Subsonic FixedWing Unmanned Aerial
Vehicles Projected Through 2025
John Frederick Gundlach, IV
Dissertation submitted to the Faculty of the Virginia Polytechnic Institute
and State University in partial fulfillment of the requirements for the degree
of
Doctorate of Philosophy
in
Aerospace Engineering
J.A. Schetz, Chair
B. Grossman
R. Kapania
E. Cliff
R. Foch
February 9, 2004
Blacksburg, Virginia
Keywords: Multidisciplinary Design Optimization, Aircraft Design, Unmanned Aerial
Vehicle (UAV), Technology Predictions
Copyright 2004, John Gundlach
MultiDisciplinary Design Optimization of
Subsonic FixedWing Unmanned Aerial Vehicles
Projected Through 2025
John F. Gundlach
Abstract
Through this research, a robust aircraft design methodology is developed for analysis and
optimization of the Air Vehicle (AV) segment of Unmanned Aerial Vehicle (UAV)
systems. The analysis functionality of the AV design is integrated with a Genetic
Algorithm (GA) to form an integrated Multidisciplinary Design Optimization (MDO)
methodology for optimal AV design synthesis. This research fills the gap in integrated
subsonic fixedwing UAV AV MDO methods. No known single methodology captures
all of the phenomena of interest over the wide range of UAV families considered here.
Key advancements include: 1) parametric Low Reynolds Number (LRN) airfoil
aerodynamics formulation, 2) UAV systems mass properties definition, 3) wing structural
weight methods, 4) selfoptimizing flight performance model, 5) automated geometry
algorithms, and 6) optimizer integration. Multiple methods are provided for many
disciplines to enable flexibility in functionality, level of detail, computational
expediency, and accuracy.
The AV design methods are calibrated against the HighAltitude LongEndurance
(HALE) Global Hawk, MediumAltitude Endurance (MAE) Predator, and Tactical
Shadow 200 classes, which exhibit significant variations in mission performance
requirements and scale from one another. Technology impacts on the design of the three
UAV classes are evaluated from a representative system technology year through 2025.
Avionics, subsystems, aerodynamics, design, payloads, propulsion, and structures
technology trends are assembled or derived from a variety of sources. The technology
investigation serves the purposes of validating the effectiveness of the integrated AV
design methods and to highlight design implications of technology insertion through
future years. Flight performance, payload performance, and other attributes within a
vehicle family are fixed such that the changes in the AV designs represent technology
differences alone, and not requirements evolution. The optimizer seeks to minimize AV
design gross weight for a given mission requirement and technology set.
All three UAV families show significant design gross weight reductions as technology
improves. The predicted design gross weight in 2025 for each class is: 1) 12.9% relative
to the 1994 Global Hawk, 2) 6.26% relative to the 1994 Predator, and 3) 26.3% relative
to the 2000 Shadow 200. The degree of technology improvement and ranking of
contributing technologies differs among the vehicle families. The design gross weight is
sensitive to technologies that directly affect the nonvarying weights for all cases,
especially payload and avionics/subsystems technologies. Additionally, the propulsion
iii
technology strongly affects the high performance Global Hawk and Predator families,
which have high fuel mass fractions relative to the Tactical Shadow 200 family. The
overall technology synergy experienced 1011 years after the initial technology year is
6.68% for Global Hawk, 7.09% for Predator, and 4.22% for the Shadow 200, which
means that the technology trends interact favorably in all cases. The Global Hawk and
Shadow 200 families exhibited niche behavior, where some vehicles attained higher
aerodynamic performance while others attained lower structural mass fractions. The high
aerodynamic performance Global Hawk vehicles had high aspect ratio wings with sweep,
while the low structural mass fraction vehicles had straight, relatively low aspect ratios
and smaller wing spans. The high aerodynamic performance Shadow 200 vehicles had
relatively low wing loadings and large wing spans, while the lower structural mass
fraction counterparts sought to minimize physical size.
iv
Acknowledgements
This research would not be possible without the support, advice, guidance, and other
contributions from a number of individuals and organizations. My wife and best friend,
Katie, provided invaluable encouragement, emotional support, practical advice, time,
personal sacrifice, and patience. I will never forget her loving dedication. My advisory
committee has shown great flexibility in accommodating an unconventional approach to
PhD research. Dr. Schetz, my advisory committee chairman, showed great
resourcefulness and creativity in making distance learning arrangements. Dr. Schetz’s
frequent requests to limit research scope enabled this effort to be completed within a
practical timeframe. Guidance and direction provided by the advisory committee has
significantly shaped the research. I am grateful to Dr. Fred Lutze, who was a great
mentor and guiding influence in my personal and educational development. I will cherish
the memory of this great man. Rick Foch shared his extensive experience with me in this
effort and during my college internships at NRL. My current employer generously
provided tuition funding, which eliminated a financial burden. L.W. championed my
quest for knowledge and understood the personal demands associated with pursuing an
advanced degree while working full time. I would also like to thank R.L., S.O., M.C.,
P.A. and R.A. for their steadfast support throughout the duration.
v
Table of Contents
Abstract ............................................................................................................................... ii
Acknowledgements............................................................................................................ iv
Table of Contents................................................................................................................ v
List of Figures.................................................................................................................... ix
List of Tables .................................................................................................................... xv
Nomenclature.................................................................................................................. xvii
Chapter 1 Introduction..................................................................................................... 1
1.1 Background......................................................................................................... 1
1.2 Problem Statement .............................................................................................. 6
1.3 Approach............................................................................................................. 6
Chapter 2 Methodology................................................................................................... 9
2.1 Methodology Overview...................................................................................... 9
2.2 Methodology Development ................................................................................ 9
2.3 Applied Methodology Architecture .................................................................. 10
2.4 Methods............................................................................................................. 11
2.5 Calibration......................................................................................................... 12
Chapter 3 Optimization ................................................................................................. 13
3.1 Optimization Overview..................................................................................... 13
3.2 Optimization Code Architecture....................................................................... 14
3.3 Design Variables............................................................................................... 16
3.4 Constraints ........................................................................................................ 17
Chapter 4 Technology Modeling................................................................................... 21
4.1 Technology Overview....................................................................................... 21
4.2 Technology Methods ........................................................................................ 23
4.3 Technology Modeling....................................................................................... 25
4.3.1 Aerodynamics ........................................................................................... 25
4.3.2 Avionics and Subystems........................................................................... 30
4.3.3 Design ....................................................................................................... 35
4.3.4 Payloads .................................................................................................... 36
4.3.5 Propulsion ................................................................................................. 39
4.3.6 Structures .................................................................................................. 42
Chapter 5 Geometry....................................................................................................... 46
5.1 Geometry Overview.......................................................................................... 46
5.2 Geometry Definition......................................................................................... 46
5.2.1 Wings and Tails ........................................................................................ 46
5.2.2 Fuselage .................................................................................................... 46
5.2.3 Propeller.................................................................................................... 48
5.3 Parametric Geometry ........................................................................................ 49
5.3.1 Parametric Fuselage.................................................................................. 49
5.3.2 Parametric Tails ........................................................................................ 52
5.4 Wing Lift Distribution Paneling ....................................................................... 53
5.5 Wing Structures Paneling ................................................................................. 54
5.6 Graphics File Generator.................................................................................... 57
vi
Chapter 6 Aerodynamics ............................................................................................... 59
6.1 Aerodynamics Overview .................................................................................. 59
6.2 Aerodynamics Code Architecture..................................................................... 60
6.3 Lift Methodology.............................................................................................. 62
6.3.1 Lift Methodology Overview..................................................................... 62
6.3.2 CG Location Determination...................................................................... 62
6.3.3 Lift Distribution Methods ......................................................................... 63
6.3.4 Pitching Moment....................................................................................... 68
6.4 Drag Methodology............................................................................................ 70
6.4.1 Drag Methodology Overview................................................................... 70
6.4.2 Profile Drag Overview.............................................................................. 71
6.4.3 Interference Drag ...................................................................................... 89
6.4.4 Fuselage Drag ........................................................................................... 91
6.4.5 Wave Drag ................................................................................................ 93
6.4.6 Landing Gear Drag ................................................................................... 94
6.4.7 Cooling Drag............................................................................................. 95
6.4.8 Miscellaneous Drag .................................................................................. 96
Chapter 7 Propulsion ..................................................................................................... 97
7.1 Propulsion Overview ........................................................................................ 97
7.2 Code Architecture ............................................................................................. 97
7.3 Propeller Methods............................................................................................. 98
7.3.1 Propeller Diameter Sizing......................................................................... 99
7.3.2 Disk Actuator Model............................................................................... 100
7.4 PistonProp Engines........................................................................................ 102
7.4.1 PistonProp Engine Deck........................................................................ 104
7.5 Turbojets and Turbofans................................................................................. 104
7.5.1 Turbojet/Turbofan Engine Decks ........................................................... 105
Chapter 8 Size, Weight, and Power............................................................................. 107
8.1 Size, Weight, and Power Overview................................................................ 107
8.2 Generic Systems.............................................................................................. 107
8.2.1 Generic Systems Overview..................................................................... 107
8.2.2 Generic Systems Weights ....................................................................... 108
8.2.3 Generic Systems Size.............................................................................. 109
8.2.4 Generic Systems Power .......................................................................... 110
8.2.5 Generic Systems Thermal Management ................................................. 110
8.3 Specialized Size, Weight, and Power Methods .............................................. 111
8.3.1 Subsystems.............................................................................................. 111
8.3.2 Avionics .................................................................................................. 116
8.3.3 Structures ................................................................................................ 118
8.3.4 Propulsion ............................................................................................... 125
Chapter 9 Flight Performance...................................................................................... 128
9.1 Overview......................................................................................................... 128
9.2 Performance Code Architecture...................................................................... 129
9.3 Performance Methodology.............................................................................. 131
9.3.1 Performance Equations ........................................................................... 131
9.3.2 Performance Constraints......................................................................... 134
vii
9.3.3 Flight Velocity Optimization.................................................................. 135
9.3.4 Climb Performance ................................................................................. 136
9.3.5 Mission Evaluation ................................................................................. 139
9.3.6 Mission Completion Criterion ................................................................ 139
Chapter 10 Calibration............................................................................................... 142
10.1 Calibration Overview...................................................................................... 142
10.2 Calibration Methodology................................................................................ 142
10.3 Calibration Cases ............................................................................................ 143
10.3.1 Global Hawk........................................................................................... 143
10.3.2 Predator ................................................................................................... 155
10.3.3 Shadow 200............................................................................................. 166
Chapter 11 Reference Year CalibrationOptimization Comparisons........................ 177
11.1 Overview......................................................................................................... 177
