turbo Expo 2013, Paper No. GT2013-96020.

[64] Bohn, D. E., Decker, A., Ma, H., and Wolff, M., 2003, “Influence of Sealing Air Mass Flow on the

Velocity Distribution In and Inside the Rim Seal of the Upstream Cavity of a 1.5-Stage Turbine,”

ASME Paper No. GT2003-38459.

[65] Roy, R. P., Xu, G., Feng, J., and Kang, S., 2001, “Pressure Field and Main-Stream Gas Ingestion in a

Rotor-Stator Disk Cavity,” ASME Paper No. 2001-GT-0564.

[66] R. P. Roy, J. F., 2005, “Experiment on Gas Ingestion Through Axial-Flow Turbine Rim Seals,” J. Eng.

Gas Turbines Power, 127(3).

[67] Barringer, M., Coward, A., Clark, K., Thole, K., Schmitz, J., Wagner, J., Alvin, M. A., Burke, P., and

Dennis, R., 2014, “Development of a Steady Aero Thermal Research Turbine (START) for Studying

Secondary Flow Leakages and Airfoil Heat Transfer,” ASME Turbo Expo 2014, Paper No. GT2014-

25570, Dusseldorf, Germany.

143

[68] Feiereisen, J. M., Paolillo, R. E., and Wagner, J., 2000, “UTRC Turbine Rim Seal Ingestion and

Platform Cooling Experiments,” 36th AIAA/ASME/SAE/ASEE Joint Propulsion Conference.

[69] Robak, C., 2014, “CFD Pre-Test Predictions: First Vane Aerodynamic Loading,” Pratt & Whitney

Internal.

[70] Clark, K., Barringer, M., Coward, A., Thole, K., Clum, C., Hiester, P., Memory, C., and Robak, C.,

2016, “Effects of Purge Distribution and Flowrates on Vane Rim Sealing,” ASME Turbo Expo 2016,

Paper No. GT2016-58098, Seoul, South Korea.

[71] Abe, T., Kikuchi, J., and Takeuchi, H., 1979, “An Investigation of Turbine Disk Cooling (Experimental

Investigation and Observation of Hot Gas Flow into a Wheelspace),” 13th International Congress on

Combustion Engines (CIMAC), Vienna, Austria.

[72] Phadke, U. P., and Owen, J. M., 1988, “Aerodynamic aspects of the sealing of gas-turbine rotor-stator

systems Part 1: The behavior of simple shrouded rotating-disk systems in a quiescent environment,”

Int. J. Heat Fluid Flow, 9(2), pp. 98–105.

[73] Gibson, J., Thole, K., Christophel, J., and Memory, C., 2015, “Effects of the Main Gas Path Pressure

Field on Rim Seal Flows in a Stationary Linear Cascade,” ASME Turbo Expo 2015, Paper No.

GT2015-43517, Montreal, Canada.

[74] Scobie, J. A., Teuber, R., Li, Y. S., Sangan, C. M., Wilson, M., and Lock, G. D., 2016, “Design of an

Improved Turbine Rim-Seal,” J. Turbomach., 138(11).

[75] Coren, D. D., Atkins, N. R., Long, C. A., Eastwood, D., Childs, P. R. N., Guijarro-Valencia, A., and

Dixon, J. A., 2011, “The Influence of Turbine Stator Well Coolant Flow Rate and Passage

Configuration on Cooling Effectiveness,” ASME Paper No. GT2011-46448.

[76] Andreini, A., Da Soghe, R., and Facchini, B., 2010, “Turbine Stator Well CFD Studies: Effects of

Coolant Supply Geometry on Cavity Sealing Performance,” J. Turbomach., 133(2), pp. 021008–

021008.

[77] Teuber, R., Li, Y. S., Maltson, J., Wilson, M., Lock, G. D., and Owen, J. M., 2013, “Computational

extrapolation of turbine sealing effectiveness from test rig to engine conditions,” Proc. Inst. Mech.

Eng. Part J. Power Energy, 227(2), pp. 167–178.

[78] Owen, J. M., 2010, “Prediction of Ingestion Through Turbine Rim Seals—Part II: Externally Induced

and Combined Ingress,” J. Turbomach., 133(3), pp. 031006–031006.

[79] Bunker, R. S., Laskowski, G. M., Bailey, J. C., Palafox, P., Kapetanovic, S., Itzel, G. M., Sullivan, M.

A., and Farrell, T. R., 2011, “An Investigation of Turbine Wheelspace Cooling Flow Interactions With

a Transonic Hot Gas Path--Part 1: Experimental Measurements,” J. Turbomach., 133(2).

