An experimental investigation of clocking effects on turbine aerodynamics using a modern

AN EXPERIMENTAL INVESTIGATION OF CLOCKING EFFECTS ONTURBINE AERODYNAMICS USING A MODERN 3-D ONE AND ONE -HALF STAGE HIGH PRESSURE TURBINE FOR CODE DISSERTATION Presented in Partial Fulfi

AN EXPERIMENTAL INVESTIGATION OF CLOCKING EFFECTS ON

TURBINE AERODYNAMICS USING A MODERN 3-D ONE AND ONE

-HALF STAGE HIGH PRESSURE TURBINE FOR CODE

DISSERTATION

Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy

in the Graduate School of The Ohio State University

ByCharles W HaldemanIV

, M.S

*****

The Ohio State University

2003

Professor Michael G Dunn, Advisor

Professor Mohammad Samimy

Dr Robert Bergholz

Graduate Program in Aeronauticaland Astronautical Engineering

ABSTRACT

This research uses a modern 1 and 1/2 stage high-pressure (HP) turbine operating

at the proper design corrected speed, pressure ratio, and gas to metal temperature ratio togenerate a detailed data set containing aerodynamic, heat-transfer and aero-performanceinformation The data was generated using the Ohio State University Gas Turbine

Laboratory Turbine Test Facility (TTF), which is a short-duration shock tunnel facility.The research program utilizes an uncooled turbine stage for which all three airfoils areheavily instrumented at multiple spans and on the HPV and LPV endwalls and HPBplatform and tips Heat-flux and pressure data are obtained using the traditional shock-tube and blowdown facility operational modes Detailed examination show that theaerodynamic (pressure) data obtained in the blowdown mode is the same as obtained inthe shock-tube mode when the corrected conditions are matched

Various experimental conditions and configurations were performed, includingLPV clocking positions, off-design corrected speed conditions, pressure ratio changes,and Reynolds number changes The main research for this dissertation is concentrated onthe LPV clocking experiments, where the LPV was clocked relative to the HPV at severaldifferent passage locations and at different Reynolds numbers Various methods wereused to evaluate the effect of clocking on both the aeroperformance (efficiency) andaerodynamics (pressure loading) on the LPV, including time-resolved measurements,time-averaged measurements and stage performance measurements A general

improvement in overall efficiency of approximately 2% is demonstrated and could beobserved using a variety of independent methods Maximum efficiency is obtained whenthe time-average pressures are highest on the LPV, and the time-resolved data both in thetime domain and frequency domain show the least amount of variation The gain inaeroperformance is obtained by integrating over the entire airfoil as the three-dimensionaleffects on the LPV surface are significant

This experimental data set validates several computational research efforts thatsuggested wake migration is the primary reason for the perceived effectiveness of LPVclocking Previous experimental work that supported those computational researchefforts was not representative of modern turbine machines, and it was unclear before thiswork started, if the main mechanism hypothesized in the previous work would translate

to more complex machines In addition, this data set provides for the designer key

insight into the energy transfer that occurs in the time-resolved data between frequencies

as a function of clocking position While it was clear that these transfers did not affectthe overall efficiency of the machine, it is for the engine designers to know if these

transfers will add to a high-cycle fatigue or other structural problem Wake migrationpredictions do nothing for the engine designer’s concerns about structural problems Thislimits their information to either experimental data sets such as this one, or full resultsfrom 3D Navier-Stokes codes, which are almost as rare as the data sets

ACKNOWLEDGMENTS

This dissertation has been a long time in the creation Getting the chance to dothe work took years Tasks such as getting the laboratory running, taking classes, andkeeping up with contract research kept this dissertation from happening sooner And asany person submitting a dissertation will tell you, it would not be possible without amajor effort from a variety of people, all of who cannot be thanked enough for theirefforts

To begin with, without the help and support of my wife Margaret, none of thiswould have been possible She has endured this work for years, when I am sure that shethought there was no end in sight (and for a long time no beginning either!) She hasmade countless sacrifices for this work and for the OSU GTL in general, and I can neverthank her enough for her support

The rest of my family and friends from all phases of my life have also been

critical in helping me to keep this work in perspective, which is sometimes not easy to dowhen you are in the middle of it I am sure that most of my friends from the U of Rexpected me to be the first to get a Ph.D and I am grateful that they have not harassed metoo much about being the last

The seeds of this work actually date back to my days at MIT and I need to thankboth Prof Alan Epstein and Dr Gerry Guenette for my initial exposure to the world ofshort-duration experimentation The time I spent building the WPAFB ATARR facilitywith Dr Charles MacArthur was also a major contributor to some of the ideas and

techniques that have come to fruition in this dissertation and I would like to thank him forhis many insights into turbine research I would also like to thank the researchers atCalspan that could not come with us to OSU for contributions they made to the variousprojects all through the 1990’s that have developed key techniques and insights that arepresent in this work For the group that came to OSU, the fact that we broke all kinds of

time records in getting the new laboratory up and running was a direct result of

everyone’s hard work Getting the laboratory functional early on made this work

possible Jeff Barton, the laboratory’s facility manager, has been instrumental in keepingthe place running and translating ideas into working projects The data contained in thisdissertation would not have been possible without his help Nor would it have beenpossible without the distinct efforts of the rest of the support staff: Ken Copley, MichaelJones, and Packy Underwood, all of whom played key roles in this project There arealso several graduate and undergraduate students that have helped in this project over theyears Matt Krumanaker, who was responsible for the detailed heat-flux gauge

calibrations, did the most extensive work

I would also like to personally thank Colin Scrivener of Rolls-Royce and Prof.Mike Giles of Oxford University for their help with UNSFLO-2D Dr Scrivener wasinstrumental in obtaining formal permission for the OSU GTL to use UNSFLO-2D andProf Giles was extremely helpful in the resurrection of the code on our machines Ilearned a great deal about how these large codes are assembled from our discussions and

am deeply indebted to him for the time he spent with me on this work

I would like to personally thank Matt Weaver, Corso Padova and Prof RezaAbhari, for their friendship, guidance, and insight over the many phases of this work Iwould also like to thank the various groups that I have interacted with at GE over thecourse of this project for their guidance, insight, and patience The GE-USA program,which is the overall program under which this research was accomplished, has proven to

be a great success There are too many people to mention specifically, but I would like toacknowledge the contributions of David Wisler, Monty Shelton, Bob Bergholz, and FredBuck for all their help in this project and in making the GE-USA program the successthat it is at OSU Their support has involved much more than the financial support GEprovides to the project In addition Fred Buck, Bob Bergholz, and David Wisler havespent a great deal of their own time in reviewing the preliminary work presented in thisdissertation and their collective suggestions have been critical in the update of this work,which hopefully will make it easier for others to digest

Finally, I would like to thank my advisor and mentor, Prof Mike Dunn for themany years of collaboration, guidance, and support in both this dissertation and our pastprojects I consider myself lucky and privileged to have been able to work so closelywith one of the founders of this area of research