11.2 Reference Case Comparisons ......................................................................... 177
11.2.1 Global Hawk 1994 Optimum Design ..................................................... 177
11.2.2 Predator 1994 Optimum Design ............................................................. 185
11.2.3 Shadow 200 2000 Optimum Design....................................................... 192
Chapter 12 Technology Case Optimization Applications ......................................... 198
12.1 Overview......................................................................................................... 198
12.2 Global Hawk Technology Impacts ................................................................. 198
12.2.1 Sizing Overview...................................................................................... 198
12.2.2 Geometry................................................................................................. 198
12.2.3 Propulsion ............................................................................................... 202
12.2.4 Weights ................................................................................................... 203
12.2.5 Performance ............................................................................................ 207
12.2.6 Technology Sensitivities......................................................................... 209
12.3 Predator Technology Impacts ......................................................................... 211
12.3.1 Sizing Overview...................................................................................... 211
12.3.2 Geometry................................................................................................. 211
12.3.3 Propulsion ............................................................................................... 215
12.3.4 Weights ................................................................................................... 216
12.3.5 Performance ............................................................................................ 220
12.3.6 Technology Sensitivities......................................................................... 222
12.4 Shadow 200 Technology Impacts................................................................... 223
12.4.1 Sizing Overview...................................................................................... 223
12.4.2 Geometry................................................................................................. 223
12.4.3 Propulsion ............................................................................................... 226
12.4.4 Weights ................................................................................................... 227
12.4.5 Performance ............................................................................................ 230
12.4.6 Technology Sensitivities......................................................................... 232
Chapter 13 Conclusions............................................................................................. 234
13.1 AV Design Methodology Conclusions ........................................................... 234
13.2 Calibration Cases Conclusions ....................................................................... 235
13.3 Technology Cases Conclusions ...................................................................... 235
13.4 Recommendations........................................................................................... 237
Appendix A Geometry Methods.................................................................................... 239
viii
Appendix B Blade Element Propeller Methods............................................................ 247
Appendix C Selected Avionics and Subsystem Weights Methods................................ 249
Bibliography ................................................................................................................... 256
List of Acronyms ............................................................................................................ 262
Formatting Notes ............................................................................................................ 264
Vita.................................................................................................................................. 265
ix
List of Figures
Figure 11 Model of German V1 Cruise Missile............................................................... 1
Figure 12 AQM34 Firefly Currently in Operational Service.......................................... 2
Figure 13 Models of the Teledyne Ryan Drones (Left to Right: Global Hawk, Compass
Cope, AQM154 Compass Arrow, AQM147H Firebee, AQM34 Firefly).............. 2
Figure 14 Models of Army Battlefield UAVs (Left to Right: Shadow 200, Outrider,
Hunter, Aquila) ........................................................................................................... 3
Figure 15 Global Hawk .................................................................................................... 3
Figure 16 Predator ............................................................................................................ 4
Figure 17 Shadow 200...................................................................................................... 4
Figure 18 Scale Comparison of Shadow 200, Predator, and Global Hawk (Left to Right)
..................................................................................................................................... 7
Figure 21 ACMAIN Code Architectre ............................................................................. 10
Figure 22 EVAL Code Architecture................................................................................. 11
Figure 31 Genetic Algorithm Code Architecture .......................................................... 14
Figure 41 Technology Parameter Ratio Model (Method 1) ............................................ 23
Figure 42 Direct Technology Parameter Lookup Model................................................. 25
Figure 43 Induced Drag Multiplication Factor Technology Trends................................ 26
Figure 44 Interference Drag Trends ............................................................................... 27
Figure 45 Wave Drag Technology Factor Trends .......................................................... 27
Figure 46 Profile Drag Multiplication Factor Trends..................................................... 28
Figure 47 Final Profile Drag Multiplication Factor Trends. .......................................... 29
Figure 48 GNC System Technology Phase. ................................................................... 31
Figure 49 GNC System Technology Trends. ................................................................. 31
Figure 410 INS Technology Phase................................................................................. 32
Figure 411 INS Technology Trends ............................................................................... 32
Figure 412 Subsystems Technology Phase .................................................................... 33
Figure 413 Subsystem Weight Technology Trend......................................................... 34
Figure 414 Satellite Communications Dish Antenna Diameter Trends ......................... 35
Figure 415 Weight Growth Technology......................................................................... 36
Figure 416 Payload Technology Phase .......................................................................... 37
Figure 417 SAR Payload Technology Trends................................................................ 37
Figure 418 EO/IR Payloads............................................................................................ 38
Figure 419 Large EO/IR Payload Performance Technology Trend............................... 38
Figure 420 Small EO/IR Payload Performance Technology Trend. .............................. 39
Figure 421 UAV Jet Engine ........................................................................................... 40
Figure 422 Jet Engine Technology Phase....................................................................... 40
Figure 423 Jet Engine Technology Trends..................................................................... 41
Figure 424 UAV Internal Combustion Engine............................................................... 41
Figure 425 Internal Combustion Engine Technology Phase.......................................... 42
Figure 426 Internal Combustion Engine Technology Trends ........................................ 42
Figure 427 Factor of Safety Technology Trends............................................................. 43
Figure 428 Structural Technology Phase........................................................................ 44
Figure 429 Structural Weight Multiplication Factor Technology Trend ....................... 45
x
Figure 51 Predator Fuselage Nose Region ..................................................................... 47
Figure 52 X45A Geometry............................................................................................ 48
Figure 53 Dragon Eye Fuselage Geometry .................................................................... 48
Figure 54 Propeller Integration on Generic Fuselage..................................................... 49
Figure 55 Jet Integration on Generic Fuselage............................................................... 50
Figure 56 Wing Fairing Geometry ................................................................................. 50
Figure 57 Radome Impacts on Fuselage Geometry........................................................ 51
Figure 58 EO/IR Ball Geometry..................................................................................... 52
Figure 59 Virginia Tech Forestry UAV .......................................................................... 53
Figure 510 Wing Aerodynamic Paneling for Virginia Tech Forestry UAV.................. 54
Figure 511 Vehicle Structural Paneling Example ........................................................... 57
Figure 512 Wing Tip Paneling Detail.............................................................................. 57
Figure 513 Example Unconventional Configuration...................................................... 58
Figure 514 TiltWing Graphics Option .......................................................................... 58
Figure 61 AEROC code Architecture.............................................................................. 61
Figure 62 Pitching Moment Subroutine Architecture .................................................... 62
Figure 63 NACA 0009 at Several Reynolds Numbers.................................................... 72
Figure 64. Selig Donnovan SD7032 at Low Reynolds Numbers.................................... 72
Figure 65 Airfoil 3D Interpolation Space ..................................................................... 73
Figure 66 Drag Polar Interpolation Diagram for Two Airfoils ...................................... 75
Figure 67 Angle of Attack Symmetry Diagram............................................................. 77
Figure 68 Curve Fit for 360Degree Angle of Attack Sweep......................................... 77
Figure 69 Example Drag polar and Lift Curve of an Airfoil Through a Large Angle of
Attack Sweep ............................................................................................................ 79
Figure 610 Upper and Lower Transition Location vs. Lift Coefficient Comparison
Between XFOIL and Empirical Equation Matching for the NACA 2412 Airfoil.... 86
Figure 611 The Pressure Drag Comparison Between XFOIL and the Empirical Equation
Matching for the NACA 2412 Airfoil ...................................................................... 86
Figure 612 Comparison of Drag Polars Between the Direct Data and Parametric Airfoil
Interpolation Methods............................................................................................... 87
Figure 613 Comparison Between XFOIL and Wind Tunnel Data. ................................ 87
Figure 614 WingFuselage Interference Geometry........................................................ 90
Figure 615 Fuselage Drag Cross Section Penalty ........................................................... 92
Figure 616 Fuselage Drag Maximum Diameter Location Penalty.................................. 92
Figure 71 Propulsion Code Architecture......................................................................... 98
Figure 72 Raymer Propeller Sizing Equation Results.................................................... 99
Figure 73 Comparison Between Predicted and Ideal Propeller Efficiencies................ 102
Figure 74 Density Ratio Impacts on Brake Specific Fuel Consumption...................... 103
Figure 81 Macci Wing Factors ...................................................................................... 121
Figure 91 Performance Code Architecture................................................................... 129
Figure 92 Flight Envelope Boundaries......................................................................... 134
Figure 93 Velocity Search Initialization....................................................................... 135
Figure 94 Standard Mission Profile............................................................................... 139
Figure 101 Global Hawk ............................................................................................... 144
Figure 102. Global Hawk Engine Thrust Performance ................................................ 145
Figure 103 Global Hawk Engine Specific Fuel Consumption Performance................ 145
xi
Figure 104. Global Hawk Avionics Weights Summary............................................... 146
Figure 105 Global Hawk Subsystems Weights Summary............................................ 147
Figure 106 Global Hawk Structural Weight Summary ................................................ 149
Figure 107 Global Hawk Weights Summary ............................................................... 151
Figure 108 Global Hawk Mass Fraction Summary...................................................... 151
Figure 109 Global Hawk Trimmed Incompressible Aerodynamic Performance......... 152
Figure 1010 Global Hawk Trimmed Drag Contributors .............................................. 152
Figure 1011 Global Hawk Altitude Profile .................................................................. 153
Figure 1012 Global Hawk Mach Profile ...................................................................... 153
Figure 1013 Global Hawk MachAltitude Envelope.................................................... 154
Figure 1014 Predator .................................................................................................... 155
Figure 1015 Predator Installed Thrust Performance...................................................... 156
Figure 1016 Predator Installed Specific Fuel Consumption Performance .................... 157
Figure 1017 Predator Avionics Weights Summary...................................................... 157
Figure 1018 Predator Subsystems Weights Summary ................................................. 159
Figure 1019 Predator Structural Weight Summary...................................................... 160
Figure 1020 Predator Weights Summary ..................................................................... 162
Figure 1021 Predator Mass Fraction Summary............................................................ 162
Figure 1022 Predator Trimmed Aerodynamic Performance ......................................... 163
Figure 1023 Predator Trimmed Drag Contributors ...................................................... 163
Figure 1024 Predator Altitude Profile ........................................................................... 164
Figure 1025 Predator Velocity Profile.......................................................................... 164
Figure 1026 Predator VelocityAltitude Envelope....................................................... 165
Figure 1027 Shadow 200.............................................................................................. 166
Figure 1028 Shadow 200 Installed Thrust Performance ............................................... 167
Figure 1029 Shadow 200 Installed Specific Fuel Consumption Performance............. 168
Figure 1030 Shadow 200 Subsystems Weights Summary........................................... 169
Figure 1031 Shadow 200 Structural Weight Summary................................................ 170
Figure 1032 Shadow 200 Weights Summary............................................................... 172
Figure 1033 Shadow 200 Mass Fraction Summary...................................................... 172
Figure 1034 Shadow 200 Trimmed Aerodynamic Performance.................................. 173
Figure 1035 Shadow 200 Trimmed Drag Contributors................................................ 173
Figure 1036 Shadow 200 Altitude Profile..................................................................... 174
Figure 1037 Shadow 200 Velocity Profile ................................................................... 175