[80] Clark, K., Barringer, M., Coward, A., Thole, K., Clum, C., Hiester, P., Memory, C., and Robak, C.,

2016, “Using a Tracer Gas to Deduce Sealing Effectiveness for an Engine Realistic Rim Seal

Geometry,” ASME Turbo Expo 2016, Paper No. GT2016-58095, Seoul, South Korea.

[81] Zhou, D. W., Roy, R. P., Wang, C.-Z., and Glahn, J. A., 2011, “Main Gas Ingestion in a Turbine Stage

for Three Rim Cavity Configurations,” J. Turbomach., 133(3), p. 31023.

[82] Robak, C., 2014, “CFD Pre-Test Predictions: Rim Seal Static Pressures,” Pratt & Whitney Internal.

[83] Pountney, O. J., Sangan, C. M., Lock, G. D., and Owen, J. M., 2013, “Effect of Ingestion on

Temperature of Turbine Disks,” J. Turbomach., 135(5), p. 51010.

[84] Cho, G. H., Sangan, C. M., Owen, J. M., and Lock, G. D., 2015, “Effect of Ingress on Turbine Discs,”

ASME Paper No. GT2015-643234.

[85] Johnson, B., Mack, G., Paolillo, R., and Daniels, W., 1994, “Turbine Rim Seal Gas Path Flow Ingestion

Mechanisms,” 30th AIAA/ASME/SAE/ASEE Joint Propulsion Conference.

[86] Owen, J. M., Zhou, K., Pountney, O., Wilson, M., and Lock, G., 2012, “Prediction of Ingress Through

Turbine Rim Seals--Part I: Externally Induced Ingress,” J. Turbomach., 134(3), pp. 031012–031012.

[87] Bayley, F. J., and Owen, J. M., 1970, “The Fluid Dynamics of a Shrouded Disk System With a Radial

Outflow of Coolant,” J. Eng. Power, 92(3), pp. 335–341.

[88] Phadke, U. P., and Owen, J. M., 1988, “Aerodynamic aspects of the sealing of gas-turbine rotor-stator

systems Part 1: The behavior of simple shrouded rotating-disk systems in a quiescent environment,”

Int. J. Heat Fluid Flow, 9(2), pp. 98–105.

144

[89] Phadke, U. P., and Owen, J. M., 1988, “Aerodynamic aspects of the sealing of gas-turbine rotor-stator

systems Part 2: The performance of simple seals in a quasi-axisymmetric external flow,” Int. J. Heat

Fluid Flow, 9(2), pp. 106–112.

[90] Hamabe, K., and Ishida, K., 1992, “Rim Seal Experiments and Analysis of a Rotor-Stator System With

Nonaxisymmetric Main Flow,” p. V001T01A060.

[91] Chew, J. W., Green, T., and Turner, A. B., 1994, “Rim Sealing of Rotor-Stator Wheelspaces in the

Presence of External Flow,” 94-GT-126, The Hague, Netherlands.

[92] Bohn, D. E., Decker, A., Ma, H., and Wolff, M., 2003, “Influence of Sealing Air Mass Flow on the

Velocity Distribution In and Inside the Rim Seal of the Upstream Cavity of a 1.5-Stage Turbine,”

GT2003-38459.

[93] Wang, C.-Z., Mathiyalagan, S. P., Johnson, B. V., Glahn, J. A., and Cloud, D. F., 2013, “Rim Seal

Ingestion in a Turbine Stage From 360 Degree Time-Dependent Numerical Simulations,” J.

Turbomach., 136(3), p. 31007.

[94] Graber, D. J., Daniels, W. A., and Johnson, B. V., 1987, Disk Pumping Test., Aero Propulsion

Laboratory, Wright-Patterson Air Force Base, Ohio 45433.

[95] Teuber, R., Li, Y. S., Maltson, J., Wilson, M., Lock, G. D., and Owen, J. M., 2013, “Computational

extrapolation of turbine sealing effectiveness from test rig to engine conditions,” Proc. Inst. Mech.

Eng. Part J. Power Energy, 227(2), pp. 167–178.

[96] Green, T., and Turner, A. B., 1994, “Ingestion Into the Upstream Wheelspace of an Axial Turbine

Stage,” J. Turbomach., 116(2), pp. 327–332.

[97] Phadke, U. P., and Owen, J. M., 1988, “Aerodynamic aspects of the sealing of gas-turbine rotor-stator

systems Part 3: The effect of nonaxisymmetric external flow on seal performance,” Int. J. Heat Fluid

Flow, (2), pp. 113–117.

[98] Johnson, B. V., Wang, C.-Z., and Roy, R. P., 2008, “A Rim Seal Orifice Model With 2 Cds and Effects

of Swirl in Seals,” pp. 1531–1541.

