Idea Transcript
WIND TUNNEL TESTING OF A SUPERSONIC CRUISE NOZZLE IN SUBSONIC EJECTOR CONFIGURATION
A Thesis Submitted to the Faculty of Purdue University by Jesse Thomas Jones
In Partial Fulfillment of the Requirements for the Degree of Master of Science in Aeronautics and Astronautics
May 2009 Purdue University West Lafayette, Indiana
ii
iii
ACKNOWLEDGMENTS
iv
TABLE OF CONTENTS Page LIST OF TABLES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
vi
LIST OF FIGURES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
vii
SYMBOLS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
xiv
ABBREVIATIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
xv
ABSTRACT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
xvi
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1
2 Literature Review . . . . . . . . . . . . 2.1 Ejectors . . . . . . . . . . . . . . 2.2 Previous Jet Engine Applications 2.3 Summary . . . . . . . . . . . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
4 4 9 12
3 Test Nozzle Geometry . . . . . . . . 3.1 CAD Model . . . . . . . . . . . 3.1.1 CATIA Design Tables . 3.1.2 Model Parameterization 3.1.3 Model Flexibility . . . . 3.2 Test Nozzle Design . . . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
13 13 14 14 15 16
4 Facility Setup . . . . . . . . . 4.1 Requirements . . . . . . 4.1.1 Existing Facility . 4.2 Facility Additions . . . . 4.3 Test Rig . . . . . . . . . 4.4 Capability . . . . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
19 19 19 20 21 23
5 Measurement Systems . . . . . . . 5.1 Seven Hole Probe . . . . . . . 5.1.1 Theory . . . . . . . . . 5.1.2 Calibration . . . . . . 5.1.3 Data Reduction . . . . 5.2 Pressure Scanner . . . . . . . 5.3 Traverse . . . . . . . . . . . . 5.3.1 Axes & Support Truss 5.3.2 Motor Controllers . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
24 24 26 29 30 31 33 33 34
v
5.4
Survey Grids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6 Flow Visualization . . . . . . 6.0.1 Smoke Wand . . 6.0.2 Tufting . . . . . . 6.0.3 Surface Sediment 6.0.4 Fluorescent Oil .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
38 38 39 40 42
7 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.1 Rig Velocity Profile . . . . . . . . . . . . . . . . . . . . 7.2 Nozzle Wake Surveys . . . . . . . . . . . . . . . . . . . 7.2.1 Case #1: C-D Nozzle, Ejector Slot Closed . . . 7.2.2 Case #2: Primary Nozzle, Clamshells Removed 7.2.3 Case #3: Clamshell AoA = 9.0◦ . . . . . . . . . 7.2.4 Case #4: Clamshell AoA = 11.5◦ . . . . . . . . 7.2.5 Case #5: Clamshell AoA = 15.0◦ . . . . . . . . 7.3 Thrust Analysis . . . . . . . . . . . . . . . . . . . . . . 7.3.1 Methodology . . . . . . . . . . . . . . . . . . . 7.3.2 Thrust Comparison . . . . . . . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
44 44 45 47 50 54 57 61 63 63 69
8 Conclusions & Recommendations . . . . . . . . . . . . . . . . . . . . . .
71
LIST OF REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . .
74
A Wake Survey Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
76
B Fluorescent Oil Images . . . . . . . . . . . . . . . . . . . . . . . . . . . .
143
C Surface Sediment Traces . . . . . . . . . . . . . . . . . . . . . . . . . . .
146
D Seven-Hole Probe Calibration Data . . . . . . . . . . . . . . . . . . . . .
149
E Parametric CAD Model of the Ejector Nozzle . E.1 Getting Started . . . . . . . . . . . . . . . E.1.1 CATIA Background . . . . . . . . . E.1.2 Modifying the Model . . . . . . . . E.1.3 Existing Designs . . . . . . . . . . E.2 Parameterization Scheme . . . . . . . . . . E.2.1 Control Plane . . . . . . . . . . . . E.2.2 Pre-Throat Flowpath Parameters . E.2.3 Principle Nozzle Geometry Controls E.2.4 Clamshell Definition Controls . . . E.2.5 Hardware Related Parameters . . . E.2.6 Aero-Profile Parameters . . . . . . E.2.7 Applications . . . . . . . . . . . . .
153 153 153 155 157 158 158 158 161 163 164 165 168
. . . . . . . . . . . . .
. . . . .
. . . . . . . . . . . . .
. . . . .
. . . . . . . . . . . . .
. . . . .
. . . . . . . . . . . . .
. . . . .
. . . . . . . . . . . . .
. . . . .
Page 36
. . . . . . . . . . . . .
. . . . . . . . . . . . .
. . . . . . . . . . . . .
. . . . . . . . . . . . .
. . . . . . . . . . . . .
. . . . . . . . . . . . .
. . . . . . . . . . . . .
. . . . . . . . . . . . .
. . . . . . . . . . . . .
vi
LIST OF TABLES Table
Page
5.1
Survey Grid Parameters . . . . . . . . . . . . . . . . . . . . . . . . . .
37
7.1
Angular Offsets of 7-Hole Probe & Test Rig . . . . . . . . . . . . . . .
53
A.1 Survey Listing & Parameters . . . . . . . . . . . . . . . . . . . . . . .
76
vii
LIST OF FIGURES Figure
Page
2.1
Early Industrial Ejector Design [1] . . . . . . . . . . . . . . . . . . . .
4
2.2
Ejector Free-Mixing Layer Attachment [4] . . . . . . . . . . . . . . .
6
2.3
Depiction of the free mixing layer between the primary and secondary ejector streams [5] . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7
2.4
Compressibility effect on turbulent shear layer spreading rate [6] . . .
8
2.5
Ejector geometry parameters . . . . . . . . . . . . . . . . . . . . . . .
10
2.6
Concorde exhaust ejector [13] . . . . . . . . . . . . . . . . . . . . . .
11
3.1
Ejector Slot Parameterization . . . . . . . . . . . . . . . . . . . . . .
15
3.2
Illustration of Model Flexibility . . . . . . . . . . . . . . . . . . . . .
16
3.3
CATIA Model of Test Nozzle . . . . . . . . . . . . . . . . . . . . . .
17
4.1
ASL Test Facility . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
20
4.2
Test Rig Support Structure on BWT Test Section . . . . . . . . . . .
21
4.3
BWT Co-axial Flow Test Rig . . . . . . . . . . . . . . . . . . . . . .
21
4.4
Cross-Section View of Test Rig . . . . . . . . . . . . . . . . . . . . .
22
5.1
Schematic of Measurement Equipment . . . . . . . . . . . . . . . . .
25
5.2
Front View of Probe . . . . . . . . . . . . . . . . . . . . . . . . . . .
27
5.3
Seven-Hole Probe & Coordinate System
. . . . . . . . . . . . . . . .
28
5.4
Test Section Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . .
35
5.5
Survey Plane Locations . . . . . . . . . . . . . . . . . . . . . . . . . .
36
6.1
External flow ingestion through ejector slot . . . . . . . . . . . . . . .
38
6.2
Separation bubble in ejector slot, α = 11.5◦ , VINF = 5 m/s . . . . . .
39
6.3
Flow reversal at ejector clamshell trailing edge, α = 11.5◦ , VINF = 5 m/s
39
6.4
Ejector Clamshell Tufts, α = 11.5◦ , VINF = 5 m/s . . . . . . . . . . .
40
6.5
Surface Sediment Trace On Clamshell Interior, α = 5.0◦ . . . . . . . .
41
viii Figure
Page
6.6
Fluorescent Oil Flow for Clamshell AoA = 5.0◦ . . . . . . . . . . . .
43
7.1
Rig Velocity Profiles . . . . . . . . . . . . . . . . . . . . . . . . . . .
45
7.2
Linear Surveys Across Clamshell Plane, X = 1.2 DEQ . . . . . . . . .
46
7.3
Case #1: Jet Wake Surveys . . . . . . . . . . . . . . . . . . . . . . .
48
7.4
Case #1: Transverse Flow, X = 1.0 DEQ . . . . . . . . . . . . . . . .
49
7.5
Case #1: Fluorescent Oil Visualization . . . . . . . . . . . . . . . . .
49
7.6
Case #2: Jet Wake Surveys . . . . . . . . . . . . . . . . . . . . . . .
50
7.7
Case #2: Convergence of primary nozzle flow along Z=0 . . . . . . .
51
7.8
Case #2: Transverse Flow, X = 1.5 DEQ . . . . . . . . . . . . . . . .
52
7.9
Case #2: Mixing Layer in the Z = 0 plane . . . . . . . . . . . . . . .
53
7.10
Case #3: Jet Wake Surveys . . . . . . . . . . . . . . . . . . . . . . .
54
7.11
Case #3: Fluorescent Oil Visualization . . . . . . . . . . . . . . . . .
55
7.12
Case #3: Transverse Flow, X = 1.5 DEQ . . . . . . . . . . . . . . . .
56
7.13
Case #4: Jet Wake Surveys . . . . . . . . . . . . . . . . . . . . . . .
57
7.14
Case #4: Surface Flow Visualization . . . . . . . . . . . . . . . . . .
58
7.15
Case #4: Transverse Flow, X = 1.5 DEQ . . . . . . . . . . . . . . . .
59
7.16
Case #4: Mixing layer in the Z = 0 plane . . . . . . . . . . . . . . .
60
7.17
Case #5: Jet Wake Surveys . . . . . . . . . . . . . . . . . . . . . . .
61
7.18
Case #5: Fluorescent Oil Visualization . . . . . . . . . . . . . . . . .
62
7.19
Case #5: Transverse Flow, X = 1.5 DEQ . . . . . . . . . . . . . . . .
62
7.20
Control Volume for integral thrust calculations . . . . . . . . . . . . .
64
7.21
Clamshell Configuration Thrust Coefficients . . . . . . . . . . . . . .
70
A.1
Survey 1: Axial Flow Speed, Wake Profile . . . . . . . . . . . . . . .
77
A.2
Survey 1: Transverse Flow & Fluctuation . . . . . . . . . . . . . . . .
77
A.3
Survey 1: Transverse Flow, Magnification . . . . . . . . . . . . . . . .
78
A.4
Survey 1: Axial Flow Speed @ Z=0 . . . . . . . . . . . . . . . . . . .
79
A.5
Survey 1: Axial Flow Speed @ Y=0 . . . . . . . . . . . . . . . . . . .
79
A.6
Survey 2: Axial Flow Speed, Wake Profile . . . . . . . . . . . . . . .
80
ix Figure
Page
A.7
Survey 2: Transverse Flow & Fluctuation . . . . . . . . . . . . . . . .
80
A.8
Survey 2: Transverse Flow, Magnification . . . . . . . . . . . . . . . .
81
A.9
Survey 2: Axial Flow Speed @ Z=0 . . . . . . . . . . . . . . . . . . .
82
A.10 Survey 2: Axial Flow Speed @ Y=0 . . . . . . . . . . . . . . . . . . .
82
A.11 Survey 3: Axial Flow Speed, Wake Profile . . . . . . . . . . . . . . .
83
A.12 Survey 3: Transverse Flow & Fluctuation . . . . . . . . . . . . . . . .
83
A.13 Survey 3: Transverse Flow, Magnification . . . . . . . . . . . . . . . .
84
A.14 Survey 3: Axial Flow Speed @ Z=0 . . . . . . . . . . . . . . . . . . .
85
A.15 Survey 3: Axial Flow Speed @ Y=0 . . . . . . . . . . . . . . . . . . .