Figure 1038 Shadow 200 VelocityAltitude Envelope................................................. 175
Figure 111 1994 Global Hawk Geometry Comparison – Top View............................ 179
Figure 112 1994 Global Hawk Geometry Comparison – Side View........................... 179
Figure 113 1994 Global Hawk Geometry Comparison – Isometric View................... 180
Figure 114 Global Hawk Geometry Comparison......................................................... 180
Figure 115 Global Hawk Summary Weights Comparison........................................... 181
Figure 116 Global Hawk Summary Mass Fraction Comparison ................................. 181
Figure 117 Global Hawk Altitude Profile Comparison................................................. 182
Figure 118 Global Hawk Mach Profile Comparison.................................................... 182
Figure 119 Global Hawk MachAltitude Envelope Used ............................................ 183
Figure 1110 Global Hawk Specific Fuel Consumption Comparison............................ 183
Figure 1111 Global Hawk Lift to Drag Comparison.................................................... 184
xii
Figure 1112 1994 Predator Geometry Comparison – Top View.................................. 186
Figure 1113 1994 Predator Geometry Comparison – Side View................................. 186
Figure 1114 1994 Predator Geometry Comparison – Isometric View......................... 187
Figure 1115 Predator Geometry Comparison............................................................... 187
Figure 1116 Predator Weights Summary Comparison................................................. 188
Figure 1117 Predator Summary Mass Fraction Comparison ....................................... 188
Figure 1118 Predator Altitude Profile Comparison....................................................... 189
Figure 1119 Predator Velocity Profile Comparison..................................................... 189
Figure 1120 Predator VelocityAltitude Envelope Used.............................................. 190
Figure 1121 Predator Specific Fuel Consumption Comparison................................... 190
Figure 1122 Predator Lift to Drag Comparison............................................................ 191
Figure 1123 2000 Shadow 200 Geometry Comparison – Top View............................ 193
Figure 1124 2000 Shadow 200 Geometry Comparison – Side View............................ 193
Figure 1125 2000 Shadow 200 Geometry Comparison – Isometric View................... 193
Figure 1126 Shadow 200 Geometry Comparison ........................................................ 194
Figure 1127 Shadow 200 Summary Weights Comparison .......................................... 194
Figure 1128 Shadow 200 Summary Mass Fraction Comparison................................. 195
Figure 1129 Shadow 200 Altitude Profile Comparison ................................................ 195
Figure 1130 Shadow 200 Velocity Profile Comparison............................................... 196
Figure 1131 Shadow 200 VelocityAltitude Envelope Used ....................................... 196
Figure 1132 Shadow 200 Specific Fuel Consumption Comparison............................. 197
Figure 1133 Shadow 200 Lift to Drag Ratio Comparison............................................ 197
Figure 121 Global Hawk Geometry Trends – Top View, Years (Left to Right): 1994,
2000, 2005, 2010, 2015, 2020, 2025....................................................................... 199
Figure 122 Global Hawk Geometry Trends – Side View, Years (Left to Right): 1994,
2000, 2005, 2010, 2015, 2020, 2025....................................................................... 199
Figure 123 Global Hawk Geometry Trends – Isometric View, Years (Left to Right):
1994, 2000, 2005, 2010, 2015, 2020, 2025............................................................. 200
Figure 124 Global Hawk Wing Span Trends ............................................................... 200
Figure 125 Global Hawk Aspect Ratio Trends ............................................................ 201
Figure 126 Global Hawk Wing Sweep Trends............................................................. 201
Figure 127 Global Hawk Thickness to Chord Ratio Trends ........................................ 201
Figure 128 Global Hawk Wing Parameter Comparisons ............................................. 202
Figure 129 Global Hawk Fuselage Length Trends........................................................ 202
Figure 1210 Global Hawk Uninstalled Thrust Trends .................................................. 203
Figure 1211 Global Hawk Thrust to Weight Ratio Trends .......................................... 203
Figure 1212 Global Hawk Weight Trends .................................................................... 204
Figure 1213 Global Hawk Mass Fraction Trends ........................................................ 204
Figure 1214 Design Gross  Obligated Weights Trends............................................... 204
Figure 1215 Global Hawk Avionics Weight Trends.................................................... 205
Figure 1216 Global Hawk Subsystems Weight Trends................................................ 206
Figure 1217 Global Hawk Payload Weight Trends....................................................... 206
Figure 1218 Global Hawk Structures Weight Trends .................................................. 207
Figure 1219 Global Hawk Altitude Profile Trends ...................................................... 208
Figure 1220 Global Hawk Mach Profile Trends .......................................................... 208
Figure 1221 Global Hawk Mission Lift to Drag Ratio Trends..................................... 208
xiii
Figure 1222 Global Hawk Mission Fuel Consumption Trends.................................... 209
Figure 1223 Global Hawk Design Gross Weight Sensitivity to Technology............... 210
Figure 1224 Predator Geometry Trends – Top View, Years (Left to Right): 1994, 2000,
2005, 2010, 2015, 2020, 2025................................................................................. 212
Figure 1225 Predator Geometry Trends – Side View, Years (Left to Right): 1994, 2000,
2005, 2010, 2015, 2020, 2025................................................................................. 212
Figure 1226 Predator Geometry Trends – Isometric View, Years (Left to Right): 1994,
2000, 2005, 2010, 2015, 2020, 2025....................................................................... 213
Figure 1227 Predator Wing Span Trends ..................................................................... 213
Figure 1228 Predator Aspect Ratio Trends ................................................................... 214
Figure 1229 Predator Thickness to Chord Ratio .......................................................... 214
Figure 1230 Predator Wing Area.................................................................................. 214
Figure 1231 Predator Fuselage Length Trends.............................................................. 215
Figure 1232 Predator Uninstalled Power Trends .......................................................... 215
Figure 1233 Predator Power to Weight Ratio .............................................................. 216
Figure 1234 Predator Weight Summary Trends............................................................ 216
Figure 1235 Predator Summary Mass Fraction Trends................................................ 217
Figure 1236 Predator Design Gross – Obligated Weight Trends................................. 217
Figure 1237 Predator Avionics Weight Trends............................................................. 218
Figure 1238 Predator Subsystems Weight Trends ........................................................ 218
Figure 1239 Predator Payload Weight Trends .............................................................. 219
Figure 1240 Predator Structures Weight Trends ........................................................... 219
Figure 1241 Predator Altitude Profile Trends .............................................................. 220
Figure 1242 Predator Velocity Profile Trends ............................................................. 221
Figure 1243 Predator Mission Lift to Drag Ratio Trends ............................................ 221
Figure 1244 Predator Mission Fuel Consumption Trends............................................ 222
Figure 1245 Predator Design Gross Weight Sensitivity to Technology ...................... 222
Figure 1246 Shadow 200 Geometry Trends – Top View, Years (Left to Right): 2000,
2005, 2010, 2015, 2020, 2025................................................................................. 223
Figure 1247 Shadow 200 Geometry Trends – Side View, Years (Left to Right): 2000,
2005, 2010, 2015, 2020, 2025................................................................................. 223
Figure 1248 Shadow 200 Geometry Trends – Isometric View, Years (Left to Right):
2000, 2005, 2010, 2015, 2020, 2025....................................................................... 224
Figure 1249 Shadow 200 Wing Span Trends............................................................... 224
Figure 1250 Shadow 200 Aspect Ratio Trends............................................................ 225
Figure 1251 Shadow 200 Fuselage Length Trends ....................................................... 225
Figure 1252 Shadow 200 Wing Loading – Aspect Ratio Relationships ....................... 226
Figure 1253 Shadow 200 Uninstalled Power Trends .................................................... 226
Figure 1254 Shadow 200 Power to Weight Ratio Trends............................................ 227
Figure 1255 Shadow 200 Weight Trends...................................................................... 227
Figure 1256 Shadow 200 Summary Mass Fraction Trends ......................................... 228
Figure 1257 Shadow 200 Design Gross – Obligated Weights Trend........................... 228
Figure 1258 Shadow 200 Avionics Weight Trends...................................................... 229
Figure 1259 Shadow 200 Subsystems Weight Trends ................................................. 229
Figure 1260 Shadow 200 Payload Weight Trends. ...................................................... 230
Figure 1261 Shadow 200 Structures Weight Trends.................................................... 230
xiv
Figure 1262 Shadow 200 Altitude Profile Trends........................................................ 231
Figure 1263 Shadow 200 Velocity Profile Trends ....................................................... 231
Figure 1264 Shadow 200 Mission Lift to Drag Ratio Trends ...................................... 232
Figure 1265 Shadow 200 Mission Fuel Consumption Trends ..................................... 232
Figure 1266 Shadow 200 Design Gross Weight Sensitivity to Technology................ 233
Figure 131 General Atomics GnatPredator Family Evolution ……………………….237
Figure A1 Wing Geometry Parameters ......................................................................... 239
Figure B1 Comparison Between Two and Three Bladed Propeller Designs………….248
xv
List of Tables
Table 31 Universal Design Variables............................................................................. 16
Table 32 Jet Propulsion Design Variables...................................................................... 17
Table 33 Reciprocating Engine Propulsion Design Variables ....................................... 17
Table 34 Constraints....................................................................................................... 18
Table 35 Mission constraint types.................................................................................. 18
Table 41 Technology Attributes..................................................................................... 22
Table 42 Attribute Ranking Description ........................................................................ 22
Table 43 Aerodynamics Technology Attributes............................................................. 25
Table 44 Airfoil Characteristics. .................................................................................... 28
Table 45 Avionics and Subsystems Technology Attributes........................................... 30
Table 46 Design Technology Attributes......................................................................... 35
Table 47 Payload Technology Attributes ....................................................................... 36
Table 48 Propulsion Technology Attributes................................................................... 39
Table 49 Structures Technology Attributes.................................................................... 43
Table 61 Primary aerodynamics product list. ................................................................. 60
Table 62 Lift Distribution Methods................................................................................. 63
Table 63 Profile Drag Methods ....................................................................................... 73
Table 64 Low Reynolds number airfoil database. .......................................................... 80
Table 65 High Reynolds number airfoil database. ......................................................... 81
Table 66 Empirical equation airfoil database. ................................................................ 82
Table 81 Component Categories and Types................................................................. 107
Table 82 Flight Control Systems Weight Methods ...................................................... 111
Table 83 Flight Control System Weight Method Functional Dependencies................ 112
Table 84 Environmental Control System Weight Methods.......................................... 113
Table 85 ECS System Weight Method Functional Dependencies ............................... 113
Table 86 Electrical System Weight Methods ............................................................... 114
Table 87 Electrical System Weight Method Functional Dependencies ....................... 114
Table 88 Fuel System Weights Methods...................................................................... 115
Table 89 Fuel System Weight Method Functional Dependencies ............................... 115
Table 810 Air Data System Factors.............................................................................. 117
Table 811 Wing Weight Methods ................................................................................ 118
Table 812 Wing Weight Method Functional Dependencies ........................................ 118
Table 813 Tail Weight Methods................................................................................... 123
Table 814 Horizontal Tail Weight Method Functional Dependencies......................... 123
Table 815 Vertical Tail Weight Method Functional Dependencies............................. 123
Table 816 Fuselage Weight Methods ............................................................................ 123
Table 817 Fuselage Weight Method Functional Dependencies ................................... 124
Table 101 Global Hawk Geometry.............................................................................. 144
Table 102 Global Hawk Propulsion .............................................................................. 144
Table 103 Global Hawk Design Technology Levels.................................................... 154
Table 104 Predator Geometry Characteristics.............................................................. 155
Table 105 Predator Propulsion Characteristics ............................................................ 156
Table 106 Predator Design Technology Levels ........................................................... 165
Table 107 Shadow 200 Geometry Characteristics ....................................................... 166
xvi
Table 108 Shadow 200 Propulsion Characteristics ...................................................... 167
Table 109 Shadow 200 Calibration weight comparison................................................ 172
Table 1010 Shadow 200 Design Technology Levels ................................................... 176
Table 111 Global Hawk Design Variables................................................................... 178
Table 112 Predator Design Variables........................................................................... 185
Table 113 Shadow 200 Design Variables..................................................................... 192
Table C1 Autopilot Factors ........................................................................................... 250
Table C2 INS Factors ................................................................................................... 251
Table C3 Camera Factors ............................................................................................. 252
xvii
Nomenclature
a Axial interference factor
Thrust ratio
a' Rotational interference factor
A(i,j) Influence coefficient in lift distribution methods
ACFunc Normalized distance from wing AC to trailing edge
alt Altitude, ft
alt
max
Maximum prescribed altitude, ft
AR Aspect ratio
AR
w
Wing aspect ratio
a
surface