VITA

1963 Born- Lexington, Massachusetts

1985 B.S Mechanical Engineering, University of Rochester, Rochester NY

1989 M.S Aeronautical and Astronautical Engineering, Massachusetts Institute of

Technology

1990 M.S Technology and Policy Program, Massachusetts Institute of Technology1986-1990 Research Assistant, MIT Gas Turbine Lab1990-1996 Research Engineer, Calspan Corporation,

Buffalo New York1996-Present Senior Research Engineer, Ohio State

University Gas Turbine Lab

PUBLICATIONS

Archival Publications

“Time-Averaged Heat-flux for a Recessed Tip, Lip, and Platform of a Transonic Turbine

Blade”; M Dunn and C Haldeman, ASME Journal of Turbomachinery, Oct

2000, Vol 122, pp 692-698

“Influence of Vane/Blade spacing on the Heat Flux for a Transonic Turbine”; M Dunn,

C Haldeman, R Abhari, and M McMillan, ASME Journal of Turbomachinery,Oct 2000, Vol 122, pp 684-691

“Influence of Vane-blade spacing on Transonic turbine Stage Aerodynamics: Part I: Time

Resolved Data and Analysis” B Venable, R Delaney, J Busby, R Davis, D.Dorney, M Dunn, C Haldeman, R Abhari, ASME Journal of Turbomachinery,Oct 1999, Vol 121, pp.663-672

“Influence of Vane-blade spacing on Transonic turbine Stage Aerodynamics: Part II:

Time Resolved Data and Analysis”; J Busby, R Davis, D Dorney, M Dunn, C.Haldeman, R Abhari, B Venable, R Delaney, ASME Journal of

Turbomachinery, Oct 1999, Vol 121, pp.673-682

“High-Accuracy Turbine Performance Measurements in Short-Duration Facilities”; C

Haldeman and M Dunn, ASME Journal of Turbomachinery, Jan 1998, Vol 120,

pg 1-9

“Phase-Resolved Surface Pressure and Heat-Transfer Measurements on the Blade of a

Two-Stage Turbine”, M Dunn and C Haldeman, Journal of Fluids Engineering,December 1995, Vol 117, pg 653-658

Refereed Conference Papers

“Influence of Clocking and Vane/Blade Spacing on the Unsteady Surface Pressure

Loading for a Modern Stage and One-Half Transonic Turbine”, C Haldeman, M.Krumanaker, and M Dunn, GT2003-38724 (accepted for publication in theASME Journal of Turbomachinery)

“Heat Transfer Measurements and Predictions for the Vane and Blade of a Rotating

High-Pressure Turbine Stage”, C Haldeman and M Dunn, GT2003-38726

(accepted for publication in the ASME Journal of Turbomachinery)

“Experimental Investigation of the Aerodynamic Effects of Clocking Vanes and Blade

Rows in a 1/3 Scale Model Turbine”, D B M Jouini, D Little, E Bancalari, M.Dunn,C Haldeman, P.D Johnson, GT2003-38872, Proceedings of the ASMEInternational Gas Turbine Institute, Turbo-Exposition, Atlanta, GA, June 15-19,2003

“Unsteady Interaction Between a Transonic Turbine Stage and Downstream

Components”, R.L Davis, J Yao, J.P Clark, G Stetson, J.J Alonso, A Jameson,C.W Haldeman, M.G Dunn, 2002-GT-30364, Proceedings of the ASME

International Gas Turbine Institute Turbo-Exposition, Amsterdam, NetherlandsJune 3-6, 2002

“The Effect of Airfoil Scaling on the Predicted Unsteady Pressure Field in a 1+1/2 Stage

Transonic Turbine and a Comparison with Experimental Results”, J.P Clark,G.M Stetson, S.S Magge, C.W Haldeman, M.G Dunn, 2000-GT-0446,

Proceedings of ASME International Gas Turbine Institute Turbo-Expo, Munich,Germany May 8-11, 2000

“Experimental and Computational Investigation of the Averaged and

Time-Resolved Pressure Loading on a Vaneless Counter-Rotating Turbine”, C

Haldeman, M Dunn, R Abhari, P Johnson, and X Montesdeoca, 2000-GT-0445,Proceedings of the ASME International Gas Turbine Institute Turbo Expo,

Munich, Germany May 8-11, 2000

“Time-Resolved and Time-Averaged Pressure and Heat-Transfer Measurements on the

Blade of a Two-Stage Turbine” M Dunn and C Haldeman, International

Symposium on Unsteady Flows in Aeropropulsion: Recent Advances in

Experimental and Computational Methods, 1994 ASME Winter Annual Meeting,Chicago, IL, November 6-11, 1994

Conference Papers

“Summary of Time-Averaged and Phase-Resolved Pressure Measurements on the First

Stage Vane and Blade of the SSME Fuel-Side Turbine”, 1994 Earth to OrbitConference, Huntsville, AL, May 1994

“The USAF Advanced Turbine Aerothermal Research Rig (ATARR)”; C Haldeman, M

Dunn, C MacArthur and C Murawski; AGARD Conference

Proceedings 527, Heat Transfer and Cooling in Gas Turbines, 1992

“Uncertainty Analysis of Turbine Aerodynamics Performance Measurements in

Short-duration Test Facilities” C Haldeman, M Dunn, J Lotsof, C MacArthur, and Lt

B Cohrs; Paper No AIAA-91-2131 AIAA/SAE/ASME/ASEE 27thJoint

Propulsion Conference, June 24-26, 1991, Sacramento, CA

FIELDS OF STUDY

Major Field: Aeronautical and Astronautical Engineering

TABLE OF CONTENTS

ABSTRACT II ACKNOWLEDGMENTS IV VITA VII LIST OF FIGURES XV LIST OF TABLES XXII NOMENCLATURE AND ABBREVIATIONS XXIV CHAPTER 1 THE RESEARCH PROGRAM AND BACKGROUND INFORMATION 1

1.1 THE RESEARCH PROJECT 1

1.1.1 Goals of the experimental program 2

1.1.2 Components of the first phase of the project 3

1.2 THE THESIS STATEMENT 5

1.3 ORGANIZATION OF THESIS 7

CHAPTER 2 CURRENT RESEARCH INTERESTS IN CLOCKING, AERO-PERFORMANCE, AERODYNAMICS, AND HEAT TRANSFER MEASUREMENTS 9 2.1 GENERAL REVIEW 9

2.2 LOW PRESSURE TURBINE VANE (LPV) CLOCKING LITERATURE REVIEW 10

CHAPTER 3 EXPERIMENTAL FACILITIES, AND THE TURBINE RIG/MODEL 16 3.1 THE OSU GAS TURBINE LABORATORY TURBINE TEST FACILITY (TTF)16