85
A.16 Survey 4: Axial Flow Speed, Wake Profile . . . . . . . . . . . . . . .
86
A.17 Survey 4: Transverse Flow & Fluctuation . . . . . . . . . . . . . . . .
86
A.18 Survey 4: Transverse Flow, Magnification . . . . . . . . . . . . . . . .
87
A.19 Survey 4: Axial Flow Speed @ Z=0 . . . . . . . . . . . . . . . . . . .
88
A.20 Survey 4: Axial Flow Speed @ Y=0 . . . . . . . . . . . . . . . . . . .
88
A.21 Survey 5: Axial Flow Speed, Wake Profile . . . . . . . . . . . . . . .
89
A.22 Survey 5: Transverse Flow & Fluctuation . . . . . . . . . . . . . . . .
89
A.23 Survey 5: Transverse Flow, Magnification . . . . . . . . . . . . . . . .
90
A.24 Survey 5: Axial Flow Speed @ Z=0 . . . . . . . . . . . . . . . . . . .
91
A.25 Survey 5: Axial Flow Speed @ Y=0 . . . . . . . . . . . . . . . . . . .
91
A.26 Survey 6: Axial Flow Speed, Wake Profile . . . . . . . . . . . . . . .
92
A.27 Survey 6: Transverse Flow & Fluctuation . . . . . . . . . . . . . . . .
92
A.28 Survey 6: Transverse Flow, Magnification . . . . . . . . . . . . . . . .
93
A.29 Survey 6: Axial Flow Speed @ Z=0 . . . . . . . . . . . . . . . . . . .
94
A.30 Survey 6: Axial Flow Speed @ Y=0 . . . . . . . . . . . . . . . . . . .
94
A.31 Survey 7: Axial Flow Speed, Wake Profile . . . . . . . . . . . . . . .
95
A.32 Survey 7: Transverse Flow & Fluctuation . . . . . . . . . . . . . . . .
95
A.33 Survey 7: Transverse Flow, Magnification . . . . . . . . . . . . . . . .
96
A.34 Survey 7: Axial Flow Speed @ Z=0 . . . . . . . . . . . . . . . . . . .
97
x Figure
Page
A.35 Survey 7: Axial Flow Speed @ Y=0 . . . . . . . . . . . . . . . . . . .
97
A.36 Survey 8: Axial Flow Speed, Wake Profile . . . . . . . . . . . . . . .
98
A.37 Survey 8: Transverse Flow & Fluctuation . . . . . . . . . . . . . . . .
98
A.38 Survey 8: Transverse Flow, Magnification . . . . . . . . . . . . . . . .
99
A.39 Survey 8: Axial Flow Speed @ Z=0 . . . . . . . . . . . . . . . . . . .
100
A.40 Survey 8: Axial Flow Speed @ Y=0 . . . . . . . . . . . . . . . . . . .
100
A.41 Survey 9: Axial Flow Speed, Wake Profile . . . . . . . . . . . . . . .
101
A.42 Survey 9: Transverse Flow & Fluctuation . . . . . . . . . . . . . . . .
101
A.43 Survey 9: Transverse Flow, Magnification . . . . . . . . . . . . . . . .
102
A.44 Survey 9: Axial Flow Speed @ Z=0 . . . . . . . . . . . . . . . . . . .
103
A.45 Survey 9: Axial Flow Speed @ Y=0 . . . . . . . . . . . . . . . . . . .
103
A.46 Survey 10: Axial Flow Speed, Wake Profile . . . . . . . . . . . . . . .
104
A.47 Survey 10: Transverse Flow & Fluctuation . . . . . . . . . . . . . . .
104
A.48 Survey 10: Transverse Flow, Magnification . . . . . . . . . . . . . . .
105
A.49 Survey 10: Axial Flow Speed @ Z=0 . . . . . . . . . . . . . . . . . .
106
A.50 Survey 10: Axial Flow Speed @ Y=0 . . . . . . . . . . . . . . . . . .
106
A.51 Survey 11: Axial Flow Speed, Wake Profile . . . . . . . . . . . . . . .
107
A.52 Survey 11: Transverse Flow & Fluctuation . . . . . . . . . . . . . . .
107
A.53 Survey 11: Transverse Flow, Magnification . . . . . . . . . . . . . . .
108
A.54 Survey 11: Axial Flow Speed @ Z=0 . . . . . . . . . . . . . . . . . .
109
A.55 Survey 11: Axial Flow Speed @ Y=0 . . . . . . . . . . . . . . . . . .
109
A.56 Survey 12: Axial Flow Speed, Wake Profile . . . . . . . . . . . . . . .
110
A.57 Survey 12: Transverse Flow & Fluctuation . . . . . . . . . . . . . . .
110
A.58 Survey 12: Transverse Flow, Magnification . . . . . . . . . . . . . . .
111
A.59 Survey 12: Axial Flow Speed @ Z=0 . . . . . . . . . . . . . . . . . .
112
A.60 Survey 12: Axial Flow Speed @ Y=0 . . . . . . . . . . . . . . . . . .
112
A.61 Survey 13: Axial Flow Speed, Wake Profile . . . . . . . . . . . . . . .
113
A.62 Survey 13: Transverse Flow & Fluctuation . . . . . . . . . . . . . . .
113
xi Figure
Page
A.63 Survey 13: Transverse Flow, Magnification . . . . . . . . . . . . . . .
114
A.64 Survey 13: Axial Flow Speed @ Z=0 . . . . . . . . . . . . . . . . . .
115
A.65 Survey 13: Axial Flow Speed @ Y=0 . . . . . . . . . . . . . . . . . .
115
A.66 Survey 14: Axial Flow Speed, Wake Profile . . . . . . . . . . . . . . .
116
A.67 Survey 14: Transverse Flow & Fluctuation . . . . . . . . . . . . . . .
116
A.68 Survey 14: Transverse Flow, Magnification . . . . . . . . . . . . . . .
117
A.69 Survey 14: Axial Flow Speed @ Z=0 . . . . . . . . . . . . . . . . . .
118
A.70 Survey 14: Axial Flow Speed @ Y=0 . . . . . . . . . . . . . . . . . .
118
A.71 Survey 15: Axial Flow Speed, Wake Profile . . . . . . . . . . . . . . .
119
A.72 Survey 15: Transverse Flow & Fluctuation . . . . . . . . . . . . . . .
119
A.73 Survey 15: Transverse Flow, Magnification . . . . . . . . . . . . . . .
120
A.74 Survey 15: Axial Flow Speed @ Z=0 . . . . . . . . . . . . . . . . . .
121
A.75 Survey 15: Axial Flow Speed @ Y=0 . . . . . . . . . . . . . . . . . .
121
A.76 Survey 16: Axial Flow Speed, Wake Profile . . . . . . . . . . . . . . .
122
A.77 Survey 16: Transverse Flow & Fluctuation . . . . . . . . . . . . . . .
122
A.78 Survey 16: Transverse Flow, Magnification . . . . . . . . . . . . . . .
123
A.79 Survey 16: Axial Flow Speed @ Z=0 . . . . . . . . . . . . . . . . . .
124
A.80 Survey 16: Axial Flow Speed @ Y=0 . . . . . . . . . . . . . . . . . .
124
A.81 Survey 17: Axial Flow Speed, Wake Profile . . . . . . . . . . . . . . .
125
A.82 Survey 17: Transverse Flow & Fluctuation . . . . . . . . . . . . . . .
125
A.83 Survey 17: Transverse Flow, Magnification . . . . . . . . . . . . . . .
126
A.84 Survey 17: Axial Flow Speed @ Z=0 . . . . . . . . . . . . . . . . . .
127
A.85 Survey 17: Axial Flow Speed @ Y=0 . . . . . . . . . . . . . . . . . .
127
A.86 Survey 18: Axial Flow Speed, Wake Profile . . . . . . . . . . . . . . .
128
A.87 Survey 18: Transverse Flow & Fluctuation . . . . . . . . . . . . . . .
128
A.88 Survey 18: Transverse Flow, Magnification . . . . . . . . . . . . . . .
129
A.89 Survey 18: Axial Flow Speed @ Z=0 . . . . . . . . . . . . . . . . . .
130
A.90 Survey 18: Axial Flow Speed @ Y=0 . . . . . . . . . . . . . . . . . .
130
xii Figure
Page
A.91 Survey 19: Axial Flow Speed, Wake Profile . . . . . . . . . . . . . . .
131
A.92 Survey 19: Transverse Flow & Fluctuation . . . . . . . . . . . . . . .
131
A.93 Survey 19: Transverse Flow, Magnification . . . . . . . . . . . . . . .
132
A.94 Survey 19: Axial Flow Speed @ Z=0 . . . . . . . . . . . . . . . . . .
133
A.95 Survey 19: Axial Flow Speed @ Y=0 . . . . . . . . . . . . . . . . . .
133
A.96 Survey 20: Axial Flow Speed, Wake Profile . . . . . . . . . . . . . . .
134
A.97 Survey 20: Transverse Flow & Fluctuation . . . . . . . . . . . . . . .
134
A.98 Survey 20: Transverse Flow, Magnification . . . . . . . . . . . . . . .
135
A.99 Survey 20: Axial Flow Speed @ Z=0 . . . . . . . . . . . . . . . . . .
136
A.100 Survey 20: Axial Flow Speed @ Y=0 . . . . . . . . . . . . . . . . . .
136
A.101 Survey 21: Axial Flow Speed, Wake Profile . . . . . . . . . . . . . . .
137
A.102 Survey 21: Transverse Flow & Fluctuation . . . . . . . . . . . . . . .
137
A.103 Survey 21: Transverse Flow, Magnification . . . . . . . . . . . . . . .
138
A.104 Survey 21: Axial Flow Speed @ Z=0 . . . . . . . . . . . . . . . . . .
139
A.105 Survey 21: Axial Flow Speed @ Y=0 . . . . . . . . . . . . . . . . . .
139
A.106 Survey 22: Axial Flow Speed, Wake Profile . . . . . . . . . . . . . . .
140
A.107 Survey 22: Transverse Flow & Fluctuation . . . . . . . . . . . . . . .
140
A.108 Survey 22: Transverse Flow, Magnification . . . . . . . . . . . . . . .
141
A.109 Survey 22: Axial Flow Speed @ Z=0 . . . . . . . . . . . . . . . . . .
142
A.110 Survey 22: Axial Flow Speed @ Y=0 . . . . . . . . . . . . . . . . . .
142
B.1
Fluorescent Oil Flow for Clamshell AoA = 0.0◦ . . . . . . . . . . . .
143
B.2
Fluorescent Oil Flow for Clamshell AoA = 5.0◦ . . . . . . . . . . . .
144
B.3
Fluorescent Oil Flow for Clamshell AoA = 9.0◦ . . . . . . . . . . . .
144
B.4
Fluorescent Oil Flow for Clamshell AoA = 11.5◦ . . . . . . . . . . . .
145
B.5
Fluorescent Oil Flow for Clamshell AoA = 15.0◦ . . . . . . . . . . . .
145
C.1
Sediment Trace for Clamshell AoA = 0.0◦ . . . . . . . . . . . . . . .
146
C.2
Sediment Trace for Clamshell AoA = 5.0◦ . . . . . . . . . . . . . . .
147
C.3
Sediment Trace for Clamshell AoA = 9.0◦ . . . . . . . . . . . . . . .
147
xiii Figure
Page
C.4
Sediment Trace for Clamshell AoA = 11.5◦ . . . . . . . . . . . . . . .
148
C.5
Sediment Trace for Clamshell AoA = 15.0◦ . . . . . . . . . . . . . . .