[99] Scobie, J. A., Sangan, C. M., Michael Owen, J., and Lock, G. D., 2016, “Review of Ingress in Gas

Turbines,” J. Eng. Gas Turbines Power, 138(12), pp. 120801–120801.

[100] Balasubramanian, J., Pathak, P. S., Thiagarajan, J. K., Singh, P., Roy, R. P., and Mirzamoghadam, A.

V., 2015, “Experimental Study of Ingestion in the Rotor–Stator Disk Cavity of a Subscale Axial

Turbine Stage,” J. Turbomach., 137(9), pp. 091010–091010.

145

Appendix

Design-Stage Uncertainty Analysis

As was mentioned previously, a design-stage uncertainty analysis was performed using the

reported accuracies of the instrumentation. It was assumed that since the measurements were to be

steady state and many samples could be acquired that the precision uncertainty would play a small

part in the final uncertainty. Precision error was unknown until real data could be acquired, so it

was neglected for this analysis so that instrumentation could be selected that minimized the

instrument bias uncertainty.

Since each of the reported measurements was composed of a number of individual

measurements, propagation of error resulted in increased uncertainty for the reported values. This

analysis was performed using the method of Figliola and Beasley [54]. Considering the linear terms

of a Taylor series expansion of a functional relationship allowed for an approximation of the error

in that final value due to the uncertainties in each measurement by taking the root sum square of

each of the sensitivity indices multiplied by the uncertainty in each measurement.

Equation A.1 shows a general function of some parameter, 𝑅, which is a function of 𝑛

variables 𝑥1 through 𝑥𝑛. The uncertainty in parameter 𝑅 is given in Equation A.2, where 𝑢𝑅 is the

approximated uncertainty in parameter 𝑅, 𝑢𝑥𝑖 is the uncertainty in variable 𝑥𝑖, and 𝜕𝑅

𝜕𝑥𝑖 is the partial

derivative of Equation A.1 with respect to variable 𝑥𝑖 (this is the sensitivity index).

𝑅 𝑓(𝑥1, 𝑥2, … , 𝑥𝑛)

(A.1)

𝑢𝑅 ±√(𝜕𝑅

𝜕𝑥1𝑢𝑥1)

2

+ (𝜕𝑅

𝜕𝑥2𝑢𝑥2)

2

+⋯+ (𝜕𝑅

𝜕𝑥𝑛𝑢𝑥𝑛)

2

(A.2)

As an example, the uncertainty in density as calculated by the ideal gas law can be

represented as shown in Equation A.3, which becomes Equation A.4 with the sensitivity indices

substituted into the equation.

146

𝑢𝜌 ±√(𝜕𝜌

𝜕𝑝𝑢𝑝)

2

+ (𝜕𝜌

𝜕𝑅𝑢𝑅)

2

+ (𝜕𝜌

𝜕𝑇𝑢𝑇)

2

(A.3)

𝑢𝜌 ±√(1

𝑅𝑇𝑢𝑝)

2

+ (−𝑝

𝑅2𝑇𝑢𝑅)

2

+ (−𝑝

𝑅𝑇2𝑢𝑇)

2

(A.4)

Design-Stage Uncertainty Equations

Several equations pertaining to the design-stage uncertainty analysis are provided in this

section for many parameters. Most of the parameters listed were measured for this research. Some

parameters are included that were not reported in this dissertation, but these will be of interest for

future measurements. This analysis was performed to specify the facility and turbine

instrumentation to ensure that the measurements for the research presented in this dissertation and

future measurements would be accurate.

It should be noted that in the following equations the subscripts 1 through 4 indicate

different axial planes in the turbine: (1) first vane inlet, (2) first vane exit or blade inlet, (3) blade

exit or second vane inlet, and (4) second vane exit.

First vane total pressure loss:

𝜁𝑝𝑡,1V 𝑝𝑡1̿̿ ̿̿ − 𝑝𝑡2̿̿ ̿̿

𝑝𝑡1̿̿ ̿̿

(A.5)

𝑢𝜁𝑝𝑡,1V ±√(𝜕𝜁𝑝𝑡,1𝑉𝜕𝑝𝑡1

𝑢𝑝𝑡1)

2

+ (𝜕𝜁𝑝𝑡,1𝑉𝜕𝑝𝑡2

𝑢𝑝𝑡2)

2

(A.6)

where 𝜕𝜁𝑝𝑡,1𝑉𝜕𝑝𝑡1

𝑝𝑡2̿̿ ̿̿

(𝑝𝑡1̿̿ ̿̿ )2

(A.7)

𝜕𝜁𝑝𝑡,1𝑉𝜕𝑝𝑡2

1

𝑝𝑡1̿̿ ̿̿

(A.8)