3.1.1 Basic Description 16

3.1.2 TTF Main characteristics and Operation 17

3.2 THE TURBINE RIG 22

3.2.1 General description and Instrumentation 22

3.2.2 Data acquisition and calibrations 26

3.3 LPV CLOCKING EXPERIMENTAL METHODS 30

CHAPTER 4 DATA DESCRIPTION AND CLOCKING ANALYSIS TECHNIQUES 32 4.1 GENERAL DATA EXAMPLE 33

4.2 MAJOR DATA ANALYSIS TECHNIQUES 34

4.2.1 Normalization of Data 35

4.2.2 Filtering 38

4.2.3 Running averages 39

4.2.4 Airfoil passing envelopes 39

4.2.5 FFT processing and Windowing 42

4.2.6 Uncertainty/error propagation 43

4.3 DATA ANALYSIS/ FACILITY MODELS 45

4.3.1 Time-delay 46

4.3.2 Speed data Processing 48

4.3.3 Mass-Flow and Temperature 49

4.4 SUMMARY 52

CHAPTER 5 AERODYNAMIC DATA AT ONE CLOCKING POSITION 54

5.1 DATA SET DEVELOPMENT AND METHODOLOGY 55

5.2 TIME-AVERAGE PRESSURE 59

5.3 TIME-RESOLVED PRESSURE ENVELOPES AND FFT’S 62

5.4 MODELING UNSTEADY PRESSURE DATA WITH UNSFLO-2D 79

5.5 REVIEW OF RESULTS 87

CHAPTER 6 LPV CLOCKING AERO-PERFORMANCE DATA 90

6.1 DATA SET DEVELOPMENT AND METHODOLOGY 91

6.1.1 Methodology 94

6.1.2 Data Presentation 99

6.2 INTEGRATED CLOCKING EFFECTS 100

6.2.1 Efficiency Measurements 100

6.2.2 Summary of Integrated-clocking Effects 108

6.3 TIME-AVERAGED EFFECTS 111

6.3.1 Clocking at Low Inlet Pressure, Re = 4.64 x106 113

6.3.2 Reynolds Number Effects on Time-Averaged Data 123

6.4 TIME-RESOLVED EFFECTS 129

6.4.1 Technique Descriptions 130

6.4.2 Time-Resolved Clocking Data for Low Pressure, Re = 4.64 x106 131

6.5 COMPARISON OF CLOCKING TECHNIQUES 163

CHAPTER 7 LPV CLOCKING MODELING AND EXPERIMENTAL INTEGRATION AND IMPLICATIONS FOR FUTURE WORK 168

7.1 LPV CLOCKING MODELS 168

7.2 INTEGRATION OF RESULTS 170

7.3 IMPLICATIONS FOR FUTURE WORK 174

7.4 GENERAL CONCLUSIONS OF RESEARCH 176

APPENDIX A HISTORICAL PERSPECTIVE: THE ESTABLISHMENT OF SHORT-DURATION FACILITIES IN GAS TURBINE RESEARCH 178

APPENDIX B UNSTEADY DATA DESCRIPTION AND ANALYSIS TECHNIQUES185 B.1 OVERVIEW: DEVELOPMENT OF A BENCHMARK DATA SET 185

B.2 DATA ANALYSIS-BASIC DEFINITIONS 188

B.2.1 Time-scales 188

B.2.2 Time-averaged vs Time-accurate Data 190

B.2.3 Time-windows 191

B.2.4 Time-accurate data: time space, encoder space, and frequency data 192

B.2.5 Uncertainty bands and noise bands 196

B.2.6 Averaging Data Points 201

B.2.7 Uncertainty/error propagation 205

B.3 MAJOR DATA ANALYSIS TECHNIQUES 206

B.3.1 Normalization of Data 206

B.3.2 Airfoil passing envelope technique development 216

B.4 SUMMARY 223

APPENDIX C DERIVATION OF AERODYNAMIC AND HEAT-FLUX DATA SET-POINTS 224

C.1 DATA SET DEVELOPMENT AND METHODOLOGY 225

C.1.1 Heat-flux Experimental Set Points 228

C.1.2 Pressure Experimental Set-Points 240

C.2 TIME-AVERAGE PRESSURE 244

C.3 TIME-AVERAGED HEAT-FLUX AT DIFFERENT REYNOLDS NUMBERS251 C.4 TIME-RESOLVED HEAT-FLUX 256

C.5 REVIEW OF RESULTS 261

APPENDIX D SUPPORTING AERO-PERFORMANCE LPV CLOCKING DATA 262

D.1 DATA SET DEVELOPMENT AND METHODOLOGY 265

D.1.1 Sub-matrix Description 265

D.1.2 Methodology 267

D.2 INTEGRATED CLOCKING EFFECTS 294

D.2.1 Measures of clocking 294

BIBLIOGRAPHY 315

LIST OF FIGURES

Figure 3.1 TTF Layout 17

Figure 3.2 Picture of TTF Looking Towards Dump Tank 18

Figure 3.3 Sketch of Overall Rig 23

Figure 3.4 Picture of Rig (Reversed from Sketch) Ready to Go Into the Dump Tank 23