148
D.1
Calibration Data: CPθ = f (θ, ψ) . . . . . . . . . . . . . . . . . . . . .
149
D.2
Calibration Data: CPψ = f (θ, ψ) . . . . . . . . . . . . . . . . . . . . .
150
D.3
Calibration Data: CPstatic = f (θ, ψ) . . . . . . . . . . . . . . . . . . .
151
D.4
Calibration Data: CPtotal = f (θ, ψ) . . . . . . . . . . . . . . . . . . . .
152
E.1
CATIA Desktop Icons . . . . . . . . . . . . . . . . . . . . . . . . . .
155
E.2
CATIA Specification Tree & Refresh Icon . . . . . . . . . . . . . . . .
156
E.3
Control Plane Parameters . . . . . . . . . . . . . . . . . . . . . . . .
159
E.4
Pre-Throat Flowpath Parameters . . . . . . . . . . . . . . . . . . . .
160
E.5
Principle Nozzle Parameters . . . . . . . . . . . . . . . . . . . . . . .
161
E.6
Supersonic Throat Definition . . . . . . . . . . . . . . . . . . . . . . .
162
E.7
Clamshell Definition Parameters . . . . . . . . . . . . . . . . . . . . .
163
E.8
Hardware Related Parameters . . . . . . . . . . . . . . . . . . . . . .
164
E.9
Ejector Slot Parameters . . . . . . . . . . . . . . . . . . . . . . . . .
166
E.10
Ejector Slot Parameters . . . . . . . . . . . . . . . . . . . . . . . . .
167
E.11
Applications in Specification Tree . . . . . . . . . . . . . . . . . . . .
168
xiv
SYMBOLS M
Mach Number
q
dynamic pressure
γ
specific heat ratio of air
DEQ
jet equivalent diameter
p0
local total pressure
p
local static pressure
pi
pressure measured on the i port of the 7-hole probe
m ˙
mass flow rate
A
Cross-sectional flow area
V
volume
Vp
pressure transducer output voltage
Tp
Temperature of pressure transducers in the PSI 9010
Ci (Tp )
PSI 9010, i-th order calibration coefficient
Crz
Re-zero correction coefficient for 9010 pressure scanner
Cspan
Rezero correction coefficient for 9010 pressure scanner
pm
The pressure measured by a transducer of the 9010 scanner
V
Flowfield velocity vector
V
Flowfield velocity magnitude
Vt
Transverse flowfield velocity vector: [v, w]
u
X-axis component of flowfield velocity vector
v
Y-axis component of flowfield velocity vector
w
Z-axis component of flowfield velocity vector
xv
ABBREVIATIONS SSBJ
Supersonic Business Jet
BWT
Boeing Wind Tunnel
HPB
High Pressure Blower
AoA
Angle of Attack
A2D
Analog-to-Digital Converter
ICL
Intelli-Command Language
GUI
Graphical User Interface
xvi
ABSTRACT Jones, Jesse T. M.S.A.E., Purdue University, May 2009. Wind Tunnel Testing of a Supersonic Cruise Nozzle in Subsonic Ejector Configuration . Major Professor: John P. Sullivan. In the design of a jet engine exhaust nozzle for civilian supersonic cruise vehicles, it is necessary to consider both the supersonic cruise efficiency and the the noise generated during terminal area operations. The convergent-divergent nozzle required for cruise efficiency, will separate in the divergent portion of the nozzle when operated at the lower nozzle pressure ratios associated with take-off and initial climb, causing unacceptable noise levels. One proposed solution is to comprise the divergent geometry of two rotatable clamshell bodies. These clamshell components would be rotated open during low nozzle pressure ratio operation, to limit the divergence and effect a subsonic ejector nozzle. These clamshells have the added benefit of functioning as thrust reversers and eliminating the need for a separate such system. This work, in general, investigates the effectiveness of a preliminary clamshell ejector design. Specifically, a seven-hole probe mounted on an automated, 2-axis traverse is used to collect wake survey data in the exhaust plume of the test ejector nozzle. This survey data is used to create velocity profiles of the jet plume and calculate thrust coefficients. In addition, several flow visualization techniques are utilized to further explore the flow structure near the geometry surface. Results indicate that the mixing layer between the primary and secondary streams is failing to attach to the ejector shroud, leading to separation and recirculation near the internal ejector shroud surfaces. Separation in these areas is causing streamwise vortices to be shed from the ejector surfaces.
1
1. Introduction Within the current constraints of the aviation industry, several companies have identified the opportunity for a supersonic business jet (SSBJ). In the pursuit of a commercially viable SSBJ, a suitable engine exhaust nozzle is a critical component of the vehicle system design. Several design constraints contribute to this assertion. Foremost, the nozzle must provide for maximum efficiency during the cruise portion of the mission profile. Given supersonic cruise conditions and the desire for minimal engine frontal area associated with supersonic flight, it is a compelling assumption that the optimal engine design will require a convergent-divergent (C-D) nozzle to accelerate the jet exhaust to supersonic speeds. In addition, any vehicle design will be subject to noise regulations in the vicinity of airports, as set by U.S. law and ICAO standards. Such restrictions are not currently specified for supersonic vehicles; however, this is only a consequence of a current absence of supersonic civilian transport aircraft in operation. It can be reasonably assumed that the subsonic operation of the proposed vehicle in the terminal area will be governed by restrictions similar to those which are currently in place for subsonic jet aircraft. These restrictions impose noise constraints on the take-off, initial climb, and approach phases of flight. In general, the performance of a jet nozzle affects these noise levels. However, given the requirement of a C-D cruise nozzle, the noise associated with the expected gross separation in the divergent portion of a C-D nozzle, operated at the lower nozzle pressure ratios (NPR’s) typical of take-off and landing, would likely violate any such noise constraints. Finally, from the perspective of customers’ utility of the vehicle, it is necessary to minimize the aircraft’s required landing distance, to maximize the number of airports at which it can operate. Though there are several means of accomplishing this, an engine equipped with reverse thrust capability is likely a required component of the design. There are numerous other nozzle design considerations
2 which are significant to the overall vehicle system, such as the effect of the nozzle’s geometry on the overall supersonic signature of the vehicle, which are omitted from this discussion as they are not covered by the scope of this work. One potential solution to the above design concerns is a clamshell-type ejector (CTE) nozzle. In such a design, the internal divergent surfaces of the C-D nozzle are substantially defined by two deployable clamshell-shaped bodies. Deploying the clamshells, by rotating them some small number of degrees aft, opens up an ejector slot and should effectively eliminate the C-D geometry which would lead to gross separation and the associated noise at low NPR’s. By rotating the clamshells completely aft, these same clamshell components can facilitate reverse thrust (not unlike the clamshell-type reverse thrust mechanisms incorporated into many current jet engines). In addition, by properly designing the wetted surfaces of the clamshells, their deployment can effect an ejector nozzle, capable of augmenting thrust, and/or further reducing jet exhaust noise by means of reducing mean shear levels in the jet exhaust plume. This solution is particularly elegant, from the vehicle system perspective, in that the deployable clamshells can address all three design concerns, raised in the prior paragraph, in one relatively simple mechanism, reducing weight, size and overall complexity of the engine. Furthermore, the subsonic ejector configuration presents an opportunity for further gains in performance and noise suppression beyond the aforementioned benefits. The overarching aim of the efforts described herein is to further the endeavor of designing a satisfactory CTE nozzle for application to a supersonic transport vehicle, as described above. Specifically, the purpose of this work is to design and fabricate a proposed CTE test nozzle, design and implement a test rig and facility modifications to enable flow visualization and wake measurements of the test nozzle, and finally to characterize the three dimensional flow structure associated with the nozzle and its exhaust plume. A 7-hole probe is used to map transverse flow patterns and axial flow profiles, identify areas of separation, and perform thrust measurement and
3 comparison. Multiple flow visualization techniques are also utilized to identify areas of separation and to obtain depictions of surface flow patterns.
4
2. Literature Review This chapter is intended to cover a cross-section of background literature, as related to this experiment. This includes some classical and more current work regarding the flow behavior of ejectors, in general. Also discussed, is prior work regarding the application of ejectors to aircraft jet engine nozzles.
2.1
Ejectors An ejector, within the context of fluid dynamics, is a means of pumping an oth-
erwise unmotivated fluid or solid particulate, secondary stream, utilizing the energy of a relatively high-speed, primary stream. Such devices, in general, are certainly not new. Crude ejection devices are described as far back as 1570, and though some related work was done in the early 19th century, credit for the invention of the ejector (or injector) is granted to Henri Giffard, with his 1858 patent [1]. The fundamental components of an ejector are depicted in Figure 2.1, showing a portion one of Henri Gifford’s early designs.
Figure 2.1. Henri Gifford’s Early Ejector Design [1]; ’a’ depicts the primary stream nozzle; ’B’ depicts the secondary stream; ’b’ depicts the mixing duct
5 The original application of Gifford’s ejector, and the variations that followed for several decades, was the continuous pumping of feed water to boilers, utilizing steam for the primary stream. As such, the volumes of empirical data that existed even as early as the publication of Reference [1] in 1894, are almost entirely related to this specific gas-liquid application. Though the transfer of energy from the primary stream to the secondary stream was realized as the base of the ejector’s operation at this early stage, the means of exchange was often referred to simply as the ’impact’ of the steam with the water [1]. To some degree this misunderstanding might have been furthered by the two-phase nature of these early ejectors (compared to the study of gas-gas ejectors); however, it was also certainly influenced by the state of understanding of fluid dynamics at the time. As is now known, fluid-dynamic mixing is the primary means of the momentum transfer in an ejector, and governs its pumping phenomenon [2, 3]. The degree of turbulent mixing of the two streams determines both ejector performance and the noise characteristics [2] of its operation. This mixing occurs in a free shear layer originating at the termination of the separation between the two streams (see Figures 2.3, 2.2) [4]. As depicted in Figure 2.2 and discussed by Der [4], much of the performance characteristics of ejectors in general, can be deduced from the behavior of the associated free shear layer and its relation to the ejector shroud walls. For example, if the mixing duct is too short, the shear layer will not attach to the ejector shroud wall, leaving flow of the secondary stream open to stronger influence by external conditions, and susceptible to separation and recirculation (see Figure 2.2). However, if the mixing duct extends significantly past the attachment point of the shear layer, frictional losses will result from the higher speed, mixed flow in contact with the extended duct. This leads to the notion of an optimum ejector length as depicted in Figure 2.2. Given the role of the topic in determining operational qualities of the ejector, further discussion of free shear layers is warranted. A depiction of a free shear layer can be seen in Figure 2.3, which is representative of that emanating from the stream
6
Figure 2.2. Ejector Free-Mixing Layer Attachment [4]
splitter of an ejector. The key characteristic of the shear layer in an ejector is its growth rate. It is largely this growth rate which determines whether or not the shear layer will attach to the ejector shroud walls for given ejector dimensions [4]. There are multiple methods of defining the shear layer thickness. Perhaps the most intuitive measure is the physical shear layer thickness, b, defined as the distance between the locations of U * (x) = 0.9 and U * (x) = 0.1 where U * (x) = (u (x)−u2 )/(u1 −u2 ). While
7
Figure 2.3. Depiction of the free mixing layer between the primary and secondary ejector streams [5]
this method may be the most intuitive, it is also more prone to experimental error [6, 7], due to its definition based on only two data points and difficulties associated with flow measurements at the lower end of a probe’s usable range (particularly as u2 → 0). Though momentum thickness is also mentioned in the literature [6], the most commonly used parameter appears to be the vorticity thickness, δω . This measure is shown in equation 2.1 as defined in Brown and Roshko [7]. δω =
u2 − u1 (δu/δy)max
(2.1)
All three of the aforementioned parameters have been shown to grow linearly in the developed region of a free turbulent mixing layer [5, 6]. In addition, parametric relationships have been established for three primary influences on mixing layer growth rate. These parameters are the density ratio (2.2), a compressibility factor, termed convective Mach number (2.3), and a velocity ratio (2.4). λρ = ρ2 /ρ1 Mc = (u1 − uc ) /a1 ,
(2.2) (2.3)
where uc = (a1 u2 + a2 u1 ) / (a1 + a2 ) λu = (u1 − u2 ) / (u1 + u2 )
(2.4)
Within the range of conditions expected for the partial scale testing and the full scale application of the nozzle, λu is shown to be the most influential parameter [5].