Surface lift curve slope, 1/deg or 1/radian
a
tail
Tail lift curve slope, 1/deg or 1/radian
a
wing
Wing lift curve slope, 1/deg or 1/radian
b(i) Influence coefficient in lift distribution methods
BHP Brake Horsepower, HP
BHP
0
Brake Horsepower at sea level, HP
BHP
h
Brake Horsepower at altitude, HP
b
ref
Reference wing span, ft
BSFC Brake specific fuel consumption, lb/HP*hr
BSFC
0
Brake specific fuel consumption at sea level, lb/HP*hr
BSFC
h
Brake specific fuel consumption at altitude, lb/HP*hr
b
w
Wing span, ft
c Local chord of interest (interference drag), ft
Blade section chord, ft
c(i) Local chord (paneling), ft
c (bar) Mean aerodynamic chord length, ft
C Cooling power, Watts
C
d
Section drag coefficient
C
D
3D drag coefficient
C
d
2D drag coefficient
C
D,cavity
Landing gear bay cavity drag coefficient
C
D,cool
Cooling drag coefficient
C
D,door
Landing gear door drag coefficient
C
d,frict
2D friction coefficient
C
D,gear
Landing gear drag coefficient
C
di
(i) Local induced drag coefficient
C
Di
Induced drag coefficient
C
Dint
Interference drag coefficient
C
Dint,wall
Interference drag coefficient due to perpendicular wall
C
Dint,lift
Interference drag coefficient increment due to lift
C
Dint,sweep
Interference drag coefficient increment due to sweep
C
Dint,incline
Interference drag coefficient increment due to inclination
C
d,press
2D pressure drag coefficient
C
D,strut
Landing gear strut drag coefficient
C
D,wheel
Landing gear wheel drag coefficient
xviii
c
flap
(i) Local flap chord, ft
C
l
Section lift coefficient
C
l
(i) Local 2D lift coefficient (paneling)
C
L
Total 3D lift coefficient per surface or total vehicle
C
L,Max
3D maximum lift coefficient
C
Lstall
3D stall lift coefficient
C
lstall
2D stall lift coefficient
C
l0
2D minimum drag lift coefficient
C
M
Total pitching moment coefficient
C
M,CL
Pitching moment coefficient due to lift
C
M,Flap
Pitching moment coefficient due to flap deflections
C
M,fuselage
Fuselage pitching moment coefficient
C
M,Fuse,Nac
Pitching moment coefficient due to fuselages and nacelles
C
M,0L
Pitching moment coefficient due to airfoil sections at zero lift
C
m,0L
(i) Pitching moment coefficient of local airfoil
C
P
Coefficient of power (conventional normalized by rotation terms)
C
PV
Coefficient of power (normalized by V terms)
c
r
c
r,w
(j) Segment root chord to wing root chord ratio
c/R
p
(j) Propeller radiusnormalized chord length
c
r,w
Wing root chord, ft
C
T
Coefficient of thrust (conventional normalized by rotation terms)
c
t
c
r,w
(j) Segment tip chord to wing root chord ratio
C
TV
Coefficient of thrust (normalized by V terms)
c
wing
Wing chord
Cx Torque force coefficient
Cy Thrust force coefficient
D Propeller diameter, ft
Drag, lbs
D
Cool
Cooling Drag, lbs
D
Dish
Communications dish diameter, ft
D
Jet
Jet engine diameter, ft
D
Pow
Power Extraction Drag, lbs
D
Ram
Ram drag, lbs
D
Tire
Tire width, ft
e Span efficiency factor
E
seg
Segment endurance, hrs
E
subseg
Subsegment endurance, hrs
F Prandtl momentum loss factor
fact(i) Interpolation factor
factpen(i) Penalty factor
F
BHP
Brake horsepower altitude correction factor
F
BSFC
Brake specific fuel consumption altitude correction factor
F
Design
Design level factor
FF Airfoil form factor
FF
Fuse
Fuselage form factor
flap(j) Flap function designator, 1=flap, 2=elevator, 3=aileron
xix
F
Tech
Final technology factor
f
type
(j) Flap type, 1=split, 2=plain, 3=slotted flap, 4=double slotted
F
Vol,Install
Installation volume factor
F
Wt,Install
Weight installation factor
G Circulation function
H
Eng
Engine height, ft
h
n
Neutral point normalized by the MAC
inc
w
Wing incidence, degrees
I
xx
XX area moment of inertia, ft
4
I
yy,mass
YY mass moment of inertia, lbm*ft
2
J Advance ratio
K Goldstein momentum loss factor
K Lift dependent profile drag constant
k
a
Aspect ratio correction factor for downwash derivative
k
tr
Taper ratio correction factor for downwash derivative
k
h
Tailplane height correction factor for downwash derivative
L
ACVT,TE
Length from the wing AC to the tail root trailing edge, ft
L/D Aircraft lift to drag ratio
L/D
fuse
Fuselage length to maximum diameter ratio
L
Door
Length of landing gear door, ft
L
Elec
Electrical wiring length, ft
L
Eng
Engine length, ft
L
fuse
Fuselage length, ft
L
H
Horizontal tail moment arm, ft
L
Jet
Length of jet engine, ft
Load
w
Lifting load of a wing, lbs
L
Overall
Overall vehicle length, ft
L
Tail
Average tail moment arm, ft
L
V
Vertical tail moment arm, ft
m(dot)
cool
Cooling air mass flow rate
M Flight Mach number
mac
W
Wing mean aerodynamic chord, ft
M
crit
Critical Mach number
M
DD
Drag divergent mach number
MF
Fuel
Fuel mass fraction
M
max
Maximum Mach number
n Index number in Prandtl LLT
Curve fit parameter in parametric airfoil method
Propeller revolutions per second, 1/sec
N Maximum value of index n in Prandtl LLT
N
Crew
Number of crew members in aircraft
N
blade
Blade count
N
FCS
Number of flight control systems
N
Gen
Number of generators
N
LLF
Limit load factor
N
strut
Number of struts
xx
N
Surf
Number of control surfaces
N
Tanks
Number of fuel tanks
N
tire/strut
Number of tires per strut
N
Ult
Ultimate load factor
P Power absorbed by propeller, HP
Component power, Watts
P
abs
Absorbed power, HP
P
Dish
Power used by dish antenna, Watts
P
Eng
Engine power, HP
P
Max
Maximum power draw of electrical system, Watts
P
rot
Wing rotation 3D point (ft, ft, ft)
P
r,w
Wing segment root point (ft, ft, ft)
P
Shaft
Shaft horsepower, HP
P
t,w
Wing segment tip point (ft, ft, ft)
P
w
Wing root 3D point (ft, ft, ft)
P/W Uninstalled sea level horsepower to design gross weight, HP/lb
q Dynamic pressure, lb/ft
2
Q
cool
Cooled power, Watts
q
max
Maximum dynamic pressure, lb/ft
2
Q
seg
Segment interference drag factor
q
Shear
Shear flow
r Radial distance, L
Distance in 3D parameter space for airfoil interpolation
R Propeller tip radius, ft
Range, nmi
R
e
/L Reynolds number per unit length, ft
1
ROC Rate of climb, ft/min
ROC
min
Minimum rate of climb, ft/min
R
p
Propeller radius, ft
r/R
p
(j) Propeller radiusnormalized radial distance
R
seg
Segment range, nmi
R
subseg
Subsegment range, nmi
R
Year
Average technology ratio per year between two reference dates
s Semispan, L
S
c
or S
Cont
Control surface area, ft
2
SF Wing shape factor
Scale factor
sm Static margin
S
ref
Reference area (defined as planform area of wing 1), ft
2
s
seg
Wing panel span increment, ft
s
tip
Wing tip semispan distance, ft
S
VTail
Vtail area, ft
2
S
w
Wing area, ft
2
S
wet,Fuse
Fuselage wetted area, ft
2
T Thrust, lbs
T
air
Air temperature
xxi
t
Bend
Bending material thickness, ft
t/c Thickness to chord ratio
T
comp
Component temperature
Tech Technology parameter
Tech
1
Technology parameter at data point 1
Tech
2
Technology parameter at data point 2
Temp Absolute static temperature at altitude
Temp
SL
Absolute static temperature at sea level
throt Throttle setting
T
Jet
Jet static thrust rating, lbs
T
max
Maximum installed thrust in flight, lbs
T
min
Minimum installed thrust in flight, lbs
T
max,SLS
Maximum installed thrust at sea level static conditions, lbs
T
Pan
Panel thickness, ft
T
req
Required thrust, lbs
TSFC Installed inflight thrust specific fuel consumption, lbm/lbhr
TSFC
rat
Installed thrust specific fuel consumption ratio
TSFC
ref
Ref. sea level static thrust specific fuel consumption, lbm/lbhr
TSFC
SLS
Installed sea level static thrust specific fuel consumption, lbm/lbhr
t
Shear
Skin thickness of shear material, ft
TVC Tail volume coefficient
T/W Uninstalled SLS thrust to design gross weight ratio
u
m
Local velocity in the streamwise (X) direction, kts or ft/sec
U
∞
Total flight velocity at infinity, kts or ft/sec
V Freestream velocity, kts or ft/sec
Flight velocity, kts
v' Vortex displacement velocity, kts or ft/sec
V
eq,Max
Maximum equivalent velocity, kts
V
Install
Installation volume, ft
3
v
m
Local velocity in the spanwise (Y) direction, kts or ft/sec
Vol
Avail
Volume available, ft
3
Vol
Dish
Radome volume occupied by dish antenna, ft
3
Vol
Eng
Engine volume, ft
3
Vol
Internal
Internal fuel tank volume, ft
3
Vol
Jet
Volume occupied by jet engine, ft
3
Vol
Margin
Volume margin, ft
3
Vol
Tank
Total fuel volume required, ft
3
Vol
Tot
Total fuel tank volume, ft
3
V
Uninstalled
Uninstalled volume, ft
3
W Local total velocity, kts or ft/sec
Weight, lbs
W
ADS
Air data system weight, lbs
W
AntiIce
Antiice system weight, lbs
W
Attach
Wing attachment weight, lbs
W
ATC
Air traffic control system weight, lbs
W
Avion
Total avionics weight, lbs
xxii
W
Bend
Bending material thickness, lbs
W
Box
Weight of wing box, lbs
W
Camera
Camera weight, lbs
W
Comms
Communications system weight, lbs
W
CS,Secondary
Weight of control surfaces and secondary structure, lbs
W
DG
Design gross weight, lbs
W
Dish
Parabolic dish weight, lbs
W
Door
Width of landing gear door, ft
W
ECS
Environmental Control System (ECS) weight, lbs
W
Elec
Electrical system weight, lbs
W
EO/IR
EO/IR payload weight, lbs
W
f
Final weight, lbs
W
fuel
Total fuel weight, lbs
W
fuelb
Fuel burned, lbs
W
Fuels
Fuel system weight, lbs
W
Fuel,wing
Weight of fuel in the wing, lbs
w
fuse
(i) Local fuselage width, ft
W
Fuse
Fuselage weight, lbs
W
Gear
Landing gear weight, lbs
W
GPS
Global Positioning System (GPS) weight, lbs
W
i
Initial weight, lbs
Wid
Eng
Engine width, ft
W
INS
Inertial Navigation System (INS) weight, lbs
W
Jet
Jet engine weight, lbs
w
m
Local velocity in the vertical (Z) direction, kts or ft/sec
W
Nac
Nacelle weight, lbs
W
Overall
Overall vehicle width (usually span), ft
W
Processor
Processor weight, lbs
W
Recip
Reciprocating engine weight, lbs
W
Recorder
Data recorder weight, lbs
W
Ribs
Weight of wing ribs, lbs
W
SAR
SAR payload weight, lbs
W
seg
Mean aircraft weight during segment, lbs
W
Tails
Overall tail weight, lbs
Wt
Install
Installation weight, lbs
W
tire
Width of tire, ft
Wt
Uninstalled
Uninstalled weight, lbs
W
Wing
Wing weight, lbs
x Nondimensional distance Ω*r/V
x(i) Intermediate coefficient in Prandtl LLT
X
AC,wing
Wing aerodynamic center location Xcoordinate, ft
x/c
flap
(j) Segment flap chord ratio
x/c
Λ,w
(j) Segment reference x/c for wing sweep
x/c
ε,w
(j) Segment reference x/c for washout
x
cg
X coordinate of the center of gravity, ft
x
c/4
(i) Local X coordinate of the panel quarter chord, ft
xxiii
x/c
Λ,p
(j) Propeller x/c reference for sweep
x/c
β,p
(j) Propeller x/c reference for twist
X
f,
Y
f,
Z
f
Fuselage leading edge x, y, z coordinates, ft
X/L
AC,wing
Fuselage x/L location of wing aerodynamic center
x/L
f
(j) Fuselage lengthnormalized xdistance from reference
X
NP
Neutral point Xcoordinate, ft
X
p,
Y
p,
Z
p
Propeller center x, y, z coordinates location, ft
X
rot,
Y
rot,
Z
rot
Wing rotation point x, y, z coordinates, ft
X
rot,f,
Y
rot,f,
Z
rot,f
Fuselage rotation point x, y, z coordinates, ft
X
rot,p,
Y
rot,p,
Z
rot,p
Propeller rotation point x, y, z coordinates, ft
X
w
, Y
w
, Z
w
Wing root leading edge x, y, z coordinates, ft
y Y coordinate, ft
y/b
flap
(j) Segment flap span ratio
Year Year
Year
1
Year at data point 1
Year
2
Year at data point 2
Year
Ref
Reference Year
y/L
f
(j) Fuselage lengthnormalized side ydistance from reference
z/L
m,f
(j) Fuselage lengthnormalized mid zdistance from reference
z/L
t,f
(j) Fuselage lengthnormalized top zdistance from reference
z/L
b,f
(j) Fuselage lengthnormalized bottom zdistance from reference
Z
wing
Zcoordinate of wing aerodynamic center, ft
Z
tail
Zcoordinate of tail aerodynamic center, ft
α Vehicle angle of attack, degrees or radians
α
twist
(i) Local washout, degrees or radians
α
0L
(i) Local zero lift angle of attack, degrees or radians
β Blade twist angle, degrees
Γ
VTail
VTail dihedral angle, degrees
Γ
w
(j) Segment dihedral, degrees
δ, δ(i) Angle of the VLM CP at zero angle of attack, degrees or radians
δ
f
(i) Flap deflection, degrees or radians
δ
flap,w
(j) Flap maximum deflection, degrees
∆C
l,flap
Change in local 2D lift coefficient due to flap deflections
∆X Difference in X direction, ft
∆y(i) Local spanwise panel difference, ft
ε Drag to lift ratio
Downwash angle, degrees or radians
ε(i) Local downwash angle, degrees or radians
ε
w
(j) Segment washout relative to wing root, degrees
ζ Displacement velocity ratio, v'/V
η
f
Flap effectiveness factor based on flap type
η
gear
Gear efficiency
η
P
Propeller efficiency
η
P,ideal
Ideal propeller efficiency
xxiv
η
P,nonideal
Nonideal propeller efficiency
η
Rad
Radiation Efficiency
η
radiation
Efficiency of radiating power
η
tip,w
(j) Spanwise location of segment tip  y
tip
/(b
w
/2)
θ(i) Normalized semispan angle for Prandtl LLT
θ
incline
Inclination angle between wing and fuselage side at intersection,
degrees or radians
θ
rot,f
Fuselage rotation angle, degrees
θ
rot,p
Propeller rotation angle, ft
θ
rot,w
Wing rotation angle, degrees
κ
a
Transonic airfoil technology factor
λ Speed ratio
Taper ratio
λ
eq
Equivalent taper ratio
Λ
c/2
Sweep at 50% chord, degrees or radians
Λ
c/4
Sweep at quarter chord, degrees or radians
Λ
p
(j) Propeller sweep angle, degrees
Λ
x,w
(j) Segment sweep at x chordwise location, degrees
u Viscosity
ξ Nondimensional propeller radius
ξ
0
Nondimensional propeller hub radius
π 3.14159…
ρ Air density, sl/ft
3
ρ
b
(j) Bottom conic parameter
ρ
h
Density at altitude, sl/ft
3
or kg/m
3
ρ
0
Density at sea level, sl/ft
3
or kg/m
3
ρ
t
(j) Top conic parameter
σ Local solidity
Air density ratio
τ(i) Local flap effectiveness factor based on flap type
τ
ult
Ultimate shear stress
φ Flow angle
φ(i) Local dihedral angle, degrees or radians
φ
t
Flow angle at the propeller tip
Ω Propeller angular velocity, radians/sec
Common Subscripts
0 Zero lift condition
Sea level
avg Average
eq Equivalent
fact Factor  multiplication
xxv
frict Friction drag
i Index
max,neg Maximum negative angle of attack (most negative)
mindat Point of minimum available data
mindrag Minimum drag condition
press Pressure drag
prof Profile drag
ref Reference case
seg Segment
stall Section, wing, or vehicle stall
table Table
target Target case
type Type
Chapter 1 Introduction
1.1 Background
Unmanned Aerial Vehicle (UAV) systems have taken many names and forms over their
long history. Unmanned aircraft have been dubbed drone, Remotely Piloted Vehicle
(RPV), UAV, Uninhabited Combat Aerial Vehicle (UCAV), Organic Aerial Vehicle
(OAV), Uninhabited Combat Aircraft System (UCAS), and Micro Aerial Vehicle
(MAV), to list a few. Whatever called, unmanned aircraft systems have an airborne
component that performs at least one mission role without a pilot onboard. Additionally,
unmanned aircraft must be capable of controlled and sustained flight, otherwise the first
cast stone is also the first unmanned aircraft. For consistency, the acronym UAV is
adopted throughout this document, as appropriate. The history of unmanned aircraft is
described by Werrell [1985], Wagner [1982 and 1992], Irwin [2000], Holder [2001],
McDaid [1997], Goebel [2003], and Munson [1988 and 1999], to name only a few
sources.
Cruise missiles and UAVs share a highly common lineage [Werrell 1985], and the
distinctions between these types of systems are minor. UAV systems began through
cruise missile and target drone developments. The first operational experiments with
cruise missiles began in World War I [Werrell 1985, McDaid 1997], but these attempts
met with little success. During World War II, Germany developed and employed
numerous V1 cruise missiles with unprecedented results [Werrell 1985]. A model of the
V1 is shown in Figure 11. Although representing the state of the art for unmanned
aircraft, the V1 is generally considered a cruise missile. Over 5,000 V1s were launched
by the Germans [Werrell 1985]. The United States developed many cruise missiles from
the mid1940’s through the late 1960’s with generally disappointing results [Werrell
1985]. The combined AirLaunched Cruise Missile (ALCM) and Surface Launched
Cruise Missile (SLCM) programs of the 1970’s produced the first practical cruise missile
programs [Werrell 1985].
Figure 11 Model of German V1 Cruise Missile
The first golden era of U.S. UAV systems was made possible by the Teledyne Ryan (now
Northrop Grumman) Firebee/Firefly drones [Wagner 1982 and 1992]. The first version
of what became the Firebee drone, shown in Figure 12, was initially developed in 1950,
2
and variants are still in service today [Wagner 1982 and 1992]. Over 20 distinct variants
of this system have been employed [Wagner 1982 and 1992]. The Firefly system
evolved into the Firebee, which conducted over 3,400 operational reconnaissance
missions over Vietnam and other regions [Wagner 1982 and 1992].
Figure 12 AQM34 Firefly Currently in Operational Service
The U.S. experienced a notable decline in UAV funding and development success during
the 1970’s and 1980’s [Wagner 1982 and 1992, McDaid 1997]. Teledyne Ryan
successfully developed two high altitude drones during the 1970’s, the AQM147
Compass Arrow and the Compass Cope, that were not fully utilized or pursued due to the
changing political climate of the early postVietnam era [Wagner 1982 and 1992]. Scale
models of the Teledyne Ryan developments are shown in Figure 13. The most
significant U.S. UAV program of the 1980’s was the Lockheed Aquila battlefield
surveillance RPV, which cost over $1 billion to develop, saw significant schedule slips,
and was ultimately never fielded due to poor performance [Fahlstrom 1998, Munson
1988, Goebel 1993]. A model of Aquila can be seen on the far right of Figure 14. The
cancellation of the Aquila program represented the nadir of U.S. UAV capabilities.