Figure 3.5 Main Flowpath 24

Figure 3.6 DAS System Preamplifiers/Power Supplies and Patch Panel 28

Figure 3.7 Main DAS systems 28

Figure 4.1 Example Data Time-Trace with Key Components Marked 33

Figure 4.2 Raw Time Data for PR42 40

Figure 4.3 Example of Time Based Ensemble Plots 42

Figure 4.4 Example of FFT Interpolation 43

Figure 4.5 Time-Delay Example 47

Figure 4.6 Wobble Model and Speed Averaging 48

Figure 4.7 Examples of Inlet Mass Flow and Temperature Calculations 51

Figure 4.8 Measured and Calculated Temperatures 51

Figure 4.9 Effect of Temperature Change on Time-Window 52

Figure 5.1 Reynolds Number and Pressure Ratio Variation for Groups 58

Figure 5.2 Pressures for All Experimental Cases 60

Figure 5.3 HPV 50% Time-Averaged Normalized Pressures with Envelopes 63

Figure 5.4 HPV 50% Span Time Averages with Absolute FFT's 64

Figure 5.5 HPV 50% Span with Normalized FFT's 65

Figure 5.6 HPB 50% Span FFT's and Ensemble Plots (A) 68

Figure 5.7 HPB 50% Span FFT's and Ensemble Plots (B) 69

Figure 5.8 HPB 50% Span FFT's and Ensemble Plots (C) 70

Figure 5.9 HPB 50% Span FFT's and Ensemble Plots (D) 71

Figure 5.10 HPB 50% Span FFT's and Ensemble Plots (E) 72

Figure 5.11 HPB 50% Span FFT's and Ensemble Plots (F) 73

Figure 5.12 HPB 50% Span FFT's and Ensemble Plots (G) 74

Figure 5.13 LPV 50% Span FFT's and Ensemble Plots (A) 75

Figure 5.14 LPV 50% Span FFT's and Ensemble Plots (B) 76

Figure 5.15 LPV 50% Span FFT's and Ensemble Plots (C) 77

Figure 5.16 LPV 50% Span FFT's and Ensemble Plots (D) 78

Figure 5.17 UNSFLO Predictions HPV and HPB 50% 85

Figure 6.1 Change in Efficiencies for Time Window "O" 102

Figure 6.2 Efficiencies based on local Group 103

Figure 6.3 Mechanical Efficiency over Time window 104

Figure 6.4 Traditional Windowing Technique Results 105

Figure 6.5 Influence of Time Window on Time-Averaged Measurements 111

Figure 6.6 Percentage Change from Average for Run 9 112

Figure 6.7 MGroup 1 Average of Time-Averaged Data for Time Window P 114

Figure 6.8 Variation Due To Clocking, MGroup 1, HPV all Spans 115

Figure 6.9 Variation Due To Clocking, MGroup 1, HPB all Spans 116

Figure 6.10 Variation Due To Clocking, MGroup 1, LPV all Spans 117

Figure 6.11 Sample integration Scheme 121

Figure 6.12 Reynolds Effect on Time-average Values for Airfoils at 50% span 125

Figure 6.13 Effect of Time Window on Run 9, LPV 50% Span 126

Figure 6.14 Cocking Effects of Time Averaged Data Due to Reynolds Number 127

Figure 6.15 Average of Normalized Pressure Envelopes 132

Figure 6.16 Variation in Envelope Size due to Clocking MGroup 1, HPV all Spans 134

Figure 6.17 Variation in Envelope Size due to Clocking MGroup 1, HPB all Spans 135

Figure 6.18 Variation in Envelope Size due to Clocking MGroup 1, LPV all Spans 136

Figure 6.19 Periodic Envelopes for LPV 50% span (0 to -20% WD) 139

Figure 6.20 Periodic Envelopes for LPV 50% span (-26 to –58 % WD) 140

Figure 6.21 Periodic Envelopes for LPV 50% span (26 to 40 % WD) 141

Figure 6.22 Variation in FFT Amplitude for HPV All spans 143

Figure 6.23 Variation in FFT Amplitude for HPB All spans 145

Figure 6.24 Variation in FFT Amplitude for LPV All spans 146

Figure 6.25 Variation Due to clocking for Fundamental Amplitude HPV all spans 148

Figure 6.26 Variation Due to clocking for Fundamental Amplitude HPB all Spans 149

Figure 6.27 Variation Due to clocking for Fundamental Amplitude LPV all Spans 150

Figure 6.28 Variation Due to clocking for First Harmonic Amplitude HPV all Spans 151

Figure 6.29 Variation Due to Clocking for First Harmonic Amplitude HPB 152

Figure 6.30 Variation Due to Clocking for First Harmonic Amplitude LPV all Spans 153

Figure 6.31 Variation Due to Clocking for Second Harmonic Amplitude HPV 154

Figure 6.32 Variation Due to Clocking for Second Harmonic Amplitude HPB 155

Figure 6.33 Variation Due to Clocking for Second Harmonic Amplitude LPV 156

Figure 6.34 Comparison of Amplitude Changes for LPV (10% span) 158

Figure 6.35 Comparison of Amplitude Changes for LPV (50 and 90% spans) 159

Figure 6.36 Comparison of Amplitude Changes for MGroup 6 at LPV 10% span 161

Figure 6.37 Comparison of Amplitude changes for MGroup 6 and LPV at 50% and 90% spans 162

Figure 6.38 Comparison of Clocking Methods 165

Figure B.1 Time-Resolved Data for 1 Revolution, Downstream Total Pressure Rake 193

Figure B.2 Average Downstream Total Pressure over 1 ms 194

Figure B.3 Power Spectra of Downstream Total Pressure over 1 Revolution 195

Figure B.4 Distribution of Deviations from Average for PTDA 198

Figure B.5 Distribution of Deviations from Average for PTDA filtered at 25KHz 199

Figure B.6 Time-Resolved Deviation data from Average for PTDA 200

Figure B.7 Distribution of Deviations from Average for PTDA (Different Filtering) 201

Figure B.8 Pressure Ratios Using Different Averaging 203

Figure B.9 Sample Pressure Calibration (Dump Tank Pressure vs Time) 208

Figure B.10 Sample Calibration Data 209

Figure B.11 Initial Pressures, Run 8 Entry 1 210

Figure B.12 Base Line Resistances 212

Figure B.13 Comparison of Heat-Flux Gauge Temperatures 213

Figure B.14 Sample Types of Heat-flux Temperatures 214

Figure B.15 Raw Time Data for PR42 217

Figure B.16 Sample Data Showing Interpolation Attenuation 219

Figure B.17 Example of Time Based Ensemble Plots 220

Figure B.18 Comparison of Techniques for Estimating Periodic Envelope 221

Figure C.1 Reynolds Number Variation for Heat-flux Experimental Points 236

Figure C.2 Pressure Ratio and Corrected Speed Variation for Design Points 237

Figure C.3 Pressure Ratio and Inlet Temperature Variations for Heat-flux Experimental Points 237

Figure C.4 Normalized Pressure Averages for Different Blowdown Conditions (HPV)245 Figure C.5 Comparison of Uncertainties for HPV Data 246

Figure C.6 Normalized Pressure For Different Blowdown Conditions (HPB and LPV)247 Figure C.7 Pressure For Shock Tube Run by Heat-flux Experimental Group 248

Figure C.8 Pressures for All Experimental Cases 250

Figure C.9 Time-Averaged Heat-Flux for 50% Span Locations on the Three Airfoil Rows as a Function of Reynolds Number 253

Figure C.10 Exponent in Stanton Number/Reynolds Number Relationship 255

Figure C.11 HPV 50% span Power Spectrum, Selected Gauges 258

Figure C.12 LPV 50% span Power Spectrum, Selected Gauges 259

Figure C.13 Comparison between Runs 260

Figure D.1 Design Parameter Variation over 60 ms and 1 Rotor Revolution 269

Figure D.2 Linear Correlation of Pressure Ratio and Speed 270

Figure D.3 Global Discharge Coefficient by Group 271

Figure D.4 Global discharge Coefficients as a Function of Choke Positions 272

Figure D.5 Percent Variation in Pressure Ratios and Speed 273

Figure D.6 Method for Calculating Flow Starting Times 274

Figure D.7 Initial Starting Times for Flow in Rig 275

Figure D.8 Design Properties as a Function of time Window for Re=4.17E6 276

Figure D.9 Relative Errors for HPV, HPB, and LPV at 50% Span 290

Figure D.10 Uncertainty Estimates for Time Average Variations for MGroup 1 293

Figure D.11 Initial Speed Accuracy 295

Figure D.12 Percent Variation in Speed for Time Window "O" 296

Figure D.13 Variation in Speed From Group Average (based on individual windows) 298 Figure D.14 Example of Possible True Variation Data 299

Figure D.15 Example of effect of Measurement Sampling on Observed Variation 300

Figure D.16 Pressure Ratio Changes for Time window “O” 301

Figure D.17 Pressure Ratio Changes for Different Groups 302

Figure D.18 Change in Efficiencies for Time Window "O" 304

Figure D.19 Efficiencies based on local Group 305

Figure D.20 Mechanical Efficiency over Time window 306

Figure D.21 Traditional Windowing Technique Results 307

Figure D.22 Frequency Estimation Accuracy for Bound Group 2 (Time Variations) 311