8
Figure 2.4. Compressibility effect on turbulent shear layer spreading rate [6]
References [5] and [7] show linear relationships for b (λu ) and δω (λu ), respectively. Though a density ratio is shown to have an effect, it is a significantly smaller influence, in the absence of density ratios on the order of 5:1 [8]. Furthermore, Brown and Roshko [7] have shown convective Mach number (the compressibility effect) to be a larger component of the change in spreading rate with Mach number. Though the compressibility effects associated with larger Mach number differences between flows are well demonstrated to have a slimming effect on turbulent shear layers, this phenomenon is shown to be nearly nonexistent for Mc values less than approximately 0.5 [5, 6, 8, 9]. This assertion is supported by a plot from Samimy shown as Figure 2.4. The ejector shear layer of both the test rig developed in this work and the proposed full scale application of the nozzle are expected to be within the range of 0 < Mc < 0.5. Though much can be understood about the behavior of the ejector’s free mixing layer from these trends, most of them are established for fully developed turbulent mixing layers. In contrast, the short mixing duct length of typical jet engine ejectors means that the shear layer growth may be influenced by the transitional phase of the
9 shear layer, about which, much less information is available. In addition, Birch and Eggers [5] suggest that the condition and thickness of the internal and external boundary layers, prior to separation from the splitting surface, are influential to subsonic turbulent mixing layer formation, particularly the transitional region. However, no parametric or empirical relations were found in the literature to quantify this affect.
2.2
Previous Jet Engine Applications In the decade following the first flight powered by a turbojet engine, increasing
combustion temperatures and the advent of afterburners lead to a need for cooling in turbojet nozzles [10]. In some of the earliest work on ejectors with application to jet engine nozzles, Huddleston et. al. explored the effectiveness of ejectors for pumping cooling air over temperature critical areas of turbojet exhaust systems. As stated by Kochendorfer et. al, much of the ejector performance data available as late as the writing of Reference [11], in 1951, was inapplicable to jet engine ejector design. At that point, most of the existing work still focused on industrial applications with significantly different geometries and flow ratios. Since the first work aimed at application of ejectors to jet engine exhaust systems, researchers have performed flow testing of numerous, generic, ejector geometries, in an effort to obtain empirical relations between characteristic geometry parameters and ejector performance. There are several such similarity parameters which are traditionally used to characterize ejector geometry for this purpose. The most pervasive of such parameters in the literature appears to be a diameter ratio Rd = Ds /Dp , and a spacing ratio Rs = Ls /Dp (see Figure 2.2). Der [4] suggests that a more appropriate definition of the spacing ratio would be posed L = Ls /∆, which has a more direct relationship to the flow physics governing the ejectors performance, namely shear layer attachment. Most of the tested designs are either axisymmetric or have basic rectangular flowpaths, so as to be well defined by the aforementioned geometry parameters, easing comparison with eachother. Unfortunately, the clamshell type ejector design
10
Figure 2.5. Ejector geometry parameters
currently under consideration is highly three dimensional, and not well described by the above geometry parameters. Ejector application to jet engine exhaust for the purpose of nozzle flow modification (as opposed to cooling) became an area of interest with the advent of supersonic transport (SST) studies. Much work was performed in the U.S. on the utilization of ejector nozzles for super sonic transport (SST) vehicles over the course of 1963-1985 under the federally funded SST, SCAR, and SCR programs. A valuable summary of this work is available in Reference [12]. Generally, two types of ejector nozzles were examined in this work. The first, termed an auxiliary inlet ejector, involves opening small doors linking external air to cavities connected to the divergent nozzle. The second type of ejector nozzle is termed a variable flap ejector. This second ejector type functions by controlling the size of a secondary passage and mixing shroud angle, with a variable flap hinged on the upstream side. These two nozzle types are not directly comparable to the current clamshell-type ejector. However, one lesson that applies, comes in the recognition that drag from the relatively large boat tail angle associated with the deployed flap ejector, tends to eliminate any positive thrust augmentation [12]. The Olympus 593 application on the Concorde SST presents the most relevant ejector geometry for comparison with this work. The Concorde’s ejector noz-
11
Figure 2.6. Concorde exhaust ejector [13]
zle was of the same fundamental shape, location, and operation as that of the current study, as shown in Figure 2.2. Furthermore, the Concorde’s ejector was intended for modification of the exhaust flow in the subsonic regime as well as to produce reversed thrust upon landing [13]. Perhaps the greatest difference between the designs is the significantly larger deployment angle of 21◦ for the OL 593 ejectors [13], compared to a preliminary design value of only 11◦ for the current geometry. This difference in angle has significant implications for boat tail drag as well as internal separation. Both the OL 593 ejectors and the current geometry are at least partly aimed at jet noise reduction. The noise suppression effect of the nozzle ejector has two potential sources. The first, as suggested by Bradshaw et. al [14], is that more thorough mixing and associated shear stress present within the ejector shrouds means more of the jet noise will be generated within the shrouds and physically shielded by them. The other mechanism of ejector noise suppression is the lower mean velocity and shear level of the jet exhaust plume, due to the mixing of the high speed exhaust with lower speed ejector air. Promoting increased mixing within the ejector nozzle is beneficial in both of these regards. This leads to the idea of combining a nozzle ejector with forced mixing at the separating geometry, to further reduce jet noise [2]
12 2.3
Summary There is a wealth of empirical data on the general subject of ejectors. However, due
to their varied applications and geometries, only a much smaller subset of the available background is relevant. In addition, though there is a sufficient body of works relating to generic ejector geometries, the highly 3-dimensional nature of the present design (see Figure 3.2) requires any comparisons with such work to be purely qualitative. For this reason, the experimentation undertaken in this work is all the more necessary. It is possible, however, to obtain better understanding of any geometry’s effect on ejector performance by observing the relationship between the spacing ratio defined by Der (Rs = Ls /∆) and attachment of the mixing layer to the ejector shroud. It is apparent that the performance of the ejector is dominated by that of the free shear layer which forms after separation of the primary and secondary streams from the primary nozzle lip. The mixing effectiveness and growth of this layer will largely determine the performance of any ejector design. In particular, there is strong impetus to ensure that the free mixing layer attaches to the ejector shroud to prevent separated flow, recirculation, and the associated noise and performance consequences. The flow parameters which affect free shear layer growth are, consequently, those best suited to guide this work.
13
3. Test Nozzle Geometry As previously mentioned, the current work is in contribution to research efforts underway at Rolls-Royce to develop a jet engine nozzle for application to a super sonic business jet. In the course of this research and design effort, Rolls-Royce has arrived at a recent design revision which is intended as the cornerstone of the present work. It was the aim of this project, to develop a unique test model which captures some of the fundamental aerodynamic features of the present Rolls-Royce design, while exploring proposed modifications and enabling fabrication and function of an effective test model.
3.1
CAD Model To fabricate the test nozzle geometry utilizing the automated CNC machining
techniques required by such complex geometry, it is necessary to start with a high fidelity CAD definition. Such a CAD model, once produced, also enables collaboration with other members of the research team pursuing CFD analysis. When endeavoring to capture some of the fundamental design characteristics of the current Rolls-Royce definition, it was proposed that the existing CAD model, with the desired modifications, be used to generate the definition for CNC and CFD applications. Unfortunately, the CAD model available to the Purdue research team lacked the necessary fidelity and existed as unparameterized, uneditable surfaces. As such, it was necessary to create a new CAD model. CATIA was chosen for this task, based on the experience and access of the Purdue research team. In addition, the ’solids’ modeling functionality of CATIA is a convenient means of ensuring that component surface definitions uniquely and completely define boundaries to 3-dimensional bodies, thus providing the necessary fidelity.
14 3.1.1
CATIA Design Tables
Given the necessity of creating a new CAD definition for the test geometry, it was desirable for the model to enable rapid modification, whether for feature changes necessary for fabrication vs. CFD, or to facilitate design revisions. It was also desirable for all members of the team, regardless of CATIA experience, to be able to work with the model. The Design Table functionality of CATIA satisfies these requirements. Design Tables allow linking parameter definitions within the CAD model to Excel spreadsheets. Within the spreadsheets, multiple versions of the same model can be defined, with different sets of parameter values. Parameters defining one design occupy one column, and subsequent columns with varied parameter values represent different designs. In this fashion, multiple working models can be maintained in the same CAD file and users can make significant modifications by interacting primarily with a simple spreadsheet of parameter values. Switching between designs is a matter of changing a design identifier in the CAD model, while creating an entirely new design only requires a new column in the spreadsheet with the desired parameter values.
3.1.2
Model Parameterization
A full listing of the parameters and the function of the Design Table is available in a tutorial created for the model and included in this volume as Appendix E. This section describes a subset of the parameters, for illustration of the general method of parameterizing the ejector nozzle geometry. Figure 3.1.2 depicts the parameterization of the ejector slot, shown in cross-section. Each tag in the figure represents a parameter value which is defined via the linked spreadsheet. Some of the more fundamental components of the design are controlled by specifying the coordinates of base points, as shown by the ’Control’ tag in the figure. Most of the other features are defined by simple angular or length offsets from the base points. The more complex aerodynamic surfaces, such as those which
15
Figure 3.1. Ejector Slot Parameterization
make up the ejector slot, are defined by splines. Each spline is defined by a series of control points, which specified by their coordinates. The points labeled ’FT#1’ and ’FT#6’ in Figure 3.1.2 are examples of the control points which define the forward, fixed ejector surface. Much of the fixed nozzle and the exterior of the clamshells have conical definitions, which require only one defining profile. The elliptic portions of the geometry, the fixed ejector slot surface and the interior clamshell surfaces, are controlled with two separate profile definitions and CATIA’s ’multi-section surface’ function. The ’top’ profile of the elliptic components is depicted in Figure 3.1.2 and is a cross-section of the center of the clamshells. The ’side’ profiles cut through the center of the hingeplates, depicted in Figure 3.2.
3.1.3
Model Flexibility
The result of the parameterization scheme is a single CAD model which is capable of representing large variations in design of an ejector nozzle. For the purposes of this work, four designs were created within the design table. The first design was the
16 reproduction of the original Rolls-Royce geometry. The basic geometric components of the model were determined with the goal of being able to reproduce this geometry. The second design created was a model scale version of the same geometry. This simple variation of the first design was utilized to test out the functionality of the Design Table and the stability of the geometric components used to achieve the model design. Once the function of the Design Table driven model was verified with the second design, the work of designing the test geometry was undertaken. A fourth design, with arbitrary modifications, was created to further demonstrate the flexibility of the Design Table driven, parametric model. Figure 3.1.3 illustrates this flexibility with a comparison of the test geometry and the arbitrary variant.
Figure 3.2. Model Flexibility: Comparison of the test nozzle geometry (top) to an arbitrary variation of the model (below).