Figure 13 Models of the Teledyne Ryan Drones (Left to Right: Global Hawk,
Compass Cope, AQM154 Compass Arrow, AQM147H Firebee, AQM34 Firefly)
3
Figure 14 Models of Army Battlefield UAVs (Left to Right: Shadow 200, Outrider,
Hunter, Aquila)
U.S. UAV systems are now in a second golden era. The Defense Airborne
Reconnaissance Office (DARO) and the Defense Advanced Research Projects Agency
(DARPA) started, among other UAV efforts, two Advanced Concept Technology
Demonstrator (ACTD) programs in 1994. The Tier II+ Global Hawk, developed by
Teledyne Ryan, is a HighAltitude LongEndurance (HALE) UAV that performs
reconnaissance missions primarily with ElectroOptical/Infrared (EO/IR) and Synthetic
Aperture Radar (SAR) payloads [Drezner 2002, OSD 2002]. A rendering of Global
Hawk is shown in Figure 15. The MediumAltitude Endurance (MAE) Predator UAV,
developed by General Atomics, flies at a lower altitude and has reduced payload capacity
than Global Hawk, but is more affordable [OSD 2002]. A rendering of Predator is shown
in Figure 16. Both systems are equipped with satellite communications systems to
provide Beyond Line Of Sight (BLOS) connectivity with remote ground stations [OSD
2002]. Predator and Global Hawk went into production and have seen extensive
operational service [OSD 2002]. Today, these two systems are the backbone of the U.S.
UAV arsenal.
Figure 15 Global Hawk
4
Figure 16 Predator
Numerous other UAV programs are either in development or operational today [OSD
2002]. The success of the Global Hawk and Predator demonstrated the maturity of
highly effective UAV systems, which helped lead the way for new UAV system
developments. While some recent UAV developments experienced failures, others have
survived. One example of a surviving UAV system still under development is the tactical
AAI Shadow 200 [Palumbo 2000]. Figure 17 shows a rendering of the Shadow 200
system.
Figure 17 Shadow 200
UAVs fill many missions and roles. The obvious distinction from a manned aircraft is
the lack of an onboard pilot. Pilots have limitations such as the need for food, water,
rest, and the intrinsic value placed on human life. Unmanned aircraft have filled roles for
“dull, dirty, and dangerous” [OSD 2002] missions. Dull typically means long
reconnaissance and surveillance missions that stress human endurance. Dirty constitutes
missions that would endanger human pilots due to exposure to radiation or
biological/chemical agents. Dangerous encompasses missions that would place human
life at risk. UAVs also offer the potential for significant cost savings over their manned
counterparts, though cost savings are not always realized.
Technology has played a key role in UAV success. The first golden era was made
possible by analog autopilot advances, lightweight inertial navigation system accuracy
enhancements, and early small jet engine technology. The second golden era is enabled
by lightweight digital electronics, improved sensor technology, and satellite
communications. This effort seeks to explore the implications of potential future
technology advances.
5
The DOD UAV Roadmaps [OSD 2001, OSD 2002], published by the Office of the
Secretary of Defense, provide comprehensive guides to UAV needs and technologies
projected into the future. These reports integrate existing initiatives and platforms into
the plan and projects what new technologies may mature during the same timeframe.
These roadmaps constitute the most definitive sources of UAV technology projections
available today.
Despite the wealth of technology information brought forward in the UAV roadmaps
[OSD 2001, OSD 2002], these sources do not attempt to directly determine the design
impacts that may result from insertion of the provided projections through aircraft design
and optimization. The descriptive plan focuses on existing systems or currently planned
initiatives. The roadmap does not integrate the technology projections into UAV system
designs.
Other reports [Carmichael 1996, (12) The National Academy of Sciences 2000, Sullivan
1998, Air Force Scientific Advisory Board 1996] describe future requirements,
technology, and implementation possibilities. Unfortunately, useful design information
and technology trends supporting the technology modeling development are usually not
available.
Nearly every UAV system developed underwent an aircraft design effort. However, the
design methods employed are rarely published with sufficient detail to provide a
complete methodology that can be transferred to other designs. The level of engineering
may not be worthy of publication in some cases. Often the design methodologies are
closely guarded as proprietary information in the highly competitive UAV industry.
Detailed methodologies are occasionally provided on conceptual design studies that do
not result in flight hardware.
Boeing provided two noteworthy papers on the design of the Boeing Compass Cope
[Brown 1980] and Condor [Johnstone 1990]. Brown [1980] describes the conceptual and
preliminary design activities, manufacturing, and flight test of the Boeing Compass Cope,
though details on specific design methodologies are limited. Johnstone [1990] describes
technical challenges and solutions obtained for the Condor HALE UAV program, but
specific design methodologies are not highlighted.
Lockheed Missiles and Space provided a series of highly detailed papers on the Solar
High Altitude Powered Platform (HAPP) design studies. Hall [1983] gives extensive
conceptual and preliminary design methodologies for the aerodynamics and propulsion
system. Hall [1984] describes a method for sizing the structures of this platform, though
with some limitations due to proprietary considerations. The Solar HAPP system was
intended to fly extended missions at high altitudes using solar power and secondary
power sources. Many methods presented in the reports have sufficient detail to be
repeatable, though many methods are tailored only to the Solar HAPP application. The
Solar HAPP was never built, so the ultimate validity of the methods is unknown.
6
Gundlach [11999] presents the design approach used for the Virginia Tech Forestry
UAV. The profile drag stripbuild up method and the lifting line theory application
described in this source are applied here. Gundlach [11999] does not describe the design
methodology in sufficient detail to be repeatable. The Forestry UAV flew with
demonstrated performance similar to the predictions.
Raymer [2002] presents a method for optimizing UCAV AVs with genetic algorithms.
The AV analysis methodology is largely based upon legacy methods from Raymer [1992]
and the associated RDS software. Raymer [2002] emphasized exploration of GA utility
for aircraft optimization. Raymer’s research appears to be limited to UCAV cases, which
are not covered here.
Naval Research Laboratories (NRL) has advanced the state of the art in Low Reynolds
Number (LRN) UAV design. NRL publications [Foch 1992] highlight key technical
issues affecting LRN AV design, though no consolidated design synthesis approach is
provided.
Sources describing UAV concepts and systems abound. Few describe the design
methodologies in sufficient depth to permit reuse. Many sources describe individual
design disciplines in isolation, particularly LRN aerodynamics. Existing research is
limited to specific design problems that are narrow in scope. There does not appear to be
any previous research covering an integrated design and optimization methodology for
relatively arbitrary Air Vehicles (AVs).
1.2 Problem Statement
An integrated MDO methodology is developed and applied to subsonic, fixedwing UAV
AV design optimization problems. The methods are calibrated against HALE, MAE, and
tactical UAV classes. Technology trends are applied to the requirements of the three
calibration cases to evaluate technology impacts on AV design.
1.3 Approach
The AV MDO methodology is applied through a selfcontained software application. All
geometry, aerodynamics, propulsion, structures, mass properties, and performance
algorithms, to name a few, are included in the provided methodology. Wherever
practical, multiple methods are provided to trade level of detail, computational speed, and
accuracy.
A Genetic Algorithm (GA) is employed to optimize the vehicles for minimum design
gross weight. The selected GA is the GA170 code, developed by David Carrol of UIUC
[Carrol 1998]. GA optimizers offer a robust optimization capability that is insensitive to
local optima and existence of smooth objective functions, relative to calculusbased
7
optimization methods. However, this robustness comes at some expense of refinement
efficiency.
The flight and payload performance requirements are held constant for all cases. This
permits the figure of merit to be minimum design gross weight. Design gross weight is
frequently used as a cost analogy by aircraft design codes, since there is a strong
correlation between vehicle weight and cost. The presumption is that the evaluated
vehicle design gross weight will decrease as technology advances. History shows that the
minimum possible AV size for a given mission does decrease with time, but the average
weight of AV within a category may actually increase with time due to enhanced
capabilities. In other words, UAV requirements typically increase with advancing
technology. Determining future requirements is beyond the scope of this effort, since this
is largely a function of evolving military doctrine, funding, politics, and technological
revolutions.
Three classes of UAV systems were selected for evaluation: the small tactical, Medium
Altitude and Endurance (MAE), and High Altitude Long Endurance (HALE), to cover a
broad range of vehicles and missions. The small tactical UAV requirement is filled by
the AAI Shadow 200 system. The MAE reference system is the General Atomics
Predator (A). The HALE reference system is the Northrop Grumman (previously
Teledyne Ryan) Global Hawk. The relative size of these systems is shown in Figure 18.
Figure 18 Scale Comparison of Shadow 200, Predator, and Global Hawk (Left to
Right)
Technology modeling is applied to the AV MDO code to demonstrate the capabilities of
the methods employed and to provide insight into technology impacts of AV design. The
8
direct, derived, and assumed technology trends, based on the roadmaps [OSD 2001, OSD
2002] and other sources, are applied to AV designs in this research. The technology
areas explored include aerodynamics, avionics and subsystems, general design, payloads,
propulsion, and structures. All technologies are ranked for substantiation, quality and
potential design impact.
The technology period under consideration spans 31 years, from 19942025. The Global
Hawk and Predator systems began development in 1994, and the Shadow 200 system
represents year 2000 technology. AV designs for 1994, and every five years from 2000
to 2025 are evaluated.
Technology impacts are evaluated by introducing advancing technologies to a reference
technology year. The AV MDO code exploits any advantages gained by the applied
technologies. The technologies are considered in combination for all years, and
individually at 1011 years after the reference technology year.
This document provides details of the AV MDO methodology and describes its
application to existing and future UAV systems. First, an overview of the methodology
and code architecture is presented. Next, the genetic algorithm optimizer and its
integration are described. Then, the technology modeling is detailed. The geometry,
aerodynamics, propulsion, systems (size, weight and power), and performance analysis
methods are presented in the next chapters. The next three chapters describe the
application of the methodology to existing UAVs, optimized UAVs with technologies
similar to the existing UAVs, and optimized UAVs utilizing advanced technologies. The
conclusions chapter summarizes key findings and provides recommendations.
9
Chapter 2 Methodology
2.1 Methodology Overview
The AV design synthesis is performed with the aid of an integrated Multidisciplinary
Design Optimization (MDO) code. The MDO code consists of an AV design code and a
GA optimizer. The AV design code includes geometry, weights, propulsion,
aerodynamics, performance, structures, systems analysis and other elements.
2.2 Methodology Development
The methodology selection process involved a search of available methods. The
strengths and limitations of several methods were evaluated by the author. In several
cases, the most promising applicable methodologies were integrated and evaluated.
Original methods were developed where none existed or existing methods had
unacceptable limitations for this application. The final methods are integrated into the
software.
The AV MDO methodology is applied in the JDES software. JDES is written in the
Fortran 90 programming language. This software contains approximately 30,000
physical lines, including spaces and internal documentation. The roughly estimated
Source Line of Code (SLOC) count is approximately 8,000. For reference, a typical
software development productivity metric is 2,000 SLOCs per equivalent head per year.
JDES, in its current form, took approximately 4 years to develop, including algorithm re
use.
A spiral development approach was taken to generate the JDES software. Important
aspects of the code were initially implemented with temporary prototype subroutines.
Once the prototype code proved functional, it was gradually replaced by improved
versions. The first refined method was the geometry definition, since the outputs are
widely used by other methods. The refined aerodynamics followed, since the
aerodynamics methods are highly complex and innovative approaches are attempted.
Other methods for the other analysis disciplines gradually replaced the prototype place
holder subroutines. Throughout development, full or partial functionality code was run to
verify modification effectiveness.
The optimizer proved to be a highly useful tool for identification and isolation of
development errors. Throughout development, running the thencurrent baseline code
helped minimize gross errors through developer inspection of results. As the code
10
became increasingly complex and mature, subtle interactions among subroutines became
increasingly difficult to identify. Once the optimizer was integrated, a new mode of error
management became possible. The optimizer tends to exploit weaknesses in the code.
Through the development and result generation process, the primary aircraft evaluation
subroutine EVAL, was called by the optimizer approximately 1.5x10
6
times.
2.3 Applied Methodology Architecture
The main subroutine in JDES is called ACMAIN. The first subroutine called from
ACMAIN is FILELIST, which contains a list of all the evaluation cases. A loop is entered
for all the cases considered. The first step within the loop is to call FILEMANAGER,
which calls all input file reader subroutines. TEMP represents a number of subroutines
that assigns, stores, and resets temporary values to data that may change during code
execution. Next, GET_OPTFILE is called to read the optimizer settings. If optimization
is required, then the OPTIMIZER master subroutine is called to execute the optimization.
Otherwise, the EVAL design analysis subroutine is called directly with the input design
variable settings. After the aircraft is evaluated or optimized, graphics are generated for
the final vehicle by the DXF subroutine. The functional flow is shown in Figure 21.
Figure 21 ACMAIN Code Architectre
The toplevel functional flow of EVAL, the primary AV design evaluation subroutine, is
shown in Figure 22. First, much of the vehicle geometry is calculated in GEOMETRY.
Next, WEIGHTS is called to evaluate the vehicle mass properties and other systems
evaluations. Then AEROC calculates all of the vehicle aerodynamic tables. Propulsion
generates a tabular engine deck. Finally, PERFORM is invoked to perform the AV
mission performance evaluations and selfoptimizations. The methods contained within
these subroutines are described in later sections.