Second vane total pressure loss:

𝜁𝑝𝑡,2V 𝑝𝑡3̿̿ ̿̿ − 𝑝𝑡4̿̿ ̿̿

𝑝𝑡3̿̿ ̿̿

(A.9)

147

𝑢𝜁𝑝𝑡,2V ±√(𝜕𝜁𝑝𝑡,2𝑉𝜕𝑝𝑡3

𝑢𝑝𝑡3)

2

+ (𝜕𝜁𝑝𝑡,2𝑉𝜕𝑝𝑡4

𝑢𝑝𝑡4)

2

(A.10)

where 𝜕𝜁𝑝𝑡,2𝑉𝜕𝑝𝑡3

𝑝𝑡4̿̿ ̿̿

(𝑝𝑡3̿̿ ̿̿ )2

(A.11)

𝜕𝜁𝑝𝑡,2𝑉𝜕𝑝𝑡4

1

𝑝𝑡3̿̿ ̿̿

(A.12)

Total-to-total efficiency:

𝜂𝑇𝑇 1 − (�̿�𝑡4/�̿�𝑡1)

1 − (�̿�𝑡4/�̿�𝑡1)(𝛾−1)/𝛾

(A.13)

𝑢𝜂𝑇𝑇 ±

√

(𝜕𝜂𝑇𝑇𝜕𝑝𝑡1

𝑢𝑝𝑡1)2

+ (𝜕𝜂𝑇𝑇𝜕𝑝𝑡4

𝑢𝑝𝑡4)2

+⋯

(𝜕𝜂𝑇𝑇𝜕T𝑡1

𝑢𝑇𝑡1)2

+ (𝜕𝜂𝑇𝑇𝜕T𝑡4

𝑢𝑇𝑡4)2

(A.14)

where

𝜕𝜂𝑇𝑇𝜕p𝑡4

−(𝛾 − 1)(T𝑡4 − T𝑡1)(𝑝𝑡4/𝑝𝑡1)

(𝛾−1)/𝛾

𝛾T𝑡1𝑝𝑡4((𝑝𝑡4/𝑝𝑡1)(𝛾−1)/𝛾 − 1)2

(A.15)

𝜕𝜂𝑇𝑇𝜕p𝑡1

(𝛾 − 1)(T𝑡4 − T𝑡1)(𝑝𝑡4/𝑝𝑡1)

(𝛾−1)/𝛾

𝛾T𝑡1𝑝𝑡1((𝑝𝑡4/𝑝𝑡1)(𝛾−1)/𝛾 − 1)2

(A.16)

𝜕𝜂𝑇𝑇𝜕T𝑡4

1/T𝑡1

(𝑝𝑡4/𝑝𝑡1)(𝛾−1)/𝛾 − 1

(A.17)

𝜕𝜂𝑇𝑇𝜕T𝑡1

T𝑡4/T𝑡1

2

(𝑝𝑡4/𝑝𝑡1)(𝛾−1)/𝛾 − 1

(A.18)

Torque-based efficiency:

𝜂𝜏 𝜏𝛺

�̇�𝑖𝑛𝑐𝑝𝑇𝑡1[1 − (�̿�𝑡4/�̿�𝑡1)(𝛾−1)/𝛾]

(A.19)

𝑢𝜂𝜏 ±

√

(𝜕𝜂𝜏𝜕𝜏

𝑢𝜏)2

+ (𝜕𝜂𝜏𝜕𝛺

𝑢𝛺)2

+ (𝜕𝜂𝜏𝜕�̇�𝑖𝑛

𝑢�̇�𝑖𝑛)2

+⋯

(𝜕𝜂𝜏𝜕𝑇𝑡1

𝑢𝑇𝑡1)2

+ (𝜕𝜂𝜏𝜕𝑝𝑡1

𝑢𝑝𝑡1)2

+ (𝜕𝜂𝜏𝜕𝑝𝑡4

𝑢𝑝𝑡4)2

(A.20)

148

where 𝜕𝜂𝜏𝜕𝜏

𝛺

�̇�𝑖𝑛𝑐𝑝𝑇𝑡1[1 − (�̿�𝑡4/�̿�𝑡1)(𝛾−1)/𝛾]

(A.21)

𝜕𝜂𝜏𝜕𝛺

−𝜏

�̇�𝑖𝑛𝑐𝑝𝑇𝑡1[1 − (�̿�𝑡4/�̿�𝑡1)(𝛾−1)/𝛾]

(A.22)

𝜕𝜂𝜏𝜕�̇�𝑖𝑛

−𝜏𝛺

(�̇�𝑖𝑛)2𝑐𝑝𝑇𝑡1[1 − (�̿�𝑡4/�̿�𝑡1)