3.2
Test Nozzle Design The first step in the design of the test nozzle was to set the overall scale. The
nozzle’s 8 inch outer diameter design took into consideration blockage effects in the test section, small scale complications to fabrication and measurement, and system
17
Figure 3.3. CATIA Model of Test Nozzle
flow capability. The 8 inch outer diameter results in a 25 sq. inch throat. A second task in the test nozzle design, was determining a means of attaching the clamshells to the fixed portion of the nozzle. It was also necessary to provide a method of setting the clamshells at desired angles of attack. For these purposes, the test geometry incorporates a set of pins which thread into the fixed geometry and slide into the clamshells. Larger, 1/4” versions of these pins are used as hinges for the clamshells. Smaller, 1/8” pins and the associated holes effect an indexing plate, allowing a set of α values (0, 1.5, 3, 5, 7, 9, 11.5, and 15 degrees). While a mechanism to achieve unlimited clamshell angle settings would be desirable, such a mechanism would be more complex and space is very limited in this portion of the scale model. One of the goals of the test nozzle design was to keep the flowpaths free of cavities and protrusions related purely to the means of attaching the test nozzle to the incorporating test rig. This was achieved by means of assembly rods which run internal to the test rig walls and screw into the mating geometry of the test nozzle. Other design modifications were aimed at easing the parameterization of the model, and preventing excessively
18 thin pieces of geometry, provided the scale. The final design of the test model nozzle is depicted in Figure 3.2.
19
4. Facility Setup 4.1
Requirements The aim of this research was to conduct coaxial flow experiments on proposed
ejector nozzle geometries in the take-off configuration. One flow was to model the mixed exhaust flow of the engine, while the other models the ambient flow of air around the engine nacelle. The existing Boeing Wind Tunnel (BWT) at Purdue University’s Aerospace Sciences Laboratory was identified as a suitable wind tunnel facility for the work. The task, then, was to incorporate a test rig and second air flow system into the test section of the BWT which would enable testing of the nozzle geometry in desired flow conditions.
4.1.1
Existing Facility
The Boeing Wind Tunnel, in its current configuration, is a nearly closed circuit, subsonic wind tunnel with a 6ft x 4ft test section. The test section is 8ft in length and is followed by a 24ft long, 6ft x 4ft, straight duct section. The straight duct section, shown in tan in Figure 4.1, is occasionally used for high Reynold’s number, boundary layer testing. However, for the purposes of this work, it simply serves as part of the tunnel circuit. The test section, shown in charcoal in Figure 4.1, is constructed of four transparent polycarbonate sheets mounted to a steel frame. The space underneath the test section is occupied by an electromechanical force-moment balance system. The side walls of the tunnel are mounted as doors, hinged at the top to facilitate loading and unloading of models during test setup. Another existing facility chosen for incorporation into the experiment setup, is a High Pressure Blower (HPB) built by the Chicago Blower Corporation. The unit is
20
Figure 4.1. ASL Test Facility
a centrifugal fan compressor-blower, specifically a Model 2T-30-28 (similar in performance to the newer Design 53, Size L2), with a 30hp motor, capable of 2950 cfm at 49.1” wg. This unit has been used in the past as the drive for a small (4in x 3in) transonic wind tunnel [15]. The HPB enclosure is depicted in brown in Figure 4.1.
4.2
Facility Additions In order to utilize the HPB as the driver for the second airstream, it was necessary
to transport the blower’s exhaust over the top of the BWT to the the test section. Ducting was sized and installed for this purpose, and can be seen in Figure 4.1. For locating the test geometry in the center of the tunnel flow and providing the secondary airstream, internal to the test rig, it was necessary to mount the test rig using a hollow strut. To accomplish this, several support pieces were designed to transfer the structural loads from the test rig to the test section frame, as well as mate to the second air stream’s transport ducting. These components can be seen in Figure 4.2. Note the two cross beams were constructed of steel for structural rigidity, while the support plate was fabricated from 6061 Aluminum for ease of machining.
21
Figure 4.2. Test Rig Support Structure on BWT Test Section
4.3
Test Rig The goal in designing the rig, was to keep the exterior as clean as possible from
assembly features, such as bolt holes. It is also necessary to incorporate some flow conditioning, particularly due to the sharp turn the flow has to navigate between the strut and the main body of the test rig. The resulting design can be seen in Figures 4.3 and 4.4.
Figure 4.3. BWT Co-axial Flow Test Rig
22 The mount plate, strut and forebody are one welded component. The aftbody tube serves as a settling duct for the flow and easy change out of this component allows for flexibility of design for future test articles. The flow conditioning is achieved via stainless steel honeycomb and screens, located in between the forward and aft tube bodies. The rig is held together with five assembly rods which thread into the test geometry at the aft end, run forward through through the walls of the rig, and are fastened with nuts at the forward end cap. The rig’s nosecone threads onto a stud mounted in the end cap to complete the clean exterior and provide a smooth external flow to the test article at the aft end of the rig. With the exception of the nosecone (which was turned from pvc) and fastening hardware, the rig components are fabricated from aluminum for weight and ease of machining. The outer diameter of the rig was set at 8in, matching the O.D of the test nozzle. Again, this scale was a compromise between making the rig small enough to provide representative flow velocities and avoid blockage affects in the test section, and large enough to facilitate fabrication and higher resolution data collection.
Figure 4.4. Cross-Section View of Test Rig
23 As shown in Figure 4.4, a Pitot probe monitors flow speed in the rig plenum. The values of Vplenum are used to normalize the wake velocities measured by the seven hole probe. Figure 4.4 also depicts the ’rig profile plane,’ the location of the rig surveys performed without the nozzle attached, to ensure an acceptable level of uniform flow through the rig. The results of the rig surveys are presented and discussed in Chapter 7.
4.4
Capability Through the course of the work described in this volume, the BWT has been
undergoing a refitting of its drive components. For this reason, all the wake surveys herein, have tunnel flow only due to that generated by the HPB ejecting into the test section. In general this produces around 5.5 m/s of flow in the test section. This condition of the tunnel limits testing to close to static ambient conditions; however, these conditions are still of great application to the design. This low-speed limitation also allows thorough validation of the structural integrity of the test rig, prior to introducing a higher energy ambient flow. The replacement drive unit is a Model FPDA-2-200-8-10-1160-300, Manufactured by TCF Aerovent. It is driven by a 3 phase, 460V, 300hp electric motor, and is capable of producing 230,000 cfm at a total pressure of 6” wg. The fan selection intent was to be capable of generating 40m/s of flow in the BWT test section.
24
5. Measurement Systems The focus of the work described in this volume is to perform wake pressure surveys using a 7-hole probe. Such wake measurements enable the construction of wake velocity maps as well as thrust analysis, via integral control volume thrust calculations. The desired measurements require three primary instrumentation system components. The first is a multihole probe, which provides a means of tapping the desired pressures at a point in the flow. The second necessary component is a means of measuring the pressure levels transmitted through the probe. Finally, due to the large number of measurement locations necessary to create an adequate wake survey, an automated 2-axis traverse is employed. The details of these three components are the topic of this chapter. The relation of these systems and their setup is depicted in the schematic in Figure 5.
5.1
Seven Hole Probe Multi-hole pressure probes have been utilized for the determination of local flow
angles, static pressure, and total pressure for several decades. A summary of many such devices can be found in the reference by Bryer and Pankhurst [16]. Initially, most multi-hole probes were operated in nulling-mode, where the probe body was rotated until the pressures read at the peripheral holes were in agreement, indicating the probe was aligned with the local flow vector. Once aligned, the flow angles could be read from the probe manipulation hardware; the total pressure was calculated as a proportion of the center port pressure, and the static pressure was calculated as a proportion of the peripheral ports’ pressure. Using probes in nulling mode provides accurate results with minimal calibration; however, it also requires complicated probe rotation hardware, and additional physical space in the test environment to
25
Figure 5.1. Schematic of Measurement Equipment
accommodate such hardware. Consequently, non-nulling methods were eventually developed. For example, a non-nulling methodology was explored by Schuab, Sharp, and Basset [17] for a five-hole probe, to completely determine the local velocity in three dimensional flows, for pitch angles within ±35◦ and yaw angles within ±40◦ . Non-nulling methodologies require more extensive calibration procedures; however, once the calibration is complete, the probe can gather data more quickly and be implemented in a more compact rig than a nulling probe. Furthermore, once a probe is calibrated, it need not be recalibrated unless its geometry is altered by superficial damage or intentional modification. In 1981, Roger Gallington, of the US Air Force Academy, published a paper detailing the development of a seven-hole probe for velocity vector determination in three-dimensional flows [18]. The primary advantages of Gallington’s probe and methodology were threefold. The first was ease of manufacture of the probe. As a
26 matter of simple geometry, seven equally sized cylinders will pack tightly within a larger cylinder, positioning one of the seven precisely in the center of the pack. This feature enables more consistent and simpler fabrication of the probe, by means of packing seven smaller, standard tubes into a standard tube of slightly larger diameter. This eliminates the need for spacers or alignment jigs during fabrication. The other advantages of the seven-hole probe, were both based in a unique sectoring methodology. This method uses different sets of pressure coefficients depending on which port of the probe read the highest pressure. This sectored application enables the probe to accurately measure large flow angles (up to ±80◦ [18]) relative to the probe axis. Another advantage of this methodology, is that it shapes the calibration space in a way as to enable relatively accurate, low-order, polynomial fitting of the calibration surfaces. Given the state of computing power at the time of development, polynomial calibration surfaces enabled efficient, computer based data reduction. Now that computing power is less restrictive, more accurate calibration data-fitting methods, such as Zilliac’s nearest neighbor interpolation [19], are typically used. Former researchers at Purdue’s Aero Sciences Laboratory have chosen the sevenhole probe [15, 20, 21] for its ease of manufacture, durability, relatively low probe volume, and its ability to measure large flow angles. For the work herein, the seven hole probe was selected primarily because of its small probe volume, durability, and the existing capability and experience.
5.1.1
Theory
A seven-hole pressure probe is a means of tapping seven surface pressures in an unknown flow over a known slender body geometry (the probe). The pressure values, and their relations to eachother, are used to solve for the flow angles, static pressure, and total pressure of the local flow in which the slender bodied probe is placed. From these measured values, the velocity vector can be calculated. As mentioned above, the probe can be used with a sectored calibration and data reduction scheme. In this
27 sectoring method, measurements from different sectors (subsets of probe ports) are used to calculate the flow properties, ignoring data from ports over which the flow is likely separated. The subset of ports used for a given measurement is determined by which port measures the highest pressure, and the sector is named for that port. The probe port numbering is shown in Figure 5.1.1. Sector seven is the only sector to use all 7 port measurements, as the flow remains consistently attached over all seven ports, when the center port measures the highest pressure.
Figure 5.2. Front View of Probe
Within the domain of sector 7, the probe can be thought of as a combination of three, traditional, three-tube pressure probes, operated in non-nulling mode. With this analogy in mind, three directional pressure coefficients (5.1,5.2) are formed from the three opposing pairs of peripheral ports [18]. As shown by Zilliac [19], these coefficients are velocity independent for smaller angles and incompressible flow; the meaning of ’smaller angles’ is dependent upon the probe’s cone angle. Previous work by Lindqvist [21] with the particular probe used in this work, has shown that it measures within sector 7 for angles up to 24◦ from the probe axis, which agrees well with Zilliac’s definition of ’small angles.’ Within the this experiment, measured flow angles were expected to be relatively small. Furthermore, given the type of flow field explored in this work, at locations where the flow may exceed the accurate quanti-
28 tative domain of sector 7, separation will be indicated. The qualitative indication of separation is the important result from such measurements; accurate quantitative flow angles in separated regions are not necessary to the purpose of the experiment. For these reasons and the associated simplicity, data reduction for this experiment was performed using only sector 7 coefficients.