11
Figure 22 EVAL Code Architecture
2.4 Methods
Multiple approaches are available for several methods. Many options have origins from
prototype methods. The simplified prototype methods were largely supplanted by the
more detailed options, but these remain to enhance analysis flexibility. The detailed
options were generated when the simple methods did not satisfactorily represent the
phenomena of interest. Unfortunately, the detailed methods are typically more
computationally demanding. By keeping both simple and detailed methods, the user can
trade between detail and computational efficiency. For several disciplines, multiple
methods are available with similar levels of detail. In this case, final method selection for
this research is based upon inspection of results and assessed appropriateness of
functional dependencies.
The aircraft design code represents a large body of original development. Some
algorithms were developed by the Author as early as 1993. The graphics file generator,
piecewise linear beam wing bending material weight estimator, and the ideal lift
distribution method algorithms are strongly inspired by the VPI&SU StrutBraced Wing
(SBW) transonic transport research publications. Other influences and applied methods
are cited throughout the methodology chapter.
The methodology chapter describes the methods utilized. Special emphasis is placed on
the methods that are applied to generate the results for this research effort. Latent code
capabilities are covered, though not extensively. Occasionally, unused procedures are
given coverage to highlight limitations to these approaches.
12
2.5 Calibration
The integrated code is calibrated against the Global Hawk, Predator, and Shadow 200. In
all cases, the empty weight, takeoff gross weight, and mission performance are matched
to the target values available from various sources. Other data is obtained for calibration,
such as component lists, subsystem descriptions, payload types, flight envelope
limitations, loading, and structural material information, to name a few. Various design
levels, technology factors, and penalty factors are applied as appropriate to facilitate the
calibration process.
Individual methodologies are validated whenever practical. Applicable experimental data
and test case references are assembled to evaluate the method validity. Some
methodologies are new or incorporate modifications that make calibration difficult, since
experimental data or method test cases are unavailable. For these new or modified cases,
calibration often occurs through comparison with alternative competing methods that are
included in the code.
13
Chapter 3 Optimization
3.1 Optimization Overview
The aircraft design code, JDES, operates under both analysis and optimization modes.
The analysis mode uses inputs provided in the input files with no optimized variable
changes except for automatic internal convergence loops. The optimization mode uses
the data from the input files and userdefined design variables.
A Genetic Algorithm (GA) is employed to optimize the AV designs. GAs are highly
capable of finding global optima within challenging design spaces. However, this comes
at some expense of efficiency in finding the refined optima. To help contend with this
limitation, custom refinement algorithms are optionally available.
A GA is used to flexibly handle a wide range of design problems, where calculusbased
methods may have limitations. The utility and limitations of GAs can be elucidated best
by comparison with the competing calculus methods. Calculusbased methods rely on
derivatives of the objective function with respect to the design variables, known as
gradients. Calculus methods require the existence of derivatives and smooth design
space. The optimal solution is the local optima, not necessarily the global optima.
Calculus methods start from a point in the design space, and the end result is typically
highly dependent on the starting conditions. Within the vicinity of local optima, calculus
methods can be very efficient at finding the refined local optima.
Genetic algorithms handle the design problem in a significantly different manner. GAs
do not require the existence of gradients. A nonsmooth design space with discontinuities
is acceptable. GAs are adept at finding the global optima, since the search is based on a
population of design points, not a single point. However, refinement of the global optima
is inefficient relative to calculus methods within the vicinity of the optima.
GAs do not operate on design variables, called parameters, directly. The parameters are
coded in binary strings of zeros and ones, and combined in larger strings. The GA
operates on the strings, and then decodes the strings into the parameters prior to
evaluating the objective function, called the fitness function. The string operators are
reproduction, crossover, and mutation. Each string represents a member of the
population. There are multiple individuals in the population for each generation. The
design variables are selected to be powerful and intuitive in driving the optimizer to the
global optima.
GAs do not handle constraints directly. Violated constraints must be converted to penalty
functions that are applied to the objective function. By comparison, many calculus
methods may handle constraints directly.
14
David Carrol’s GA170 [Carroll 1998] optimizer was selected for this research. This
code, developed at University of Illinois at Urbana Champaign (UIUC), is freeware and
written in Fortran 77, both key attributes contributing to its selection. The code was
originally written in Fortran 77, but the language was modified to Fortran 90 by the
Author for this application. The GA170 code is a true GA, using the methods described
above.
The basic principles of GAs are wellestablished. A more extensive treatment of GA
theory similar to that applied here can be found in Goldberg [1989].
3.2 Optimization Code Architecture
David Carrol’s GA170 code required significant modification and additional code to be
useful for this application. The standard GA170 code acts as a main subroutine, but the
UAV code has a main subroutine that controls the high level execution of the code.
GA170 was modified to act as a subservient subroutine called by ACMAIN via the
OPTMIMIZER subroutine. New code was written for conditioning the design variables,
constraints, fitness function, and optimizer search refinement.
Figure 31 Genetic Algorithm Code Architecture
Figure 31 shows the GA code architecture. The shaded blocks represent new code
developed by the Author, and the white blocks represent existing or slightly modified
code contained within David Carrol’s GA170. The functions represented by the blocks
are not necessarily contained within a single subroutine or as a distinct subroutine.
ACMAIN
EVAL
OPTIMIZER
GA Inputs
DV, Constr
Inputs
GA170
Record
Fitness
Objective
EVAL
Set DV
FUNC
EVAL
Code DV
GA String
Operators
Decode DV
Reset DV
Bounds
Penalty
15
The ACMAIN code, the main subroutine, reads the generic GA setting data and controls
the execution of the GA. The generic GA setting data includes population size, number
of generations, and probability of mutation and crossover settings. The ACMAIN code
will simply call EVAL, the primary aircraft evaluation subroutine, when in the analysis
mode. When in the optimization mode, ACMAIN will call the OPTIMIZER subroutine.
OPTIMIZER is the primary optimization control subroutine. It reads the specialized GA
inputs, such as design variable limits and constraint information. Once the GA data is
input, OPTIMIZER controls the execution of GA170. Optimizer can call GA170 multiple
times with refined design variable limits such that the global optima may be refined.
Once the best solution is found, OPTIMIZER calls EVAL to evaluate the best design and
write detailed design information to output files.
GA170 contains the GA algorithm. The GA creates an initial population of individuals
based on the design variable limitations. These individuals are represented by genotypes
of bit strings. GA170 performs the GA operations on these strings, decodes the strings
into design variables, and evaluates the fitness of the individuals. The fitness value for
each individual is found in subroutine FUNC. This procedure occurs for the specified
number of generations. The fittest individual in the final generation is output.
Subroutine FUNC combines the objective function and the constraint penalty functions to
produce the fitness value for GA170 to utilize. FUNC calls EVAL with the design
variable settings. EVAL calculates the objective function, and evaluates the individual
constraint values and penalties. The constraint penalties are summed by FUNC and
combined with the objective function to produce the fitness function. FUNC calls a
subroutine to record the most fit members of populations over all generations that do not
have violated constraints. The optimized solution and the fit individual array are sent
back to Optimizer after the final generation is run.
OPTIMIZER has the option of refining the permissible solution set within the solution
space based on the knowledge of the fit individual array. In this scenario, the minimum
and maximum design variable limits are brought closer to the minimum and maximum
design variable values for the fittest individuals with no violated constraints. This
procedure may be performed many times until either the maximum number of iterations
is reached or the difference between the minimum and maximum fitness values reaches a
specified tolerance. This new feature is not applied to the final cases, since the baseline
GA algorithms are sufficient.
The GA170 code is a highly useful GA. Substantial new code was required to integrate
GA170 into a flexible and userfriendly UAV optimizer. The final product fully satisfies
the needs of the research problem set.
Although randomization is used by the GA, the results are repeatable. The random
number generator selects the same random numbers starting at the initiation of software
execution. The optimized solutions are identical when evaluated in the same order.
16
3.3 Design Variables
The universally applicable design variables are shown in Table 31.
Variable Description
b
W
Wing span
AR
W
Wing aspect ratio
λ Wing taper ratio
t/c Wing thickness to chord ratio
MF
Fuel
Fuel mass fraction
Table 31 Universal Design Variables
The Global Hawk case additionally provides wing quarter chord sweep as a design
variable. Sweep is a unique design variable to this case, because higher Machnumber
Global Hawk contends with compressibility effects. Sweep is one way to counteract
compressibility impacts on drag.
The geometry is a strong contributor to vehicle weight. The optimizer seeks to minimize
the vehicle design gross weight while not violating constraints. The geometry design
variables applied here were selected to have the greatest impact. Some of the trades
performed by the optimizer with the design variable selections are described here.
The aspect ratio drives aerodynamic efficiency at the expense of structural weight. High
aspect ratio wings have lower induced drag and higher available total lift coefficient for a
given wing area than low aspect ratio wings. Exclusive of high lift devices and for a
fixed airfoil, higher aspect ratio wings tend to have lower wing areas due to the increase
in maximum lift coefficient. High aspect ratio wings have a higher structural weight than
low aspect ratio wings of the same area due to higher bending moments and shear loads
in the wing spar. Aerolastic phenomena are more dominant in high aspect ratio wings,
though these effects are not evaluated here.
The taper ratio affects the induced drag and structural efficiency. Wing taper permits a
wing to approach the ideal elliptical lift distribution more closely than without taper,
which reduces the induced drag. At low Reynolds numbers, the wing tip Reynolds
number is lower than the average chord Reynolds number, which can drive up the profile
drag. With a constant wing area and wingspan, a reduced taper ratio increases the root
chord length, which increases the wingfuselage fairing size in the parametric geometry.
The increased fairing size impact on drag can have positive or negative drag
contributions. Taper improves structural weight, because the absolute airfoil thickness is
higher where the bending moment is highest. The increased area moment of inertia
permits the wing skin thickness to be reduced for a given loading.
Wing sweep can delay the onset of compressibility and reduce the impacts of wave drag,
but at the cost of a reduction in laminar flow and less structural efficiency. Wing sweep
trades are documented in Gundlach [21999].
17
Increasing the airfoil thickness to chord ratio increases structural efficiency, but tends to
adversely affect the wing profile drag and hastens the onset of compressibility effects.
Transonic wings tend to manipulate wing sweep and airfoil thickness to chord ratio
simultaneously.
The wing, tail, and fuselage weights are impacted by scale. The optimizer tends to
minimize fuselage length and wing span such that the volume constraints are not violated.
The parametric tail geometry sizes the tail surfaces as a function of tail moment arm, so
the fuselage length reduction is tempered by tail size. Additionally, the parametric
geometry used for all optimization cases limits the minimum fuselage length via
constraints that are functions of span and average chord, as described in the geometry
section.
The propulsion design variable for jet aircraft are shown below in Table 32. Note that
the thrust to weight ratio, T/W, and SeaLevel Static (SLS) thrust, T
max,SLS ,
may not be
used simultaneously. The thrust is evaluated at the uninstalled SLS maximum thrust
conditions. Uninstalled refers to the advertised engine characteristics that do not take
into account for engine installation impacts due to inlets or nozzles. The weight in T/W
is the takeoff gross weight of the aircraft.
Variable Description
T/W Uninstalled SLS thrusttoweight ratio at
TOGW
T
max,SLS
Uninstalled SLS thrust
Table 32 Jet Propulsion Design Variables
The propulsion design variables for reciprocating engines are shown in Table 33. Note
that P/W and P
Shaft
may not be used simultaneously. The power is the uninstalled sea
level static brake power. Uninstalled refers to the advertised characteristics of the engine
that does not account for aircraft installation. The weight in P/W is the takeoff gross
weight of the aircraft.
Variable Description
P/W Uninstalled brake powertoweight ratio at
design gross weight
P
Shaft
Uninstalled brake horsepower
Table 33 Reciprocating Engine Propulsion Design Variables
3.4 Constraints
Several constraints are used by the optimization code. Unique penalty functions are
calculated for violated constraints. The constraint types are listed below in Table 34.
The user can select which constraints are imposed prior to running the code.
18
Constraint Description
Mission Complete Constraint violated if prescribed mission is not
completed due to performance limitations or lack
of fuel
Mission Constraints Constraints violated if the set of mission
constraints (Table 35) is not fully satisfied over
the complete prescribed mission
Fuel Volume Constraint violated if the required fuel volume is
greater than the available allocated volume
Table 34 Constraints
The mission constraint types are listed in Table 35.
Constraint Number Description
1 Stall (C
L
> C
L,Max
)
2 Insufficient thrust for minimum specified rate of climb
3 Insufficient thrust for level flight
4 Dynamic pressure limit (q > q
Max
)
5 Mach limit (M > M
Max
)
6 Engine operating limits (type dependent)
7 Altitude limit
8 Weight error (Weights < 0)
9 Lift lower limit (C
L
< C
L,Min,Table
)
Table 35 Mission constraint types
The fitness function is a combination of the objective function and the penalty functions.
The GA attempts to maximize the fitness. If the objective is to minimize the weight, and
the weight is the objective function, then the fitness should be the negative of the
objective function. A negative fitness with a lower magnitude is considered a greater
number, so lowering the negative magnitude is a means of maximizing the fitness.