Figure D.23 Frequency Estimation Accuracy for Bound Group 1 (Clock D Variations) 312

xxiFigure D.24 Uncertainty Ranges for FFT Analysis 314

LIST OF TABLES

Table 3.1 Instrument Count/Location 25Table 3.2 Clocking Position Mapping (Text to Numeric) 30Table 5.1 Experimental Conditions Used for Heat-Flux and Pressure Data 57Table 5.2 Overall Variation in “Constant” Conditions For all Experimental Groups 59Table 6.1 Entry 1 Experiments 92Table 6.2 Comparison of Time window Selection on Design Properties 97Table 6.3 Integral Effects of Time-Averaged Data Variation by Span on the LPV 122Table 6.4 Integral Effect for Time-Averaged Data based on Reynolds Number 128Table 6.5 Integral Effect for Percent Variation of Time-Averaged Data based on

Reynolds Number 129Table 6.6 Integral Effects of Envelope Changes by Span on the LPV 137Table 6.7 Comparison of FFT amplitudes over blade surface 160Table B.1 Summary of Calibration Results for one Sub-matrix 210Table B.2 List of number of Heat-Flux Sensors Requiring Pre-processing 215Table C.1 Experimental Conditions Used for Heat-Flux and Pressure Data 227Table C.2 Experimental Conditions for Group 2 231Table C.3 Experimental Conditions for Group 3 232Table C.4 Experimental Conditions for Group 4 233Table C.5 Experimental Conditions for Group 5 234

Table C.6 Overall Variation in Heat-flux Conditions 235Table C.7 Experimental Group 3 at Lower Corrected Speed Conditions 239Table C.8 Experimental Conditions for Group 1, 1a 241Table C.9 Experimental Conditions For Group 6 242Table C.10 Overall Variation in Conditions For all Experimental Groups 243Table D.1 Entry 1 Experiments 265Table D.2 Time Window "O" Design Values 278Table D.3 Time Window "O" Design Values for Modified Groups 279Table D.4 Entry 1 Time Window Listings 280Table D.5 Time Windows for Matching Equivalent clock positions 281Table D.6 Time Window "P" Design Values for Modified Groups 283Table D.7 Time Window "N" Design Values for Modified Groups 284Table D.8 Time Window "M" Design Values for Modified Groups 285Table D.9 Comparison of Time window Selection on Design Properties 286Table D.10 Clocking Position Mapping (Text to Numeric) 297

NOMENCLATURE AND ABBREVIATIONS

TTF Turbine Test Facility at the Ohio State University Gas Turbine Lab (OSU GTL)FAV Fast Acting Valve – This is the valve that separates the driven tube from the testsection on the TTF

HPV High Pressure Turbine Vane

HPB High Pressure Turbine Blade

LPV Low Pressure Turbine Vane

FFT Fast Fourier Transform (numerical algorithm)

CHAPTER 1 THE RESEARCH PROGRAM AND BACKGROUND INFORMATION

1.1 The Research Project

This dissertation examines the aerodynamic effects of Low Pressure TurbineVane (LPV) clocking in a stage and one half full-scale rotating turbine operating atdesign corrected conditions This is the first time this type of work has been done wherethe effects measured locally on the airfoils have been compared to the overall machineperformance This also represent the first time that highly detailed efficiency changeshave been measured using a short-duration facility This dissertation is part of a large,multi-year and multi-faceted research effort conducted at The Ohio State University(OSU) Gas Turbine Laboratory (GTL) that was designed to provide a variety of

experimental results used for the development of advanced research and developmentdesign systems GE Aircraft Engines (GEAE) as part of the GEAE University StrategicAlliance (USA) Program supports this effort Within GEAE, there is a wide variety of

“customers” for the detailed physical information generated from this experimentalprogram These range from design support groups requiring applicable data to validatecomputational models, to engineers trying to understand some of the basic physics

involved in advanced engine concepts Sometimes the main use of the information is just

to answer a question experimentally; “Does this design produce less heat-transfer at thetip?” or “Does this design result in destructive high-cycle fatigue?” Other times it is toverify that a specific manufacturing process could produce an observed behavior in anengine Another very basic need is to generate a quality data set that matches the propercorrected conditions, with defined control variables becoming the basis for either betterunderstanding of the physics, or just the beginning of an empirical design tool In thissection, the individual goals of the experimental program will be outlined and the

components of the first phase of this research effort will be discussed which is the basis

of this dissertation

1.1.1 Goals of the experimental program

The research effort with which this dissertation is associated is a complex year program utilizing a modern stage and 1/2 high-pressure turbine for a variety ofresearch purposes The turbine design is a highly loaded turbine with extremely strong3D effects on each airfoil row, and an interconnecting “S” duct between the HPB and theLPV The hardware is configured to make changing components “relatively1

multi-” easy inbetween builds so that different types of hardware configurations can be investigated.The ultimate goal of this research activity is to obtain experimental results for complexengine geometries operating at design corrected conditions in a fully cooled mode (vane,blade, endwalls, etc.) The current work utilizes the uncooled version of the hardware.The data from these experiments and the future cooled experiments will be used in one ofthree ways:

1) As general design code verification In this case, data is taken in a knownconfiguration and is used to validate design codes used in the design of turbine hardware

2) As a verification of industrial research codes and techniques It is not unusualfor the industry to have several different types of codes that have similar goals, butperhaps do the task in different ways There is a steady migration of the industrial

research codes into the design code classification Currently, the industry is capable of agreat deal more in their research codes than they are in the design codes, due to thecomplex nature of these research codes and the amount of computer resources required tomake them operate

3) Finally, as a valid data set from which empirical theories, new models, or justbasic knowledge is developed This is not a small matter, particularly in the area ofcooling flow where for many cases it is unclear how the various components of blowing

1

Relatively is a key word For most facilities, the change of an airfoil or instrumentation

on an airfoil requires a complete teardown of the rig The same is true for this rig, butdue to improvements in the overall instrumentation capabilities and rotor wiring, this can

be done in a matter of weeks and not months

ratios, temperature ratios, and measured heat flux relate to overall design properties such

as cooling film effectiveness

The research program was separated into two distinct phases at the beginning,although since then it has developed further The first stage was the design and

construction of rig hardware and the initial measurement program that utilized uncooledengine hardware The second stage is the adaptation of the hardware to a fully cooledsystem with a temperature profile generator, turbulence generator, and traversing rings.The first phase was designed to obtain data (of which this dissertation is a part) to be used

as verification of the design tools and industrial research codes, and as a reference dataset for the cooling experiments The future cooling experiments are designed to providedata sets to be used for the second and third modes (described above)

1.1.2 Components of the first phase of the project

This dissertation, as discussed previously, is concerned with the measurementprogram associated with the first phase of the research effort: the uncooled experiments

As described in section 3.1 the OSU GTL Turbine Test Facility (TTF) uses either actualengine parts or “rig” parts, thus the hardware used is the same physical size as engineparts “Rig” hardware is a term used in the industry to imply that the parts have beenmodified slightly for their experimental facilities (rigs), which generally operate at

slightly different conditions than the real engines One example is that an uncooled vanering will often have the vanes rotated slightly so that the mass flow through the uncooledmodel will match the core mass flow through the cooled engine parts Another example

is that blades might be tipped a little differently so that they run at the same relative gap

as an engine operating at high temperatures and speeds Generally, the differences

between “rig” hardware and engine hardware are quite small Sometimes the TTF willoperate a rig configuration that is not an operating engine, but rather a specific piece ofhardware designed either to make computational work simpler or so that a large group of

industrial government sponsors can share the coordinates and the data [1].