CPa =
p4 − p1 p3 − p6 p2 − p5 , C Pb = , and CPc = p7 − p¯ p7 − p¯ p7 − p¯
(5.1)
6
1X p¯ = pi 6 i=1
(5.2)
Zilliac provides equations 5.3 for calculating directional pressure coefficients related to the more familiar pitch and yaw rotations, using the directional coefficients from (5.1). The coordinate system for the 7-hole probe matches that of the overall test setup and is depicted, along with flow angle definitions, in Figure 5.1.1.
CPθ = CPa +
CPb − CP cc 1 and CPψ = √ (CPb + CPc ) 2 3
(5.3)
Figure 5.3. Seven-Hole Probe & Coordinate System The preceding equations describe the coefficients relating pressure distributions to flow angles. What remains is the coefficients relating those same distributions and
29 the ports’ pressure values to local static and total pressure values. The equations for these coefficients, provided by Zilliac [19] are given in (5.4).
CP t =
p7 − p t p¯ − ps and CP s = p7 − p¯ p7 − p¯
(5.4)
The above equations complete the set of coefficients necessary to utilize the sevenhole probe within sector 7. Their application in calibration of the probe and subsequent reduction of experimental data is explained in the following sections.
5.1.2
Calibration
In order to calibrate the probe, it is placed in a known flow, and measurements are taken with the probe at known angles relative to the flow. Typically calibration sets are composed of a rectangular matrix defined by increments in pitch and yaw, over the range of expected experimental flow angles or over the domain of the desired sector(s). Equations 5.1-5.4 are applied to the calibration measurements. The known flow values and the calculated coefficients form calibration curves of the form shown in 5.5.
Cθ = f (θ, ψ) Cψ = f (θ, ψ)
(5.5)
CP t = f (θ, ψ) CP s = f (θ, ψ) The calibration of the probe used in this work was performed by Jens Lindqvist [21]. The data covers pitch and yaw domains of ±44◦ in increments of 4◦ for a total of 529 calibration points. This data, in the form of equations (5.5) can be seen in Appendix D; the plots are limited to the approximate bounds of sector 7 (±24◦ ).
30 5.1.3
Data Reduction
After the experimental data is collected, in the form of seven pressure measurements at each desired location in the flow, a data reduction process must be performed to arrive at the desired data values: flow angles, static pressure, and total pressure. The steps of this process are detailed below. 1. Calculate the directional coefficients for the pressure measurements at each physical location, using equations 5.1. 2. Using the calculated direction coefficients and the calibration curves, solve for the flow angles. The equations necessary for these calculations would be of the form θ = f (CP θ , CP ψ ), where as the calibration data is in the form of (5.5). R To effect the desired functional inverse, Matlab’s griddata function was used,
with a linear, triangular based interpolation. 3. Once the flow angles are known, the next step is determining the total and static pressure coefficients from the flow angles. The same griddata function is used for this step; however, the calibration data is used in its original form (5.5). 4. Using the known pressure coefficients found in the previous step, the total and static pressures are solved for, using equations 5.4, 5.2 and the measured pressure data. 5. With the total and static pressures calculated above, the flow speed can be calculated using Bernoulli’s equation. V =
p 2 ∗ (ptotal − pstatic )/ρ
(5.6)
6. With the flow speed known, equations 5.7 can be used to calculate the flow component speeds per the coordinate system specified in Figure 5.1.1.
31
u = V cos(θ) cos(φ) v = −V cos(θ) sin(ψ)
(5.7)
z = −V sin(θ) As an additional note on data reduction, the following is a procedure for dealing with measurements of severe flow angles or reversed flow. For separated regions where severe angle or reversed flow may exist, the probe is quantitatively non-functional. However, the pressure probe is capable of qualitatively identifying these regions as x¯ >= p7 . At these grid points, the data reduction procedure assigns a velocity value of zero. This is done to prevent singularities or complex numbers in the reduction equations, and to prevent gaps in the presented velocity profiles.
5.2
Pressure Scanner To efficiently measure the multiple pressure levels tapped by the probe, a pressure
scanner is necessary. For these experiments, a Pressure Systems Inc., Model 9010, pressure scanner was utilized. The unit includes 16 individual pressure transducers (channels) read by a common analog-to-digital converter (A2D). The unit will return values for all 16 channels or any specified subset per scan command. The PSI 9010 is capable of internally averaging from 1 to 255 A2D samples per scan command. The unit’s accuracy and performance specifications are declared based on the default setting of 32 samples; this setting was preserved for the measurements in this experiment. Each transducer module in the scanner unit contains its own calibration data as a function of voltage and temperature. This data is created and stored to the transducers at the factory, prior to assembly of the scanner unit. The scanner’s processor uses a third order polynomial to convert the pressure transducers’ voltage signal to engineering pressure units. The formula for this operation is shown in equation 5.8.
32
Pm (Pp , Tp ) = [C0 (Tp ) − Crz + C1 (Tp ) ∗ Vp + C2 (Tp ) ∗
Vp2
+ C3 (Tp ) ∗
(5.8) Vp3 ]
∗ Cspan
Using the calibration data, the scanner dynamically updates the temperature dependent coefficients of the formula to correct for temperature drift. With this method, the scanner limits thermal error to ±0.001%/◦ C. The PSI 9010 is also capable of insitu span and rezero calibration. The manufacturer recommends rezero calibration approximately once per four hours of continuous operation. As a matter of consistency, a rezero calibration was conducted prior to every wake survey experiment. The unit’s accuracy specifications specify that span recalibration is necessary once every 12 months. Since the transducer modules used in this experiment were recently purchased and calibrated at the factory, a span calibration was not necessary during the course of this work. Following the preceding calibration guidelines, the PSI 9010 declares a static accuracy of ±0.10% of full scale. This accuracy specification includes combined errors due to non-linearity, hysteresis, and non-repeatability per ISA S51.1. The transducer modules used for this experiment had a full scale range of ±2.5psid, corresponding to a static accuracy of ±0.0025psi. The software driver for the PSI 9010 was created using Labview. This driver converts graphical user interface (GUI) commands to the Optimux protocol commands recognized by the scanner’s firmware and handles communication of such commands between the host computer and the scanner unit. One feature of note for the scanner’s driver is a multiple scan setting. This setting determines the number of sequential ’scan’ commands sent to the PSI 9010 per survey grid point. Recording multiple measurements per grid point enables calculation of more accurate mean pressure values. In addition, statistical data can be calculated from the multiple measurements. A fluctuation intensity, K, was calculated in this way, according to the equations in (5.9) and (5.10).
u = u¯ + u0
(5.9)
33 q (u0 )2 K= u
(5.10)
The fluctuation intensity is the standard deviation of the temporal domain, pressure measurements as a fraction of their mean. This value is calculated for each data point in the surveys and reported as a 2D profile. Though the formation of the K parameter is the same as that of the traditional turbulent intensity parameter, the PSI 9010’s scan frequency of 20Hz is too low to capture broadband turbulence; hence the choice of a different parameter name. However, the fluctuation intensity is still valuable for indicating the location of shear layers and separated flow in the exhaust wake, as is shown in the data.
5.3
Traverse In order to achieve the large number of probe measurements at accurate grid lo-
cations, an automated, 2-axis traverse was implemented above the BWT test section. This section is a description of the components and setup of the traverse.
5.3.1
Axes & Support Truss
Each axis of the traverse is composed of an aluminum beam, steel tracks fitted to the beam, a drive belt and pulleys, a movable platform, a drive motor, and stepdown gearbox. Movement in an axis is effected by the drive motor, through the 5:1 gearbox, acting to turn the drive pulley. The movable platform, riding on the steel rails and fixed to the drive belt, is then moved when the drive pulley turns. The horizontal axis is fixed to the base of the support truss (see Figure 5.4). The vertical axis is fixed to the movable platform of the horizontal axis. As such, the position of the entire vertical axis moves horizontally when a horizontal move is executed. The base of the truss serves as a stable foundation for the traverse axes, fixing the traverse components to the BWT test section. The upper track of the support truss serves to help in alignment of the vertical axis as well as suppressing the vibration
34 of the probe support. A translating probe holder was fabricated for the traverse to enable measurements over a range of X-axis locations without having to move the support truss and traverse axes. The traverse indexing motors are both Model M34209 manufactured by American Precision Industries (API). The vertical axis motor shaft is equipped with an electrically driven, power-off brake. The brake is a Model B341-01 also by API.
5.3.2
Motor Controllers
The two indexing motors which manipulate the traverse are each driven by separate P315X-H micro-indexing motor controllers made by API. The controllers are connected to the host computer via a daisy-chained RS-435 connection. The controllers offer a variety of parameters to control motor operation, such as top speed, acceleration time, deceleration time, base speed, etc. Prior work was done by Campbell [20] to set these parameters at time efficient values which minimized vibration of the probe support. The values Campbell arrived at were preserved for this work. The micro-indexing controllers, used to drive the indexing motors, communicate using Intelli-Command Language (ICL). A Labview software driver was created to convert GUI commands to ICL commands and handle communication of those commands between the host computer and the controllers. The driver was written to enable both axes to move simultaneously. This enables diagonal movements and makes the surveys more time efficient. The driver also included functionality to continuously display position, and ensure the traverse had stopped moving before proceeding to other actions.
35
Figure 5.4. Test Section Setup
36 5.4
Survey Grids In characterizing locations within a jet plume, a common normalization quotes
distances as a fractional number of ’jet equivalent diameters,’ DEQ . The definition of DEQ can vary for complex nozzle geometries; the definition used for the scale test model and data herein is presented in Equation 5.11. The value of DEQ is used to normalize data locations in the Y-Z survey planes, as well as the X-axis location of the survey planes.
At = π ∗
DEQ 2
2 (5.11)
DEQ = 5.642inches The translating probe holder fabricated for the traverse permits full wake surveys at probe placements from X = 1.0-3.5 DEQ , where X = 0 is at the fixed throat of the nozzle. Within this range, planes at X = 1.0, 1.5, 2.0, and 3.0 DEQ were selected for surveys. In addition, the probe holder can be positioned inside the nozzle to perform internal surveys as far forward as X = 0.42 DEQ , depending on the clamshell setting. The relation of these planes to the test rig is shown in Figure 5.5.
Figure 5.5. Survey Plane Locations
The resolution and size parameters for the selected survey grids are shown in Table 5.4. While a fine grid could be used over the entire survey area and at each axial
37 plane; this was not necessary. Preliminary surveys showed the flow outside the jet exhaust wake to lack significant spatial gradients, as expected. In addition, the time required for full high resolution grids would be prohibitive. The grids used for most of the wake surveys required about 1 hour and 45 minutes each. A higher resolution Cartesian grid was used to capture the more dramatic spatial gradients within the jet plume, while a lower resolution, polar grid was utilized to capture the more spatially consistent flow outside the jet plume. Different grids were employed at each axial location, as the spreading of the jet plume require the higher resolution grid to grow in size as the survey planes step aft from the nozzle. The total number of data points was kept roughly consistent by increasing the spacing of the finer resolution grid to approximate the spreading angle of the jet plume. Table 5.1 Survey Grid Parameters Inner Grid (x,y) X=
Outer Grid (r,θ)
dx
dy
max(r)
dr
dθ
max(r)
Total
[in]
[in]
[in]
[in]
[deg]
[in]
N points
0.42 DEQ
0.12
0.12
3.152
–
–
–
2047
1.0 DEQ
0.13
0.13
5.0
0.5
7.5
9.0
4853
1.5 DEQ
0.16
0.16
6.0
geom(9)
5.0
9.0
5069
2.0 DEQ
0.18
0.18
6.9
geom(7)
5.0
9.0.