Similarly, if the violated constraint penalties are represented in terms of weight, then the
penalty contribution to the fitness function is the negative of the penalty. The fitness
function is:
( )
∑
+ − = Penalty Objective Fitness
19
The objective function is made to be on the order of 10
0
(or unity). When the objective
function is purely a function of design gross weight (W
DG
), then the objective function is:
Ref
DG,
DG
W
W
Objective =
The penalty functions are intended to be referenced to the same order of magnitude. For
constraints that vary, the penalty function takes the form of a quadratic equation with a
step function starting when the constraint is violated. The step function ensures that there
is an immediate penalty for a violated constraint. The quadratic equation increases the
penalty more aggressively as the magnitude of constraint violation increases. Each
penalty has a userdefined weighting parameter. The variable penalty function is:
) (Constraint f Param = When constraint is violated
( )
Wt
Penalty Param F Param F Step Penalty ⋅ ⋅ + ⋅ + =
2
2 1 When constraint is violated
Other constraints take the form of discrete differences. In other words, the constraint is
either satisfied or not satisfied. There is no variation in the degree of constraint violation.
The penalty parameter is:
1 = Param
when a constraint is violated. Discrete constraint penalty functions take the following
form:
Wt
Penalty Param F Penalty ⋅ ⋅ = 1
when a constraint is violated.
The penalty functions for each of the constraints are described below.
Mission Completion
The mission completion penalty function is called when the mission completion is less
than unity.
1 − = MissComp Param
The penalty function is:
( )
Wt
Penalty Param Param Penalty ⋅ + + =
2
1 . 0
20
when the constraint is violated
Performance Constraints
The performance constraint penalty function is found by:
Violated
N Param =
The penalty function is:
Wt
Penalty Param Penalty ⋅ =
Fuel Volume Constraints
The wing volume penalty function is found by:
Type Avail,
Type Margin,
1
Vol
Vol
Param + =
when the margin for a given type of volume is negative. The type refers to wings and
fuselage, and each type is evaluated individually.
( )
Wt
Penalty Param Param Penalty ⋅ + + =
2
1 . 0
when the constraint is violated.
21
Chapter 4 Technology Modeling
4.1 Technology Overview
Technology modeling into the future is an art, since technology trends that endure for
more than a few short years are rare. One of the most famous exceptions, Moore’s Law,
states that the number of transistors on a microprocessor doubles approximately every 18
months. This Law has remained notably accurate since 1965, and the form of this
projection is expected to hold true into the coming two decades [OSD 2002].
Most technology trends tend to have high levels of uncertainty. Many trends are based
upon the possible, not the likely. It is possible to accelerate a technology with aggressive
funding, priority, or inspiration. However, technologies may also experience a
maturation period prior to garnering sufficient attention to stimulate investment.
Predicting enabling funding relationships can be as challenging as predicting technology
breakthroughs. Nascent technologies may demonstrate impressive performance under
laboratory conditions. However, this does not mean that the embryonic technology is
suitable for operational applications. Transforming the laboratory technology into a
fieldready product may result in performance alteration due to practical integration
considerations.
To counter the time lag between demonstrated technology and operational capability, it is
important to capture impacts of technology emphasis and level of design in the
predictions. This research attempts to accommodate both considerations. Long term
technology projections contain high uncertainty. Expectations of accuracy become
increasingly unrealistic as the time into the future extends. In light of the inherent
limitations associated with divining the future, this research endeavors to describe what
may potentially happen, not what will.
The technologies are grouped into aerodynamics, avionics/subsystems, design, payloads,
propulsion, and structures categories. The individual technologies composing the
technology categories were qualitatively ranked for substantiation, quality, and
aggressiveness. The summation of all technology categories is shown in Table 41.
Details of each technology are described throughout this section.
22
Table 41 Technology Attributes
The ranking standards for each attribute are shown in Table 42.
Table 42 Attribute Ranking Description
23
As can be seen from the above tables, there is only one directly applicable existing
technology trend with a high relative chance of success and anticipated high impact on
the optimization objective function.
4.2 Technology Methods
Two primary methods of technology modeling are applied to the design code. These are
technology parameter ratios (Method 1) and direct technology parameter lookup (Method
2).
The technology parameter ratio, Method 1, is used whenever the ratio of technologies is
multiplied or divided by a baseline value. An example is a technology factor applied to
an input baseline avionics weight. The procedure is shown graphically in Figure 41 (A
B).
(A) (B)
Figure 41 Technology Parameter Ratio Model (Method 1)
A single technology curve composed of discreet technology versus technology year data
points is used. The technology phase lag, Phase, is the time difference between
availability of the technology and its application. Phase is found through a lookup
procedure based on phase lag versus technology level input data. The technology curve
is shifted to the future by the time amount Phase, to obtain an adjusted technology curve.
The technology for a given lookup year is:
( )
1
1
) (
Year Year
Year
R Tech Year Tech
−
⋅ =
where Year is automatically bound by Year
1
and Year
2
in the adjusted technology curve.
In the case where Year lies beyond the final Year
2
date, the following extrapolation
occurs:
24
( )
2
2
) (
Year Phase Year
Year
R Tech Year Tech
− −
⋅ =
Whenever Year is before the earliest Year
1
date, the technology is set to the earliest
technology level without further extrapolation.
The technology ratio between Year
1
and Year
2
, R
Tech
, is found by:
1 2
1
1
2
) (
) (
Year Year
Tech
Year Tech
Year Tech
R
−


.

\

=
The technology factor, F
Tech
, is found by:
) (
) (
Ref
Des
Tech
Year Tech
Year Tech
F =
where Year
Des
is the design year and Year
Ref
is the reference year of the baseline. For
example, the Global Hawk program began in 1994, so the reference year is 1994. If the
Global Hawk requirements were evaluated in 2005, then the design year would be 2005.
The baseline characteristic of interest is multiplied or divided by the technology factor, as
appropriate.
The direct technology parameter lookup model, Method 2, is suitable for design
technologies that are not ratios. For example, the structural design factor of safety as a
function of design year and design level is appropriate for a lookup function. The design
level ranges from 01, where 0 represents the lowest design level and 1 is the highest
design level. The equivalent of a phase lag is built into the design level, because design
level of 1 equates to the minimum phase lag, and design level 0 represents the equivalent
of the maximum phase lag. This lookup procedure is shown graphically in Figure 42.
25
Figure 42 Direct Technology Parameter Lookup Model
4.3 Technology Modeling
4.3.1 Aerodynamics
The aerodynamics technologies are divided into induced drag, interference drag, wave
drag, and profile drag. For convenience, the attributes of these technologies are repeated
in Table 43.
S
u
b
s
t
a
n
t
i
a
t
i
o
n
Q
u
a
l
i
t
y
A
g
g
r
e
s
s
i
v
e
n
e
s
s
Aerodynamics
Induced Drag 4 3 2
Interference Drag 3 2 2
Wave Drag 3 2 2
Profile Drag 3 2 3
Table 43 Aerodynamics Technology Attributes
Induced Drag
The induced drag technology is modeled as a factor applied to the minimum theoretical
induced drag found through the Trefftz plane analysis. The trends are assumed to
converge to the ideal as time progresses. This assumption is based on the advancement
26
of Computational Fluid Dynamics (CFD) in wing design optimization. The assumed
technology trends are shown in Figure 43.
Induced Drag Multiplication Factor vs.
Design Year
0.95
1
1.05
1.1
1.15
1.2
1920 1940 1960 1980 2000 2020 2040
Design Year
I
n
d
u
c
e
d
D
r
a
g
M
u
l
t
i
p
l
i
c
a
t
i
o
n
F
a
c
t
o
r
Des Lev 0
Des Lev 0.5
Des Lev 1
Figure 43 Induced Drag Multiplication Factor Technology Trends
Interference Drag
Hoerner [1992] interference drag methods are used, as described later in the
aerodynamics section. Hoerner notes that the interference drag of a strut junction may be
reduced to 10% of the data presented if properly faired. This trend is assumed to hold for
all other interference drag contributors from 1940 through 1995. Some improvement in
aggressive design levels are permitted between 1995 and 2025. The general trend is for
the interference drag multiplication factors to approach the minimum value as time
progresses. This assumption is justified by the proliferation of CFD codes and design
methods. Much of Hoerner’s data and methods originated in Germany during World War
II. Therefore, the starting date is 1940. The interference drag multiplication factor is
applied to the standard Hoerner interference drag estimates. The trends are shown in
Figure 44. Due to the significant impacts of the interference drag on vehicle geometry
optimization, the interference drag design level is set to 1 for all cases.
27
Interference Drag Multiplication Factor vs.
Design Year
0
0.2
0.4
0.6
0.8
1
1.2
1920 1940 1960 1980 2000 2020 2040
Design Year
I
n
t
e
r
f
e
r
e
r
e
n
c
e
D
r
a
g
M
u
l
t
i
p
l
i
c
a
t
i
o
n
F
a
c
t
o
r
Des Lev 0
Des Lev 0.5
Des Lev 1
Figure 44 Interference Drag Trends
Wave Drag
The wave drag is modeled through the transonic airfoil technology factor used in the
Korn equation [Grasmeyer 1998]. The factor for a Boeing 747 is 0.89, and for a Boeing
777 with supercritical airfoil sections of 1995 technology year is 0.955 [Grasmeyer
1998]. These commercial transport values are used to estimate the stateoftheart trends.
One noteworthy assumption is that the lowest technology line for 1995 through 2025 is
assumed to be consistent with the technology associated with the 747. The transonic
airfoil technology factor trends are shown in Figure 45.
Wave Drag Technology Factor vs. Design
Year
0.7
0.75
0.8
0.85
0.9
0.95
1
1960 1980 2000 2020 2040
Design Year
T
r
a
n
s
o
n
i
c
A
i
r
f
o
i
l
T
e
c
h
n
o
l
o
g
y
F
a
c
t
o
r
Des Lev 0
Des Lev 0.5
Des Lev 1
Figure 45 Wave Drag Technology Factor Trends
28
Profile Drag
The profile drag technology is modeled as a multiplication factor to the profile drag.
Data was compiled for high performance airfoil sections over a wide time span for three
Reynolds numbers covering the majority of the range of interest here. The referenced
airfoils represent assumed typical stateoftheart for the airfoil design year. Table 44
shows the selected airfoils.
Airfoil Year Reynolds
Number
Airfoil L/D max Source
1949 60,000 NACA 0009 26.3 Selig [1989]
1980 60,000 FX 60100 45.5 Selig [1989]
1985 60,000 HQ 2.59 47 Althaus [1985]
1922 300,000 Clark Y 71 Selig [1989]
1980 300,000 FX 63137 93 Selig [1989]
1949 3,000,000 NACA 2415 95 Abbot [1949]
1992 3,000,000 SM 701 100 Somers [192]
Table 44 Airfoil Characteristics.
The profile drag multiplication factor is the ratio of the inverse of the lift to drag ratios
referenced to the starting year condition for a given Reynolds number. These factors for
the selected airfoils are plotted in Figure 46.
Profile Drag Multiplication Factor vs. Design
Year
0
0.2
0.4
0.6
0.8
1
1.2
1900 1920 1940 1960 1980 2000
Design Year
P
r
o
f
i
l
e
D
r
a
g
M
u
l
i
t
p
l
i
c
a
t
i
o
n
F
a
c
t
o
r
Re=3M
Re=300K
Re=60K
Figure 46 Profile Drag Multiplication Factor Trends
The technology trends derived from the available data are highly Reynolds number
dependent. The 3 million Reynolds number case represents the flight regime experienced
by small, lower speed manned aircraft such as general aviation types. By the first
reference date of 1949, much development and testing had been performed on airfoils
29
operating in these Reynolds numbers, spurred on by two world wars and 44 years of
commercial air transportation.
By comparison, the lower Reynolds number airfoils, especially near 60,000, had
relatively few applications beyond model aircraft. It was not until the late 1970s that the
low Reynolds number regime (Re < 300,000) received serious attention by technical
hobbyists and engineers. Today, there are multiple, highquality, low Reynolds number
databases available, which are frequently used by competitive model sailplane enthusiasts
and UAV developers. Furthermore, high performance boundary layer codes, such as
XFOIL, are freely available. These codes can realistically model and aid in the design of
low Reynolds number airfoils.
A single function that is Reynolds number independent is desired for the profile drag
multiplication factor. The UAVs under consideration typically operate with wing chord
Reynolds numbers near the 300,000 to 3,000,000 range. Therefore, a curve
approximately splitting the two trends was employed. The level of design was also
adopted to bound the variation of design. The profile drag multiplication factor of 1
represents the state of the art in 1949. The presumption is that after the year 1990, the
poorest airfoil designs employed on UAVs will be the equivalent of the stateoftheart in
1949 for an equivalent Reynolds number. The stateoftheart airfoils in 2025 are
assumed to have a 20% improvement over the stateoftheart in 1949 for a given
Reynolds number, as a best judgment. Figure 47 shows the resulting assumed airfoil
technology trends.
Profile Drag Multiplication Factor vs. Design
Year
0.6
0.8
1
1.2
1.4
1940 1960 1980 2000 2020 2040
Design Year
P
r
o
f
i
l
e
D
r
a
g
M
u
l
t
i
p
l
i
c
a
t
i
o
n
F
a
c
t
o
r
Des Lev 0
Des Lev 0.5
Des Lev 1
Figure 47 Final Profile Drag Multiplication Factor Trends.
The ratio of profile drag multiplication factors is applied as a multiplication factor to the
calculated profile drag at a given design level.
30
Laminar Flow
The laminar flow technology affects transition for 2D airfoils and 3D wings with
sweep. The applied technologies are independent of year, and the formulation takes a
different form than other technologies. The laminar flow technology impacts are
described in the Aerodynamics section.
4.3.2 Avionics and Subystems
Technology trends for various types of avionics are evaluated. The attributes of these
technologies are repeated in Table 45.