The actual engine or rig hardware is generally mounted in a device that holds theturbine stage Sometimes the disk may be engine hardware, often it is not since making aheavier disk will reduce the acceleration rate and simplify the measurement program

The overall system would be referred to in the engine industry as a rig, but in the

academic world, it is often called a model However, there is nothing small about thesemodels, as they are full sized, high-speed rotating rigs In this case, the overall “model”was thirteen ft long and four ft in diameter

For the first phase of the experiments, the uncooled model was heavily

instrumented with both surface pressure and heat-flux sensors on all three airfoil rows(the high pressure vane and blade and the low pressure vane) over three different spans.There were both types of sensors installed on the endwalls of both vane rings, on therotor platform and on multiple and differing blade tip configurations The main goals ofthe experiments were: (1) to obtain detailed surface pressure distributions as a function oflow-vane clocking (more about that later) and as a function of corrected speed and

Reynolds number and (2) to obtain detailed heat-transfer distributions at design point for

a range of Reynolds numbers In addition, research was performed with different shroudconfigurations to examine their effect on blade unsteady pressure loadings

Much of the data acquired from this wide variety of experiments will be analyzedand used as part of the work of other OSU GTL students Some examples are blade tipbehavior, heat-flux scaling, three-dimensional effects and performance calculations.Because of limitations of the experimental apparatus, instrumentation, and data

acquisition channels, the first phase of the experiment (called Build 1) was split into twodifferent experimental assemblies (called entries) that occurred about a year apart Entry

1 was done primarily to obtain an aerodynamic database and using primarily pressuresensors These were the experiments that dealt with the clocking data discussed in thisdissertation Entry 2 was performed to obtain a detailed heat transfer database as afunction of Reynolds number To create a consistent database, both the entry 1 and entry

2 data need to be examined together However, in this dissertation, the entry 2 heat-fluxdata is presented only in the appendices since the actual data does not directly impact themain dissertation, which is aeroperformance measurement of Low Pressure Turbine Vane(LPV) clocking Varying the design corrected conditions during this experimental matrixhad the effect of making this rig operate like different engines in the field, which is also

of great interest

1.2 The Thesis Statement

This dissertation is an experimental examination of LPV clocking on

aeroperformance These results are critical since it represents the first data published for

a high-pressure ratio turbine where clocking effects can be observed both on the localairfoil surface pressures and over the entire machine in terms of measured efficiencydifferences The discussion addresses several shortcomings in previous experimentalwork by more fully documenting the uncertainty analysis, which is so critical to theinterpretation of the final results In addition, it resolves several unresolved issues frompast work done at The Ohio State University Gas Turbine Lab (OSU GTL) where

clocking effects were observed using one method of investigation, but could not beconfirmed using a separate measure from the same data set The effects of LPV clockingfor an uncooled turbine stage are substantial, yielding about a ±2% change based on therelative position of the High Pressure Turbine Vane (HPV) and the LPV These resultsare consistent whether one looks at the time averaged data or the time resolved data, or ifone looks at the local pressure on the LPV, or the overall machine performance Thiswork supports conclusions drawn by previous research that suggest that the two primarymechanisms by which LPV clocking operates are the wake propagation throughout themachine, and the decrease in unsteady effects (time-resolved pressure envelopes) whichreduce the loses due to reducing the velocities at the LPV and the turbulence generated atthe LPV However, this data set provides an important piece of information which is notaccounted for in these models which is the frequency transfer that occurs in the time-resolved data due to clocking which will impact blade designers not so much from anefficiency stand-point, but rather form a life-cycle perspective

It is important to realize that resolving accurately a ±2% variation in a propertyrequires much higher resolution of the underlying measurements For a ±2% efficiencymeasurement to be repeatable and believable requires essentially an order of magnitudeincrease in the accuracy (0.2%) of the underlying measurements This was essentiallynot possible without better understanding of the underlying facility operation,

improvements in the data reduction process, and improvements to the basic

instrumentation The results of this work not only are significant for the underlying

results surrounding the effectiveness of clocking in highly radial divergent turbine stages(“S” ducts), but also in the fact that to achieve these measurements required that theshort-duration facilities be used in a mode where efficiencies can be measured to thislevel of precision This has long been a goal for these types of facilities, but has proved

to be an elusive one, and this work represents a milestone in that regard

To support the LPV clocking results, this dissertation will also address severalsupporting issues

1 First, it will be the platform for documenting the main characteristics of the dataset for others to use In this regard, some might be interested in time-averageproperties, others may be interested in high-cycle fatigue, and still others may beinterested in the data for the cooling studies yet to come The data presented, bothaerodynamic and heat-flux that will generally be limited to 50% span, will bebased on specific techniques outlined so that others may interpret the data

correctly In addition, some work will be done to explain how multiple runs withdiffering instruments can be combined to create one data set This is critical sincethis technique is one of the new innovations that allows improvements in theunderlying measurements

2 Secondly, the techniques developed with respect to the uncertainty analysis, andthe data reduction techniques will be discussed in detail These are critical sinceone of the main failure of past published work related to clocking has been in theproper interpretation of the uncertainty analysis

This data set can be used to determine the effect of clocking on aeroperformance

in multiple ways: from an overall stage efficiency perspective and based on individualairfoil pressure distributions in both a time resolved and time-averaged reference frame

It will be demonstrated that these measurements yield similar results Clearly there aremany more topics associated with this measurement program that could be investigated inmore detail than will be presented in this dissertation, but those will be left for others toexplore The experimental data acquired in the course of this work has shown that

experimentalists using short-duration facilities have progressed to the point where the

data acquisition/processing and the stability of the facilities allows them to be used asmajor sources of data for physics-based modeling and code verification In fact, themajor technical questions regarding turbine development (unsteady effects, coolingissues, etc) can only be done in environments where the full time-resolved nature of theflow can be investigated During the course of this dissertation, the data set quality,particularly with respect to the uncertainty analysis, has quantified how well the designset point can be established in a short-duration facility The investigations into the effect

of LPV clocking have forced the data resolution to new levels that has brought to

conclusion a long-standing process where the short-duration facility has been transformedfrom a qualitative research tool into a quantitative one In addition, the ability of theadvanced 3D codes to predict the complicated pressure distribution in a 3D turbine areimproving, but simplified flow models such as the UNSFLO-2D code (used at the OSUGTL by special arrangement with Rolls-Royce England) can be used to rapidly obtaintime-averaged and unsteady aerodynamic and heat-transfer predictions An example ofwhich is shown in this work