5141
3.0 DEQ
0.24
0.24
9.0
–
–
–
4421
geom(#) - indicates geometric spacing of # grid points.
38
6. Flow Visualization This chapter describes the methods of flow visualization that were employed to better understand the 3D structure of the flow, in particular, that near the geometry’s surface. Much of the methodology was inspired by or directly from Yang [22], which was found to be a valuable resource for flow visualization work. Though some of the flow visualization results are presented in the following sections, others are saved for Chapter 7 to supplement the indications of the 7-Hole probe data.
6.0.1
Smoke Wand
Figure 6.1. External Flow Ingestion Through Ejector Slot, α = 11.5◦ , VINF = 5 m/s
A smoke wand was utilized for preliminary visualization of the flow around the test rig. Smooth flow was indicated over the exterior of the rig. Satisfying the most basic measure of an ejector design, Figure 6.0.1 shows external flow being ingested into the ejector slot. Unfortunately, the smoke also indicated a separation bubble on the forward, fixed surface of the ejector slot (Figure 6.0.1). Of particular interest, Figure 6.0.1 depicts external flow passing around the trailing edge of the clamshell, before reversing direction and flowing upstream into the ejector nozzle. This flow pattern indicates separation along the internal clamshell surface.
39
Figure 6.2. Separation bubble in ejector slot, α = 11.5◦ , VINF = 5 m/s
Figure 6.3. Flow reversal at ejector clamshell trailing edge, α = 11.5◦ , VINF = 5 m/s
6.0.2
Tufting
In the first of several methods of surface flow visualization, one of the clamshells was fitted with tufts of white thread. The thread used was medium duty, sewing thread, composed of a polyester core and a cotton wrapping. Methodology for attaching the tufts was taken from [22]. Figure 6.0.2 shows the flow over the exterior and ejector surfaces of the clamshell to be smooth and in general alignment with the freestream flow. Tufting results from the internal clamshell surface are discussed in Chapter 7.
40 Working with the tufts described above suggested that the chosen thread was slightly more rigid than desired for this geometry and flow speed regime. Future efforts might employ a finer thread of natural material such as silk, which is more limp, or a thinner monofilament nylon. In addition, the test geometry’s scale, surface flow speeds, and complex flow structure suggest a need for spatial resolutions which are challenging to obtain with tufting.
Figure 6.4. Ejector Clamshell Tufts, α = 11.5◦ , VINF = 5 m/s
6.0.3
Surface Sediment
In an effort to obtain a higher spatial resolution visualization of the surface flow, surface sediment tracing [22] was pursued. Surface sediment tracing is achieved by mixing various types of fine powder (sediment) with an evaporative liquid petroleum. This liquid suspension is applied to the surface of the geometry with any desirable means, e.g. a paint brush. As the air flows over the body surface, the suspension aligns itself with the surface flow patterns, and the air gradually evaporates the liquid petroleum. Once the petroleum base has completely evaporated, the sediment is left behind in a stable pattern depicting the flow structure at the body surface. As described by Yang [22], the best choice of petroleum base depends on many factors, including flow speed, gravity effects, as well as time constraints associated with test setup and facility operation. For the tests covered in this work, kerosene was used
41 as the base. Yang also suggests a variety of specific sediment powders. For the work herein, powdered chalk was found to be a suitable sediment. The results of the chalk sediment tracing are the first look at the severity of the internal flow separation and the portion of the clamshells inner surface dominated by reverse flow. The sediment trace seen in Figure 6.0.3 shows several important flow features. Two spiral nodes are clearly visible near the center of the clamshell surface. The top and bottom extremities of the clamshell’s inner surface have portions nearly wiped clean of sediment (seen as light triangular patches), indicating high speed flow at the surface. However, the angle of the pattern indicates the flow is at a significant angle to the jet axis. The majority of the internal clamshell surface is not wiped clean and does not posses patterns parallel to the jet flow, indicating separation over most of this surface. Finally, the pattern parallel to the jet axis, near the trailing edge of the clamshell, corresponds to the region where reverse flow was indicated with the smoke wand.
Figure 6.5. Surface Sediment Trace On Clamshell Interior, α = 5.0◦
42 Though the surface sediment tracing achieved satisfactory results for lower clamshell angles (α = 9◦ ), the flow patterns often seemed masked by gravity effects. This was partly a struggle to achieve the appropriate mixture viscosity. However, the continuous evaporation of the kerosene solution made testing a time sensitive endeavor; consistent results were challenging to obtain. For these reasons, surface oil flow was pursued, as described in the following section. The full series of chalk sediment tracings can be seen in Appendix C.
6.0.4
Fluorescent Oil
Surface oil flow visualization involves placing patterns or smooth coatings of oil on an exposed surface and recording the movement of the oil as it is affected by the flow. Initially, artist pigment and linseed oil was used to attempt such visualizations. However, this effort struggled from viscosity, evaporation, and contrast issues similar to the surface sediment tracings. An alternative method, described in Reference [22], utilizes fluorescent oil to overcome the problem with contrast. The fluorescent oil is excited with ultraviolet light, and filmed through a green-pass optical filter. Using this method, the excitation light is mostly eliminated and significant contrast is captured in the oil patterns. Though this method is not explicitly a solution to the viscosity and evaporation concerns, the particular fluorescent oil obtained for this work was 150 cSt polyalkylene glycol (PAG), which proved to be much more stable and a viscosity quite suitable for the geometry and flow speeds of this experiment. The results of the fluorescent oil visualization for α = 5.0◦ are shown in Figure 6.6. The result agrees well with the surface sediment tracing in Figure 6.0.3, while showing much greater detail and contrast. The full series of fluorescent oil visualizations are shown in Appendix B and discussed in Chapter 7.
43
Figure 6.6. Fluorescent Oil Flow for Clamshell AoA = 5.0◦
44
7. Results 7.1
Rig Velocity Profile The test rig, described in Section 4.3, incorporates a section of metal honeycomb
and a means of securing screen sections adjacent to the honeycomb for flow conditioning. Prior to using the rig for testing of the nozzle geometry, the screen size, shape, and blockage necessary to achieve an acceptable level of uniform flow was determined experimentally. Velocity surveys of the rig were performed in the test nozzle attachment plane, shown in Figure 4.4, utilizing the 7-Hole probe and traverse setup. The pressure scanner was set to average 32 A/D samples per measurement and 10 measurements are taken at each spatial grid point. The recorded value is the mean of these 10 measurements (an overall average of 320 samples). The survey was taken using a Cartesian grid with 0.1in spacing in the y and z axes. The measurement system is programmed with a 0.1 second settling time to allow for pressure stabilization after each traverse movement. The velocity profiles of the test rig with no screens, full screens, and shaped, partial screens can be seen in Figure 7.1. The spatial standard deviation, normalized by the spatial mean of the flow gives a quantitative indication of the test rig flow uniformity. These parameters are defined in Equations 7.1 and 7.2.
u = u¯ + u0 q (u0 )2 K= u
(7.1) (7.2)
As expected the unconditioned flow was unacceptable; however, the commonly used, full area dense screens were also ineffective for the pressure drop available.
45
(a) No Screens: K = 0.209
(b) Full Area Screens:
(c) Shaped Partial Screens:
K = 0.143
K = 0.047
(d) Axial Flow Speed (m/s)
Figure 7.1. Rig Velocity Profiles
Consequently, shaped partial screens were fitted to the rig in a trial and error process. The final rig profile is shown as Figure 7.1(c).
7.2
Nozzle Wake Surveys Several preliminary, linear surveys were conducted using the 7-Hole probe, to
gain an understanding of the nozzle’s behavior over the range of available ejector clamshell angles. Examples of these linear surveys (straight line grids in the plane of the clamshell profile definition) are presented in Figure 7.2. The survey data indicated that the ejector suffered significant separation at lower clamshell angles (α Customize. Then, from the dialogue box, select ’Restore All Contents’, click ’OK’, select ’Restore Position’, click ’OK’, and then close the dialogue box. 2. Check for hidden toolbars: when toolbars are fixed to the gray areas along the screen edges, they are considered docked. If too many toolbars are docked in a row, CATIA hides some of them. CATIA indicates that some toolbars are hidden by showing a ’>>’ sign at the bottom of vertical docks or the right side of horizontal docks (See Figure E.1.1). To see the hidden toolbars, remove tool
155 bars from that dock (drag and drop the toolbar header, placing the toolbar in the center area of the screen) until the ’>>’ symbol on that dock disappears.
Figure E.1. CATIA Desktop Icons
CATIA is capable of showing models in several different ways. Some of these visualization methods are less than desirable for the present model. To change the rendering style, select the desired style from the ’View’ toolbar (See Figure E.1.1). When getting started with this model, the ’Shading with Edges without Smooth Edges’ is probably the best view, but feel free to explore the other styles. Finally, CATIA has a Specification Tree edit mode. This mode allows the user to adjust the size and position of the Tree with the same mouse commands that would otherwise modify the view of the model. This mode is entered by clicking with the mouse on the lines that make up the Specification tree. It is easy to accidentally click on one of the lines when attempting to click on an entity of the tree. When this happens, the model goes dark and freezes. To exit the Specification Tree edit mode, simply click again on one of the lines of the Tree.
E.1.2
Modifying the Model
The design table allows for quick generation of multiple versions of the same parameterized model. Each row of the design table contains a separate parameter, while each column represents a different overall design (set of parameter values). A
156 new overall design can be created by adding a column with the desired parameter values to the spreadsheet. To modify a design, it is only necessary to change the parameter values in the spreadsheet, save the spreadsheet, and refresh the model in CATIA (See Figure E.1.2). To switch designs within CATIA, all that is required is changing the specified design number by double clicking on the Design Table object (See Figure E.1.2). It is possible to specify a set of parameters that conflict with eachother, such that the model cannot represent them (the model ’breaks’). This is indicated by error messages after executing a ’Refresh’ command. For those less familiar with CATIA, it would be best to go back to the last version of the model that worked, and gradually change one parameter at a time. Sometimes after a model ’breaks,’ the parameterization becomes unstable. It may be necessary to exit CATIA, without saving, and reopen the model.
Figure E.2. CATIA Specification Tree & Refresh Icon
Figure E.1.2, shows the location on the CATIA desktop of some of the objects necessary to interact with the model. Note that the location of the refresh button may differ, but its appearance will be the same (though it is grayed out when the model does not need refreshing).
157 It is important to double-click on the ’Nozzle Assembly’ entity at the top of the specification tree prior to refreshing the model; this ensures that CATIA is in the Assembly Design Workbench, and that the entire model, not just one of its components, is being updated. CATIA will automatically detect changes to the design table each time you save the excel file. CATIA indicates it has detected a design table change with a pop-up dialogue box, which the user must click ’OK.’ Note that it may take several seconds for CATIA to recognize the change. This also means that every time you modify the design table, it is necessary to save the excel file for CATIA to recognize the modification.