(𝛾−1)/𝛾]

(A.23)

𝜕𝜂𝜏𝜕𝑇𝑡1

−𝜏𝛺

�̇�𝑖𝑛𝑐𝑝(𝑇𝑡1)2[1 − (�̿�𝑡4/�̿�𝑡1)

(𝛾−1)/𝛾]

(A.24)

𝜕𝜂𝜏𝜕𝑝𝑡1

(𝛾 − 1)𝜏𝛺(�̿�𝑡4/�̿�𝑡1)

(𝛾−1)/𝛾

𝛾�̇�𝑖𝑛𝑐𝑝𝑇𝑡1𝑝𝑡4[1 − (�̿�𝑡4/�̿�𝑡1)(𝛾−1)/𝛾]2

(A.25)

𝜕𝜂𝜏𝜕𝑝𝑡4

−(𝛾 − 1)𝜏𝛺(�̿�𝑡4/�̿�𝑡1)

(𝛾−1)/𝛾

𝛾�̇�𝑖𝑛𝑐𝑝𝑇𝑡1𝑝𝑡1[1 − (�̿�𝑡4/�̿�𝑡1)(𝛾−1)/𝛾]2

(A.26)

Vane aerodynamic loading (non-dimensional surface pressure):

𝜉 𝑝

𝑝𝑡1̿̿ ̿̿

(A.27)

𝑢𝜉 ±√(𝜕𝜉

𝜕𝑝𝑢𝑝)

2

+ (𝜕𝜉

𝜕𝑝𝑢𝑝)

2

(A.28)

where 𝜕𝜉

𝜕𝑝

1

𝑝𝑡1̿̿ ̿̿

(A.29)

𝜕𝜉

𝜕𝑝

𝑝

(𝑝𝑡1̿̿ ̿̿ )2

(A.30)

Flow coefficient:

Φ 𝑉𝑥,2Ω𝑟

(A.31)

𝑢Φ ±√(𝜕Φ

𝜕𝑉𝑥,2𝑢𝑉𝑥,2)

2

+ (𝜕Φ

𝜕Ω𝑢Ω)

2

+ (𝜕Φ

𝜕𝑟𝑢𝑟)

2

(A.32)

where 𝜕Φ

𝜕𝑉𝑥,2

1

Ω𝑟

(A.33)

149

𝜕Φ

𝜕Ω −

𝑉𝑥,2Ω2𝑟

(A.34)

𝜕Φ

𝜕𝑟 −

𝑉𝑥,2Ωr2

(A.35)

Loading coefficient:

ψ 𝑐𝑝Δ𝑇𝑡(Ω𝑟)2

(A.36)

𝑢ψ ±√(𝜕ψ

𝜕𝑇𝑡2𝑢𝑇𝑡2)

2

+ (𝜕ψ

𝜕𝑇𝑡3𝑢𝑇𝑡3)

2

+⋯

(𝜕ψ

𝜕Ω𝑢Ω)

2

+ (𝜕ψ

𝜕𝑟𝑢𝑟)

2

(A.37)

where 𝜕ψ

𝜕𝑇𝑡2

𝑐𝑝(Ω𝑟)2

(A.38)

𝜕ψ

𝜕𝑇𝑡3 −

𝑐𝑝(Ω𝑟)2

(A.39)

𝜕ψ

𝜕Ω −

2𝑐𝑝Δ𝑇𝑡Ω3𝑟2

(A.40)

𝜕ψ

𝜕𝑟 −

2𝑐𝑝Δ𝑇𝑡Ω2𝑟3

(A.41)

Blade inlet axial Reynolds number:

𝑅𝑒𝑥 𝜌2𝑉𝑥,2𝐶𝑥,𝐵

𝜇2

(A.42)

𝑢𝑅𝑒𝑥 ±√(𝜕𝑅𝑒𝑥𝜕𝜌2

𝑢𝜌2)2

+ (𝜕𝑅𝑒𝑥𝜕𝑉𝑥,2

𝑢𝑉𝑥,2)

2

+ (𝜕𝑅𝑒𝑥𝜕𝜇2

𝑢𝜇2)2

(A.43)

where 𝜕𝑅𝑒𝑥𝜕𝜌2

𝑉𝑥,2𝐶𝑥,𝐵𝜇2

(A.44)

𝜕𝑅𝑒𝑥𝜕𝑉𝑥,2

𝜌2𝐶𝑥,𝐵𝜇2

(A.45)

𝜕𝑅𝑒𝑥𝜕𝜇2

−𝜌2𝑉𝑥,2𝐶𝑥,𝐵(𝜇2)