1.3 Organization of Dissertation

For a work of this size, complexity, and various audiences, one could envision avariety of ways to present the data and to organize the dissertation A decision has beenmade to take this work and draw the main connections to the LPV clocking work

throughout the dissertation, and to subjugate some of the supporting information, such asthe uncertainty analysis and the heat-flux data, to supporting appendices This

dissertation is split into two different sections, each being composed of chapters andsome supporting appendices The first section provides the background for those readerswho are not familiar with short-duration facilities or the development of this type of timeresolved data set Chapter 2 highlights the current research interests and reviews thecurrent literature with respect to clocking studies Chapter 3 deals specifically with themain facility (the TTF) and the experimental hardware used in this research Chapter 4will finish this section with a general discussion of the description of unsteady data andthe main techniques that are used for analysis and data presentation throughout the

dissertation Two appendices support this section Appendix A discusses the

development of the short-duration facility and may be useful to those interested in theconnections between the different researchers in this field and the rise in importance ofthe short-duration facility for turbine research work Appendix B discusses in more detailsome of the topics associated with time-resolved measurements introduced in Chapter 4

The second section of the dissertation is composed of Chapters 5 and 6, whichpresent the results of this work Chapter 5 discusses the development of the data sets andwill show the time averaged and time resolved pressure distributions at one clockingposition The heat-flux data, which was also taken at this clocking position, is provided

in Appendix C The data presented in Chapter 5 supports the data shown in Chapter 6,which details the changes dues to clocking, and forms the core of the analysis Chapter 6also has a supporting appendix (D), which discusses the uncertainty analysis used forpresentation of the data

This system is not quite as linear as some might like By putting the supportingwork in the appendices, the main discussion is not as easily lost in a sea of details

However, the reader is cautioned that some of the details in these appendices are not justhousekeeping items, but are critical to the understanding of the data An attempt has beenmade to make the appendices complete chapters so there are a few cases where figuresare duplicated in both the chapters and the appendices, as this adds to the readability ofthe document

CHAPTER 2 CURRENT RESEARCH INTERESTS IN CLOCKING, AERO-PERFORMANCE,

AERODYNAMICS, AND HEAT TRANSFER MEASUREMENTS

2.1 General Review

A detailed review of the current state of convective heat transfer and

aerodynamics has been given recently by Dunn as a IGTI Scholar lecture [2] This papercovers in great depth the developments in these areas and the relationships to variousfacilities In this dissertation, the main scientific analysis will be aerodynamic clockingeffects, and the literature review for that topic will be provided shortly As an

introduction, the main research trends that this phase of the project hoped to influencewill be briefly addressed

For the aerodynamic and heat transfer phases of this research, the primary focuswas on code verification This is because the geometry is well known and the conditionsare designed to duplicate specific flight conditions The one exception to this was theheat-transfer in the tip region, where a great deal of work has been done in non-rotatingreference frames Bunker provided an excellent review of this topic [3] as well as beingone of its major contributors Realistically, the questions of proper grid resolutions andtechniques for handling the rotating nature of the calculations have not been resolved tothe point where the design or research codes have been used enough to claim that they areready for the verification stage Rather a more complete data set was required to help dosome of the preliminary checkouts of the codes and the computational procedures

A second major topic related to clocking was the development of the capability toperform efficiency measurements (aero-performance) in a short duration facility Thiswork has had a long line of research done on the subject dating to [4] and more recentlydone by [5] This work has shown that aero-performance measurements can be made in

these facilities, and there was a specific goal to develop multiple systems of measuringefficiency (thermodynamic and mechanical) that will yield similar answers in a systemthat is believable to the engine industry Thus, one major goal was the development ofthis data set so that the proper analysis could be done on the experimental results todeduce efficiencies

A final issue in this data set was the incorporation of several advances in

instrumentation and data acquisition/reduction techniques These had been developedover the course of the four preceding programs, and the goal was to improve the overallmeasurement resolution Several techniques were tried, and these included increasedresolution of the speed, better calibration of the heat-flux sensors and the improvedamplifiers All of these techniques were utilized in this set of experiments in preparationfor the cooled experiments discussed in Chapter 1

2.2 Low Pressure Turbine Vane (LPV) Clocking Literature review

The concept of “clocking” has evolved over the years, but the basic idea is tochange the position of one blade row relative to another blade row One can imagine that

in this case, the viscous losses could be reduced if the second vane were placed in thefluid area with high eddy losses, instead of occupying an area which was relatively cleanflow (such as would occur if the leading edge of the second vane was in the middle of thefirst vane row passages) The idea being that the greater the amount of eddy losses, theless useable energy there is in the flow to do work The major clocking work can

arguably be stated to start with Denton [6] In this paper, Denton lays the conceptualgroundwork for why clocking should work through his description of the mechanismsinvolved in “blade boundary layer loss”2

Clocking has often been done in high pressureturbines by changing the position of the second vane relative to the first [7], but it canalso be done with rotor blades as was done in the Westinghouse ATS turbine where theblade rows of the first two stages of a high pressure turbine were clocked relative to eachother [8] The potential benefits of clocking can be applied to many areas as will bediscussed, but the first efforts were generally contrived to improve overall turbine

2

Pg 14 of ref 6 Denton, J.D., Loss Mechanisms in Turbomachines ASME Journal

of Turbomachinery, 1993(115): p 621-658

efficiency One of the key pieces of information that was needed in the development ofthis hypothesis is the fact that the upstream vane wakes propagate through the rotor stagerelatively intact down into the second vane area This was first shown for a transonicturbine by Dunn [9] In fact, if the rotor completely mixed up the first vane wakes, onewould not hypothesis any effect from clocking

Clocking research has been widespread and has taken several forms with bothexperimental and numerical studies Dunn [2] has provided a good review of the

clocking studies involving heat transfer and aeroperformance

As noted above one of major works in this area was reported in a two part paperauthored by Huber et al [7], [10] In this study the effect of clocking was investigated on

a relatively low pressure ratio turbo-pump (1.42 over full two stages of turbine) usingpurely inlet and exit flow field measurements (total temperature and total pressure), andthe measured torque generated by the machine as suggested by Haldeman and Dunn [4,11] These measurements separate into the two commonly referred to measurements ofefficiency: thermodynamic and mechanical This study showed that both the mechanicaland thermodynamic measurements had the same sinusoidal shape with approximately a0.8% variation in efficiency over the range of clocking positions at the midspan location(although there was on offset between these two methods) However, what was alsoobserved in this work was a distinct variation in which position provided the highestefficiency based on spanwise location of the measurements Averaging the spanwisemeasurements of figure 8 (Huber et al [7]) (which can be done since each measurement

is at the center of an equal area) still yields a minimum at position 5 and a max at position

3 (instead of 2), but the overall variation is now 0.1 % (a factor of 8 less) with a statedaccuracy of the measurements is 0.07% (although the procedure used to arrive at thisnumber is not given) Huber et al [7] also examined the amplitudes of the FFT at thefundamental blade passing frequency of the exit rake data Huber showed (at least atmidspan), that the amplitude also changed with clocking position with their position 5having the lowest magnitude and their position 3 having the highest This was not theirmaximum efficiency at midspan location but is close Another interesting result wasobtained when looking at the Reynolds number effects on clocking which showed a

3 (once again they were switched) In addition, looking at an integral effect over allspans once again switched the relative importance of positions 2 and 3 and reduced theoverall effect by a factor of 8, down to a level of the same order as the uncertainty in themeasurements (which were not defined as being either a standard deviation or a range).