E.1.3
Existing Designs
Four designs already exist in the design table associated with this model. The first is an attempt to exactly match the provided Rolls-Royce geometry: PDA72010/72011. As the definitions/design intent for these geometries was not available, this model must, strictly speaking, be considered an approximation. However, every attempt was made to make it exact. Close visual inspection and measurable dimensions match those of the model in all locations where the PDA72010/72011 geometry is unambiguous and measurable. The second design is a scaled version of the original with small changes to the fixed throat location. The third design reflects some changes made for the purposes of fabricating a test model for Task 7a. The fourth design is a model-scale trial design. Though it is not entirely viable, it demonstrates some of the capability of the model and allows users to begin modifying parameters to see how the model reacts, without changing the three relevant designs.
158 E.2
Parameterization Scheme
E.2.1
Control Plane
The origin of the coordinate system was chosen to lie on the centerline of the nozzle, in the plane of the supersonic throat. The x-axis is the centerline of the nozzle and flow through the nozzle is in the positive x-direction. The positive y-direction goes through one of the fixed sideplates, while the positive z-direction passes through the center of one of the clamshells. While many features, such as the flowpath definition forward of the throat are linked to the supersonic throat plane at the origin, it was convenient to choose a different anchor for many of the aerodynamic features of the model. The beginning of the ejector slot on the outside of the fixed geometry is the foundation for the parameterization of these features. Setting the radius of the geometry at this plane directly limits or determines much of the other geometry. This is a consequence of the spatial interrelation of the internal and external flowpaths, integral to an ejector nozzle. As seen in the figure below, there are two parameters that manipulate the control plane: 1. Throat 2 ControlPlane Offset (in): As seen in Figure E.2.6 and discussed in the Aero Profiles section, changing this value affects the radial position of all the aero profile spline control points. It can be difficult to change this value by any significant amount unless all related parameters are scaled together. 2. ControlPlane Radius (in): Similar to #1 above, but affecting the axial position.
E.2.2
Pre-Throat Flowpath Parameters
The model is intended to capture the geometry as far forward as the beginning of the mixer (while not modeling the mixer itself). The controls for this portion of the design consist of various axial planes at which the flowpath radius can be specified. The flowpath profile between these points is linear and no provision is made for
159
Figure E.3. Control Plane Parameters
filleted transitions at these points, as they are not represented this way in most of the supplied geometry. The parameters that control this geometry are listed below and shown in Figure E.2.2. While the parameters have been motivated and named according to the PDA72010 geometry, they can be used in general fashion (so long as axial ordering is preserved) to create whatever pre-throat flowpath is desired. The parameters can even be given nearly identical values to achieve a straight or simple conical flowpath. 1. Throat 2 Prethroat Offset (in): The prethroat plane contains the beginning of the final convergence leading to the fixed throat. 2. Prethroat Radius (in): The prethroat is an axisymmetric feature with a radius assigned by this parameter. 3. Throat 2 PostMixerPlane (in): This plane defines the end of the mixer convergence. 4. PostMixerFlowpath Radius (in): The post-mixer transition is an axisymmetric feature with radius assigned by this parameter.
160 5. Throat 2 MixerConvergence (in): The Mixer Convergence plane contains the transition from the initial mixer duct angel to the mixer contraction duct angle. 6. MixerConvergence Radius (in): The Mixer Convergence is an axisymmetric feature with radius assigned by this parameter. 7. Throat 2 PreMixerPlane (in): Defines the axial location of the mixer start of the mixer. 8. PreMixerFlowpath Radius (in): Defines the flowpath radius ahead of the mixer. 9. Throat 2 Interface Offset (in): Sets the total axial distance from the throat to the forward-most definition of the nozzle model. 10. Interface OR (in): This parameter defines the outer radius of the foremost edge of the geometry. 11. Interface IR (in): This parameter defines the inner radius of the foremost edge of the geometry.
Figure E.4. Pre-Throat Flowpath Parameters
161 E.2.3
Principle Nozzle Geometry Controls
The principle nozzle parameters are those that most directly affect the geometry of the cruise configuration of the nozzle. These parameters are identified in the list and Figures below.
Figure E.5. Principle Nozzle Parameters
1. FixedThroat Offset (in): This offset sets the axial distance from the clamshell throat to the fixed throat. This directly determines the area of the fixed throat. Unfortunately there is no closed form equation to allow direct specification of the fixed throat area while maintaining appropriate geometric relations with the rest of the throat geometry. This value also affects the fixed throat lip radius. 2. ThroatFillet Radius (in): This is the radius of the supersonic throat, part of which is contained in the clamshells and part of which is on the fixed sideplates. 3. SideCut AftAngle (deg): This angle affects the nozzle’s effective A9 value as well as the extent to which the clamshells can be opened (for thrust reverse mode).
162 4. AftCone Angle (deg): This is the contraction angle of the aft part of the nozzle, beginning at the Control Plane. 5. ExternalTangent Offset (in): The external tangent is the location, aft of the control plane, on the exterior of the clamshells, where the external LE aerodynamic surface becomes tangent to the aft cone surface as described above. 6. MaxInternalDiffusion Angle (deg): This angle is the maximum diffusion angle that occurs in the divergent portion of the supersonic nozzle; this angle occurs in the Y=0 plane. 7. Throat MinorRadius (in): This is the semi-minor axis length of the ellipse defining the supersonic throat. 8. Throat MajorRadius (in): This is the semi-major axis length of the ellipse defining the supersonic throat.
Figure E.6. Supersonic Throat Definition
163 E.2.4
Clamshell Definition Controls
The side profile and hinge action of the ejector clamshells are controlled by the parameters depicted below.
Figure E.7. Clamshell Definition Parameters
1. Hinge AftOf CThroat (in): This dimension is the axial offset from the supersonic throat plane to the clamshell hingeline. 2. HingeLine Offset (in): This is the radial offset of the clamshell hingeline from the nozzle centerline. 3. SideCut AftRadius (in): This is the radius of the side profile, aft curve. This affects the mating geometry on the fixed nozzle. 4. SideCut FwdRadius (in): This is the radius of the side profile, forward curve. This affects the mating geometry on the fixed nozzle, as well. 5. HingeStop Angle (deg): This is the angle defining the lower forward geometry of the clamshell geometry and the mating surface on the fixed nozzle.
164 E.2.5
Hardware Related Parameters
The parameters shown below are those that are primarily concerned with the physical hardware characteristics and less with the aerodynamic behavior of the model.
Figure E.8. Hardware Related Parameters
1. HingePlane Offset (in): This is the radial distance from the centerline of the nozzle to the pivot plane on the sides of the clamshell. It is basically half the width of the clamshell with allowance for the Hinge Gap. 2. Hinge 2 Tail Length (in): This is the axial measure from the clamshell hingeline to the end of the fixed nozzle side plates. 3. AssemblyFeatures Active: This is a logic variable that allows specialization of the design used for fabrication of the test nozzle. It activates/deactivates features that are not desirable or applicable on other designs. This variable is intended to be ’true’ only in the third column. However, for those familiar with Catia, new parameters, linked to the appropriate ’Activity’ variables in Catia, could be created to allow construction of specific features on other individual designs without affecting the versatility of the Design Table driven model.
165 E.2.6
Aero-Profile Parameters
There are two surface definitions inherent to the ejector nozzle geometry that are complex enough as to require spline/control-point definition. These surfaces are the forward part of the clamshell and the outer lip of the fixed nozzle (ejector slot). Each of these surfaces is characterized by spline profiles at the ’top’ and ’side’ (in the y and z planes). Each defining spline is anchored by end points that are controlled by previously mentioned parameters, and has several control points in the middle, which are specified as an (x,y) coordinate pair. Since these features of the geometry are interrelated to several parameters, it can be difficult to change any one of the parameters by a significant value without adjusting all of the related parameters. The coordinate system used for all of these control points is the traditional airfoil coordinate system (different from the rest of the nozzle model), with positive-x in the flow direction, and positive-y in the positive radial direction of the nozzle. Also, in the traditional manner, the origin of the coordinates is at the leading edge of the airfoil (clamshell leading edge). Note that in the image below, the clamshell is slightly open. This is only for illustrative purposes. The fixed spline profile shown in Figure E.2.6 and the Clamshell spline profile shown in Figure E.2.6 are both specified in the same coordinate system, with the clamshell in the fully closed position. This facilitates proper ’clamshell closed’ orientation and spatial relation of the two ejector slot surfaces. For example, in design #3, the inner clamshell leading edge and lower portion of the ejector slot are intended to match up exactly. This is accomplished by setting the first 5 control points of each spline to the same coordinate values. The Leading Edge point of the clamshell is one of the clamshell spline control points, however the parameter nomenclature for this point is special, as these values affect other features, as well: 1. CT LE Inset (in): This is the radial inset difference, from the control radius to the clamshell leading edge point.
166
Figure E.9. Ejector Slot Parameters
2. LE Offset (in): This is the axial offset of the clamshell leading edge from the Control Plane. The parameter names defining the rest of the control points follow a specific convention. The trailing letter specifies which coordinate the dimension is in. The number specifies which control point within a given spline the parameter belongs to. The point numbering for all splines, starts nearest the throat and moves radially outward. The first two letters of the control-point parameter names specify to which spline the control point parameter belongs. The key for these first two letters follows below. Though Figures E.2.6-E.2.6 both show the ’top’ profiles, the side profile definitions are similar. • F = fixed ejector slot surface • C = clamshell leading edge surface • S = side (within the fixed sideplate, at the maximum diameter of the throat, in the y-plane) • T = top (at the center of the clamshell, at the minimum diameter of the throat, in the z-plane)
167
Figure E.10. Ejector Slot Parameters
1. CT LE Radius (in): This value sets the leading edge radius of the clamshell. In the Figure E.2.6 above, a section of the orange-highlighted line contains two white dots, near the throat. This portion of the clamshell definition profile (not part of the spline) is linear, tangent to the constant-radius throat, and intersects the first control point of the clamshell spline. The spline, in turn, is set to be tangential to this linear segment, at the first control point. The mating geometry on the outer portion of the fixed nozzle ejector slot surface is defined in a similar fashion. By this definition, setting the first control point on both the fixed nozzle and clamshell ejector slot surfaces to be identical ensures that the clamshells will mate/seal at the throat when the clamshell is closed, even if it is desired that the rest of the fixed nozzle and clamshell ejector slot surfaces don’t touch in the closed position. As such, this portion of these surfaces is most directly controlled by moving the location of the first control point on each spline.
168 E.2.7
Applications
Two other features of the model which might be of some use are under the ’Applications’ menu of the specification tree. The first is a section view tool. Right clicking the Section View entity and selecting ’Hide/Unhide’ from the pop-up menu will control the display of a 2D section view in the separate window (explained below), as well as an outline of the section cut in the model window. Double clicking on the Section view entity in the tree will open an additional window within CATIA that displays a 2D section cut as well as a dialogue box that lets you adjust the sectioning parameters. Note that this separate window with 2D section cut will respond to changes in the model such as adjusting the clamshell AoA. Right clicking on the section view entity (in the tree) and selecting ’Activate/Deactivate Section Cut’ will cut away the model at the sectioning plane to leave a 3D sectioned model.
Figure E.11. Applications in Specification Tree
The second application is an interference check. There are two interference checks already prepared to check each of the clamshells against the fixed portion of the nozzle. Given the current design scheme, both should give identical results and only checking one is necessary. To run the interference check, double click on either the
169 ’Lower Clamshell’ or ’Upper Clamshell’ interference checks to bring up the dialogue box. Then hit ’Apply’ in the lower corner of the box. The results of the application will be displayed shortly. It is important to check the magnitude of any interference. Collisions can be difficult to completely eliminate given such complex mating surfaces. However, when the parameters are set correctly any collisions should be an order of magnitude less than expected machining tolerances.