2

(A.46)

150

Viscosity (Sutherland’s law):

𝜇 𝜇𝑟𝑒𝑓 (𝑇

𝑇𝑟𝑒𝑓)

3/2(𝑇𝑟𝑒𝑓 + 𝑆0)

(𝑇 + 𝑆0)

(A.47)

𝑢𝜇 ±

𝜕μ

𝜕𝑇𝑢𝑇

(A.48)

where

𝜕μ

𝜕𝑇 𝜇𝑟𝑒𝑓

(𝑇𝑟𝑒𝑓 + 𝑆0)

(𝑇 + 𝑆0)[1 2⁄ (𝑇 𝑇𝑟𝑒𝑓

⁄ )3/2

+3𝑆0

2⁄ (𝑇 𝑇𝑟𝑒𝑓⁄ )

1/2

]

(A.49)

Rotational Reynolds number:

𝑅𝑒𝜙 𝜌𝑟𝑐Ω𝑏

2

𝜇𝑟𝑐

(A.50)

𝑢𝑅𝑒𝜙 ±√(𝜕𝑅𝑒𝜙

𝜕𝜌𝑟𝑐𝑢𝜌𝑟𝑐)

2

+ (𝜕𝑅𝑒𝜙

𝜕Ω𝑢Ω)

2

+ (𝜕𝑅𝑒𝜙

𝜕𝜇𝑟𝑐𝑢𝜇𝑟𝑐)

2

(A.51)

where

𝜕𝑅𝑒𝜙

𝜕𝜌𝑟𝑐 Ω𝑏2

𝜇𝑟𝑐

(A.52)

𝜕𝑅𝑒𝜙

𝜕Ω 𝜌𝑟𝑐𝑏

2

𝜇𝑟𝑐

(A.53)

𝜕𝑅𝑒𝜙

𝜕𝜇𝑟𝑐 −

𝜌𝑟𝑐Ω𝑏2

(𝜇𝑟𝑐)2

(A.54)

Sealing effectiveness:

𝜀𝑐 𝑐 − 𝑐∞𝑐𝑠 − 𝑐∞

(A.55)

𝑢𝜀𝑐 ±√(𝜕𝜀𝑐𝜕𝑐

𝑢𝑐)2

+ (𝜕𝜀𝑐𝜕𝑐∞

𝑢𝑐∞)2

+ (𝜕𝜀𝑐𝜕𝑐𝑠

𝑢𝑐𝑠)2

(A.56)

where 𝜕𝜀𝑐𝜕𝑐

1

𝑐𝑠 − 𝑐∞

(A.57)

𝜕𝜀𝑐𝜕𝑐∞

−𝑐 − 𝑐∞

(𝑐𝑠 − 𝑐∞)2

(A.58)

𝜕𝜀𝑐𝜕𝑐𝑠

𝑐 − 𝑐∞

(𝑐𝑠 − 𝑐∞)2

(A.59)

151

Non-dimensional mass flow rate:

𝐶𝑤 �̇�

𝜇𝑟𝑐𝑏

(A.60)

𝑢𝐶𝑤 ±√(𝜕𝐶𝑤𝜕�̇�

𝑢�̇�)2

+ (𝜕𝐶𝑤𝜕𝜇𝑟𝑐

𝑢𝜇𝑟𝑐)2

(A.61)

where 𝜕𝐶𝑤𝜕�̇�

1

𝜇𝑟𝑐𝑏

(A.62)

𝜕𝐶𝑤𝜕𝜇𝑟𝑐

−�̇�

(𝜇𝑟𝑐)2𝑏

(A.63)

Non-dimensional mass flow rate:

𝜙 �̇�

2𝜋𝑠𝑐𝜌Ω𝑏2

(A.64)

𝑢𝜙 ±

√

(𝜕𝜙

𝜕�̇�𝑢�̇�)

2

+ (𝜕𝜙

𝜕𝑠𝑐𝑢𝑠𝑐)

2

+⋯

(𝜕𝜙

𝜕𝜌𝜌)

2

+ (𝜕𝜙

𝜕ΩΩ)

2

(A.65)

where 𝜕𝐶𝑤𝜕�̇�

�̇�

2𝜋𝑠𝑐𝜌Ω𝑏2

(A.66)

𝜕𝜙

𝜕𝑠𝑐 −

�̇�

2𝜋𝑠𝑐2𝜌Ω𝑏2

(A.67)

𝜕𝜙

𝜕𝜌 −

�̇�

2𝜋𝑠𝑐𝜌2Ω𝑏2

(A.68)

𝜕𝜙

𝜕Ω −

�̇�

2𝜋𝑠𝑐𝜌Ω2𝑏2

(A.69)

Design-Stage Uncertainty Values

The nominal values of the uncertainty calculations are shown in several tables on the

following pages. Due to the proprietary nature of many of these measurements the values were

generalized for this appendix.