If the main data of figure 9 and 10 showed position 3 as being the highest efficiency, thedata would be more consistent, but the sine-pattern would be disturbed Despite thesetechnical limits, it was clear that one could resolve a difference due to clocking at a givenspan location both in the frequency and the time domain, and that this variation is notgreat (on the order of 1% maximum variation) In addition, while a 0.07% measurementerror seems quite acceptable, the definition of the this uncertainty band and its level have

to be reduced in order to better resolve the variations that occur between spans

At about the same time as this paper, two others papers by Dorney et al [12, 13]also appeared While [12]and [13] are both numerical studies, they point to some

interesting aspects of clocking that one might be able to observe experimentally, even ifthe experiment was not quite the same as the numerical investigation In [12], Dorney et

al examine the affect of combustor hot streaks on a 1 1/2-stage turbine In this case, theclocking was of the first vane relative to the hot spots generated by the combustor Themain results in this study were that the positioning of the first vane changed which

surface of the rotor and second vane had the higher temperatures In [13], Dorney’s workfocuses on modeling a relatively low pressure ratio turbine from the United TechnologiesLarge Scale Rotating Rig (LSRR) The conclusions from this work were that a 2%change in efficiency could be seen due to clocking and that the relative unsteadinesslevels increase on the second vane at the clocking positions that correspond to maximum

efficiency Determining if this result would apply to higher work turbines in fully

corrected conditions was one of the goals of this study

There have been other papers dealing with different aspects of clocking in

turbines Eulitz et al [14] used unsteady 2D Reynolds average equations with a oneequation turbulence model to examine the effects of clocking Their results show about a0.4% change in time average efficiency with clocking, but this was one of the first papers

to look at how the time average and the fundamental and first and second harmonics ofblade passing change amplitude with clocking position Another important point,

although not well developed is that clocking seems only to influence the second vane andnot any other parts of the turbine Johnston and Fleeter [15] examined the effect ofclocking on heat transfer measurements, although their work included pressure

measurements as well Their major findings reflect both the strong three-dimensionalcomponents of clocking (i.e the effect varies with span location), and the fact that thetime resolved data for pressure and heat-flux in often dissimilar along the wetted distance

of the airfoil and along the span The effect on multi-stage low turbines was numericallyinvestigated using a 3D time accurate code by Arnone et al [16] While the main goal ofthis study was to investigate the effects of turbulence modeling on the low turbine, theyfound that the efficiency range estimate was about 0.7% and that there were no majorchanges due to Reynolds number effects in the results Combination studies involvingboth numerical and experimental work have begun to be published outside of the US.Reinmoller et al [17] used completely different instrumentation techniques, separatingthe unsteady from the steady components

At the OSU GTL, clocking experiments were done in three major experimentalprograms (to date) In order of performance, these were the Westinghouse ATS program1994-1996 [8] the Pratt and Whitney MTFE (Mid Thrust Family of Engines) program,1994-1997 [18], [19] and then most recently this work Contractual requirements limitedthe amount of analysis that was done both on the Westinghouse and MTFE data, but therewas a distinct effort made to improve the measurements and the experiments throughoutthese programs, building upon the knowledge gained in the previous programs Tracingthe uncertainty estimates through these programs one can see that as the measurement

to a lack of trying, but the data reduction technique in use at that time and the lack ofrepeat conditions made it hard to bound the perceived clocking effects Therefore, theresults presented in these works were presented as “differences in percent” as opposed tothe “percent variations” It was these limitations that guided the direction of the work inthis dissertation, and the techniques developed here will be used to re-examine past datasets to improve the data resolution.

The major advancements in measurement accuracy that affect the clocking

measurements in this program come primarily from the increase in the number and

absolute pressure level of the pressure sensors, and improvements in the speed

measurements Many different data reduction techniques were used both in the

Westinghouse rig and the MTFE rig to estimate the acceleration of the rotor from theabsolute position of the rotor (encoder data) This reduced to a problem of obtaining anaccurate second derivative of the position signal These techniques included (but are notlimited to) developing orthogonal polynomial fit routines, least median fit routines andjitter routines, described in Appendix C What was most interesting in these methods isthat despite the relatively complex nature of the routines, none could provide repeatableestimates of the acceleration, which of course brought into question the basic

measurements Several different techniques were tried in both the Westinghouse and theMTFE programs (as outlined in Appendix D) that ultimately forced a change in the datareduction procedure with the current program, and provided the insight needed to resolvethe clocking effects in the multiple ways that was achieved

Summarizing the current state of clocking research; several studies existed outsidethe data generated by the OSU GTL which showed clocking effects, but experimentalwork was tied either to very low pressure ratio machines, or to rigs that were not

operating at proper operating points In addition, the rig geometry was relatively benign,

without the “S” ducts that connect the modern high-pressure turbine with the

low-pressure vane In addition, while experimental work had shown clocking effects, thesewere very small, and the uncertainty analysis was not sufficiently developed to interpret

if the observed characteristics were within the repeatability of the measurements Thefact that integrating over the entire airfoil in the case of Huber et al [7] reduced thecalculated effect, and the location of maximum efficiency changed (although not much)indicated that the effect was probably real, but it seemed as though the whole picture wasnot being observed properly At the OSU GTL, several different rigs had been used inclocking experiments which were of more modern turbine design (higher pressure ratio)and were much more three-dimensional Clocking effects had been observed in all cases

by examining the pressure distribution on the airfoils, but correlating these measurements

to an overall measurement of the machine efficiency was elusive

Thus, one of the goals for the experiment was to improve the measurement

accuracy to the point where aeroperformance measurements of clocking (both from amechanical and thermodynamic perspective) could be made and to compare these withtime-resolved and time-averaged measurements of clocking affect on the airfoils Inaddition, we wanted to see if one technique might show clocking effects more

substantially than another The desire was to improve both the measurements and thedata reduction to the point where the uncertainty analysis could properly bound themeasurements to a resolution that would allow accurate comparisons between clockingpositions Also, we wanted to use the time honored experimental technique of

confirming the answers by verifying the results by using independent measures andprocedures Finally, we wanted to see if the results from the MTFE turbine stage

(particularly the influence of the spanwise position on clocking results) would be

replicated on a different machine