152

Table A.0.1 Uncertainty in Vane Aerodynamic Loading

Parameter Measurement Description Uncertainty

[% meas]

Contribution of Individual Measurement to Uncertainty

First vane aerodynamic

loading (~1.0), 𝑝

𝑝𝑡1 (Δ𝑝1𝑉 + 𝑝𝑎𝑡𝑚)

(Δ𝑝𝑡1 + 𝑝𝑎𝑡𝑚)

Measuring Δ𝑝𝑡1 and Δ𝑝1𝑉 with 50 psi diff transducers (±0.05% FS),

referenced to atmospheric pressure, 𝑝𝑎𝑡𝑚, measured with a barometer

(±0.08% FS) 0.06%

uΔpt1

= 50% u

Δp1V = 50%

upatm

= 0%

First vane pressure loading

(~0.6),

(Δ𝑝1𝑉 + 𝑝𝑎𝑡𝑚)

(Δ𝑝𝑡1 + 𝑝𝑎𝑡𝑚)

Measuring Δ𝑝𝑡1 with 50 psi and Δ𝑝1𝑉 with 30 psi diff transducers (±0.05%

FS), referenced to atmospheric pressure, 𝑝𝑎𝑡𝑚, measured with a barometer

(±0.08% FS) 0.06%

uΔpt1

= 48% u

Δp1V = 47%

upatm

= 5%

Second vane pressure loading

(~1.0), 𝑝

𝑝𝑡3 (Δ𝑝2𝑉 + 𝑝𝑎𝑡𝑚)

(Δ𝑝𝑡3 + 𝑝𝑎𝑡𝑚)

Measuring Δ𝑝𝑡3 and Δ𝑝2𝑉 with 30 psi diff transducers (±0.05% FS),

referenced to atmospheric pressure, 𝑝𝑎𝑡𝑚, measured with a barometer

(±0.08% FS) 0.07%

uΔpt3

= 52% u

Δp2V = 48%

upatm

= 0%

Second vane pressure loading

(~0.6), 𝑝

𝑝𝑡3 (Δ𝑝2𝑉 + 𝑝𝑎𝑡𝑚)

(Δ𝑝𝑡3 + 𝑝𝑎𝑡𝑚)

Measuring Δ𝑝𝑡3 with 30 psi and Δ𝑝2𝑉 with 15 psi diff transducers (±0.05%

FS), referenced to atmospheric pressure, 𝑝𝑎𝑡𝑚, measured with a barometer

(±0.08% FS) 0.07%

uΔpt3

= 58% u

Δp2V = 33%

upatm

= 9%

153

Table A.0.2. Uncertainty in Efficiency

Parameter Measurement Description Uncertainty

[% meas]

Contribution of Individual Measurement to Uncertainty

Total-to-total efficiency, ηTT

Measuring Δ𝑝𝑡1 and Δ𝑝1𝑉 with 50 psi diff transducers (±0.05% FS),

referenced to atmospheric pressure, 𝑝𝑎𝑡𝑚, measured with a barometer

(±0.08% FS) 0.25%

upt1

= 7% u

pt4 = 13%

uTt1

= 33% u

Tt4 = 47%

Torque-based efficiency, ητ

Measuring 𝜏 with dynamometer (±1.2%), 𝛺 (±0.1%FS) with dynamometer, �̇�𝑖𝑛 with venturi (±0.3-5% reading), 𝑝𝑡1 with 100 psia abs transducers (±0.08% FS), 𝑝𝑡4 with 50 psi abs transducers (±0.08% FS), and 𝑇𝑡1 and 𝑇𝑡4 with TC's (±1R)

1.1%

uτ = 84%

uΩ = 1%

u�̇�𝑖𝑛 = 9% u

Tt1 = 1%

upt1

= 2% u

pt4 = 2%

Torque-based efficiency, ητ

Measuring 𝜏 with a torque meter (±0.2%), 𝛺 (±0.1%FS) with

dynamometer, �̇�𝑖𝑛 with venturi (±0.3-5% reading), 𝑝𝑡1 with 100 psia abs

transducers (±0.08% FS), 𝑝𝑡4 with 50 psi abs transducers (±0.08% FS), and

𝑇𝑡1 and 𝑇𝑡4 with TC's (±1R)

0.46%

uτ = 13%

uΩ = 4%

u�̇�𝑖𝑛 = 52% u

Tt1 = 7%

upt1

= 13% u

pt4 = 12%

154

Table A.0.3. Uncertainty in Flow Characterization Parameters

Parameter Measurement Description Uncertainty

[% meas]

Contribution of Individual Measurement to Uncertainty

Blade inlet axial Reynolds number, Re

x Measuring V

x2 with 5HP (max of 1 m/s or 1% meas), ρ: p with 100 psia

abs transducer (±0.08%FS) and T with TC's, and μ using Sutherland's law 1.5%

UVx2

= 7% u

ρ = 13%

uμ = 33%

Rotational Reynolds number, Re

φ Measuring Ω with dynamometer (±0.1% FS), ρ: p with 100 psia abs transducer (±0.08%FS) and T with TC's, and μ using Sutherland's law

0.3%

uΩ = 15%

uρ = 59%

uμ = 26%

First vane supply mass flow rate, ṁ

1VP Measuring Q with swirl flow meter (±0.5% FS accuracy), density: pressure

with 100 psia abs transducer (±0.08%FS) and temperature with

thermocouples 0.55%

First vane supply mass flow rate, ṁ

1VP Measuring Q with turbine flow meter (±1% FS accuracy), density:

pressure with 100 psia abs transducer (±0.08%FS) and temperature with

thermocouples 1.03%

First vane supply mass flow rate, ṁ

1VP Measuring ṁ directly with coriolis flow meter (±0.1-0.2% FS accuracy) 0.15%

155

Table A.0.4. Uncertainty in Sealing Effectiveness

Parameter Measurement Description Uncertainty Contribution of Individual

Measurement to Uncertainty

Sealing effectiveness, εc~0.05 Measuring 𝑐 and 𝑐∞ with low range (1000 ppm) and 𝑐𝑠 with high range

(10,000 ppm) (±1.0%FS) ±0.011

uc = 1%

uc∞ = 0% u

cs = 99%

Sealing effectiveness, εc~0.30 Measuring 𝑐∞ with low range (1000 ppm) and 𝑐 and 𝑐𝑠 with high range

(10,000 ppm) (±1.0%FS) ±0.013

uc = 67%

uc∞ = 0% u

cs = 33%

Sealing effectiveness, εc~0.60 Measuring 𝑐∞ with low range (1000 ppm) and 𝑐 and 𝑐𝑠 with high range

(10,000 ppm) (±1.0%FS) ±0.012

uc = 86%

uc∞ = 0% u

cs = 13%

Sealing effectiveness, εc~0.90 Measuring 𝑐∞ with low range (1000 ppm) and 𝑐 and 𝑐𝑠 with high range

(10,000 ppm) (±1.0%FS) ±0.011

uc = 98%

uc∞ = 1% u

cs = 1%

Sealing effectiveness, εc~1.0

Measuring 𝑐∞ with low range (1000 ppm) and 𝑐 and 𝑐𝑠 with high range

(10,000 ppm) (±1.0%FS) ±0.011

uc = 99%

uc∞ = 1% u

cs = 0%

156

VITA

Kenneth P. Clark

Kenneth Clark completed his Bachelor of Science in Mechanical Engineer at Brigham

Young University (BYU) in Provo, Utah in April 2009. He then began his graduate work under the

guidance of Dr. Steven Gorrell at BYU, performing high-fidelity computational fluid dynamics

simulations to study the interaction of compressor blade rows. He completed his Master of Science

degree in Mechanical Engineering at BYU in April 2011, and then began his doctoral work at the

Pennsylvania State University (Penn State), working with Dr. Karen Thole. Before graduating from

BYU, Ken was awarded a National Defense Science and Engineering Graduate (NDSEG)

Fellowship from the United States Department of Defense. Dr. Thole was working closely with

Pratt & Whitney and the US Department of Energy National Energy Technologies Laboratory

(DOE-NETL) to develop a new, state-of-the-art turbine research facility. Since the new facility was

still being designed and there was no research being performed, there was no funding for a student.

Because of the NDSEG Fellowship Ken was able to begin working on the new Steady Thermal

Aero Research Turbine (START) Laboratory for his PhD. Ken’s major roles were to develop the

test and instrumentation plan for the first phase of testing in the START Lab. He was able to

commission and run the rig, perform the experiments, and collect and analyze the data. For his

work on the START lab, Ken was awarded the Allan J. Brockett Student Award from Pratt &

Whitney in August 2015. In 2015 Penn State awarded Ken the College of Engineering

Distinguished Teaching Fellowship, which involved teaching the senior-level gas turbines class.

Ken completed his Doctor of Philosophy in Mechanical Engineering in December 2016, and

accepted an engineering position working at Pratt & Whitney in the compressor aerodynamics

group, where he hopes to contribute to producing efficient and dependable gas turbine engines.