S
u
b
s
t
a
n
t
i
a
t
i
o
n
Q
u
a
l
i
t
y
A
g
g
r
e
s
s
i
v
e
n
e
s
s
Avionics/Subsystems
Autopilot 3 1 1
INS 3 1 1
Subsystems 2 3 1
Communications 4 3 2
Table 45 Avionics and Subsystems Technology Attributes
Guidance, Navigation and Control
The Guidance, Navigation and Control (GNC) avionics includes the INS, flight
computers, and varying flight critical sensors. Werrell [1985] noted that a 1970 cruise
missile GNC weight is 115 pounds. Data for similar functionality GNC systems from
manufacturer specification sheets was used to generate the overall GNC trends. Figure
48 and Figure 49 show the GNC technology phase and technology trends, respectively.
The GNC trend is applied to all avionics except for the air data system and INS.
31
GNC Technology Phase vs. Technology
Level
0
2
4
6
8
10
0 0.2 0.4 0.6 0.8 1
Technology Level
T
e
c
h
n
o
l
o
g
y
P
h
a
s
e
,
Y
e
a
r
s
Figure 48 GNC System Technology Phase.
GNC Weight vs. Technology Year
0.1
1
10
100
1000
1970 1980 1990 2000 2010
Technology Year
G
N
C
W
e
i
g
h
t
,
P
o
u
n
d
s
Figure 49 GNC System Technology Trends.
Inertial Navigation System
UAV inertial navigation systems (INS) have seen much progress over the past half
century. In 1960, a cruise missile INS weighed approximately 300 pounds [Werrell
1985]. By 1970, the same functionality system weighed only 29 pounds, enabling a
miniaturized cruise missile revolution [Werrell 1985]. The trends carry the same
functionality forward to current technology, using manufacturer specification sheets. The
assumed guidance system technology phase and technology trends are shown in Figure
410 and Figure 411, respectively.
32
INS Technology Phase vs. Technology
Level
0
1
2
3
4
5
0 0.2 0.4 0.6 0.8 1
Technology Level
T
e
c
h
n
o
l
o
g
y
P
h
a
s
e
,
Y
e
a
r
s
Figure 410 INS Technology Phase
INS Weight vs. Technology Year
0.01
0.1
1
10
100
1000
1960 1970 1980 1990 2000 2010
Technology Year
I
N
S
W
e
i
g
h
t
,
P
o
u
n
d
s
Figure 411 INS Technology Trends
Processors
Much of the avionics revolution is made possible through the advancement of processor
technology. The DOD UAV Roadmap [2002] has tracked commercial processor speed
through time. This technology trend was adopted directly in this research, though not
used to generate the results. There is a phase lag between the availability of commercial
processors and the application to aircraft.
Although a processor performance trend is available for technology modeling within the
design code, it is not utilized here. The INS and GNC weight trends are utilized to
estimate avionics weights. The processor speed is not proportional to actual avionics
weight, because demand placed on processors increases as processor capability improves.
33
The demands placed on processors typically increases faster than the improvements in
processor speed [Noor 1997]. The GNC weight trends are more representative of
attainable weight reductions.
Subsystems
Trends found in Noor [2000], which present Defense and Aerospace study [1995] results,
are used for subsystems technology modeling. For longrange, transonic transports, the
anticipated total gross weight reduction due to subsystems contributions is estimated at
9% [Defense and Aerospace 1995]. This reduction, however, is not the subsystems
weight reduction, but the total synthesized aircraft weight reduction. The subsystems
contribution is found by a simple mass fraction analysis. A very simple sizing
relationship for a given aircraft class is:
Tech Subs Subs Subst E Fuel Payload
TO
F MF MF MF W
W
, ,
1
1
⋅ − − −
=
−
Numbers representative of the Boeing 777 are used as an estimate for the 1995 mass
fraction numbers. The fuel mass fraction is assumed to be 0.4. The mass fraction of the
empty weight less subsystems is assumed to be 0. 42. The mass fraction of the 1995
subsystems is assumed to be 0.05, when the subsystems weight multiplication factor is
equal to 1. In order to achieve a 9% reduction in takeoff gross weight due to subsystems
contributions, the subsystems weight multiplication factor must equal 0.75. The 1960
subsystems weight multiplication factor is assumed to be 1.1, or 10% greater than the
1995 subsystems weight, since subsystems technology has improved between these dates.
This technology trend is assumed to be applicable to UAVs in absence of better
subsystems weight technology models. All UAVs considered here are much smaller than
transonic transports, but competing subsystems technology information suitable for a
model is unavailable. The assumed subsystems technology phase and multiplication
factor trends are shown in Figure 412 and Figure 413, respectively.
Subsystems Technology Phase vs. Technology
Level
0
1
2
3
4
5
6
0 0.2 0.4 0.6 0.8 1
Technology Level
T
e
c
h
n
o
l
o
g
y
P
h
a
s
e
,
Y
e
a
r
s
Figure 412 Subsystems Technology Phase
34
Subsystems Weight Multiplication Factor vs.
Design Year
0.6
0.8
1
1.2
1.4
1960 1970 1980 1990 2000 2010 2020
Design Year
S
u
b
s
y
s
t
e
m
s
W
e
i
g
h
t
M
u
l
t
i
p
l
i
c
a
t
i
o
n
F
a
c
t
o
r
Figure 413 Subsystem Weight Technology Trend
Communications
The most significant communications system impact on UAV design is the antenna for
satellite communications, if so equipped. In addition to the weight and power of the
antenna and gimbals, the antenna takes up considerable volume and drives the shaping of
the fuselage. The diameter of dish antennas, as seen on the Global Hawk and Predator, is
driven primarily by the antenna gain requirements. The gain requirements are in turn
driven by the desired data rates, radiated power, and beam width requirements. These are
largely driven by the satellite constellation that the UAV is communicating with.
It is difficult to predict the advancement of communication satellites through 2025. This
is largely driven by satellite project funding. It is even more difficult to translate the
communication satellite capabilities into UAV antenna diameter. The assumption that
there will be no communication satellite or related UAV antenna technology
advancement is bolder than assuming that technology will progress. The DOD UAV
Roadmap [OSD 2002] predicts that data compression, preprocessed payload data, and
more efficient usage of bandwidth will act to relieve the communications challenges
facing UAVs. No direct impacts on aperture sizing are provided. In light of the OSD
predictions, an assumption was made that for equivalent collection functionality, the dish
antenna diameter on UAVs will decrease to 50% of the 1994 values by 2010 and to 25%
of the 1994 value by 2025. This relationship is shown in Figure 414. It is assumed that
no technology phase applies to this assumed trend. This trend affects the Global Hawk
and Predator families only.
35
Comms Antenna Diameter Ratio vs. Technology
Year
0
0.2
0.4
0.6
0.8
1
1.2
1990 1995 2000 2005 2010 2015 2020 2025
Technolodgy Year
C
o
m
m
s
A
n
t
e
n
n
a
D
i
a
m
e
t
e
r
R
a
t
i
o
Figure 414 Satellite Communications Dish Antenna Diameter Trends
4.3.3 Design
The only technology attribute in the design technology category is weight growth. For
convenience, the attributes of this technology are repeated in Table 46.
S
u
b
s
t
a
n
t
i
a
t
i
o
n
Q
u
a
l
i
t
y
A
g
g
r
e
s
s
i
v
e
n
e
s
s
Design
Weight Growth 4 3 2
Table 46 Design Technology Attributes
Weight Growth
The overall vehicle weight growth design technology is assumed to follow the trend
presented in Figure 415. Nearly every aircraft experiences weight growth in the
development process as the aircraft goes from early design to operations. The typical
weight growth from project goahead to first flight is 7% [SAWE 1996]. The weight
growth can be substantially higher than this. The F111 experienced 17% growth from
goahead to the end of the first phase research and development, and 23% growth over
the first five years after goahead [SAWE 1996]. By comparison, the C5A experienced
only 2.9% weight growth from contract award to the delivery date [SAWE 1996]. The
assumed lower and upper bounds from 1930 through 1990 from design year to operations
are 3% and 20%, respectively. By 2025, the assumed weight growth upper and lower
bounds are assumed to change to 1% and 15%, respectively. The technology trend
36
assumes that the combined impacts of improving integrated mass properties tools and
evolving systems engineering practices containing requirements creep will limit the
weight growth encountered during development.
Weight Growth Technology
0%
5%
10%
15%
20%
1930 1950 1970 1990 2010 2030
Design Year
W
e
i
g
h
t
G
r
o
w
t
h
t
o
O
p
e
r
a
t
i
o
n
s
Des Lev 0
Des Lev 0.5
Des Lev 1
Figure 415 Weight Growth Technology
4.3.4 Payloads
Only a small subset of potential payload types is considered for this research.
Technology trends for Synthetic Aperture Radar (SAR), ElectroOptical (EO), Long
Wave Infrared (LWIR), and MidWave Infrared (MWIR) are incorporated. The source
of all payload technology trends is the 2000 DOD UAV Roadmap [OSD 2001].
For convenience, the attributes of these technologies are repeated in Table 47.
S
u
b
s
t
a
n
t
i
a
t
i
o
n
Q
u
a
l
i
t
y
A
g
g
r
e
s
s
i
v
e
n
e
s
s
Payloads
SAR Payload 1 1 1
Small EO/IR 2 2 2
Large EO/IR 2 2 3
Table 47 Payload Technology Attributes
37
All payloads are assumed to follow the payload technology phase relationship shown in
Figure 416. The high technology level phase lag is zero, because it is possible to
develop a payload and a UAV concurrently.
Payload Technology Phase vs. Technology
Level
0
1
2
3
4
5
6
0 0.2 0.4 0.6 0.8 1
Technology Level
T
e
c
h
n
o
l
o
g
y
P
h
a
s
e
,
Y
e
a
r
s
Figure 416 Payload Technology Phase
SAR Payloads
The SAR payload technology trends are shown in Figure 417. The two trends are the
payload relative performance and relative performance to weight. Note that the relative
performance to weight improves much faster than the relative performance. The relative
performance per weight is used for SAR payload scaling.
SAR Relative Technology vs. Technology
Year
1
10
100
1000
10000
1980 1990 2000 2010 2020 2030
Technology Year
R
e
l
a
t
i
v
e
T
e
c
h
n
o
l
o
g
y
M
e
t
r
i
c
Rel Perf/Wt
Rel Perf
Figure 417 SAR Payload Technology Trends
38
ElectroOptical / Infrared (EO/IR) Payloads
(A) (B)
Figure 418 EO/IR Payloads
The DOD UAV Roadmap has performance technology trends for large and small Electro
Optical (EO) payloads and large and small MidWave Infrared (MWIR) payloads. The
relative performance metric is the ratio of ground resolved distances, a measure of
imaging performance, of the initial capability over the new capability. Most EO and
Infrared (IR) payloads are combined into single units with an assumed even relative
weight and functionality division, so the technology trends were averaged. Two
examples of EO/IR balls are shown in Figure 418 (AB). The large EO and large MWIR
payload trends were combined for the large EO/IR trend, and the small EO and small
MWIR payload trend were combined for the small EO/IR trend. In both cases, the
MWIR data was extrapolated to match the first EO technology date. The resulting curves
for large and small EO/IR payloads are shown in Figure 419 and Figure 420,
respectively.
The EO/IR payload weight per performance is considered constant. This assumption
may be conservative if the EO/IR payload exhibits similar behavior to the SAR.
Large EO/IR Technology
1
1.05
1.1
1.15
1.2
1.25
1.3
1.35
1.4
1970 1980 1990 2000 2010 2020 2030
Technology Year
R
e
l
a
t
i
v
e
P
e
r
f
o
r
m
a
n
c
e
Figure 419 Large EO/IR Payload Performance Technology Trend
39
Small EO/IR Technology
1
2
3
4
5
6
7
8
1990 1995 2000 2005 2010 2015 2020 2025
Technology Year
R
e
l
a
t
i
v
e
P
e
r
f
o
r
m
a
n
c
e
Figure 420 Small EO/IR Payload Performance Technology Trend.
4.3.5 Propulsion
The basic propulsion technology trends were derived directly from data provided in the
DOD UAV Roadmap [2002]. Both jet engines and internal combustion engine trends are
included.
The attributes of these technologies are repeated in Table 48.
S
u
b
s
t
a
n
t
i
a
t
i
o
n
Q
u
a
l
i
t
y
A
g
g
r
e
s
s
i
v
e
n
e
s
s
Propulsion
Jet Propulsion 1 2 1
Reciprocating Engines 1 2 1
Table 48 Propulsion Technology Attributes
40
Jet Engines
Figure 421 UAV Jet Engine
The HALE family considered here uses jet engines for propulsion, so appropriate
technology trends must be applied. An example of a jet engine intended for UAV
application is shown in Figure 421. The jet engine technology phase and trends are
presented in Figure 422 and Figure 423, respectively. The jet engine technology phase
is based on the assumed time from engine technology development to fielding in an
aircraft.
Jet Engine Technololgy Phase vs.
Technology Level
0
5
10
15
0 0.2 0.4 0.6 0.8 1
Design Level
T
e
c
h
n
o
l
o
g
y
P
h
a
s
e
,
Y
e
a
r
s
Figure 422 Jet Engine Technology Phase
41
Jet Technology Factors vs. Technology
Year
0
0.5
1
1.5
2
2.5
3
3.5
4
1980 1990 2000 2010 2020 2030
Technology Year
T
e
c
h
n
o
l
o
g
y
F
a
c
t
o
r
s
T/W fact
SFC fact
Figure 423 Jet Engine Technology Trends
Internal Combustion Engines
Figure 424 UAV Internal Combustion Engine
The tactical and MAE UAV classes considered here utilize internal combustion engines.
An example of an internal combustion engine intended for UAV applications is shown in
Figure 424. The internal combustion technology phase and technology trends are
presented in Figure 425 and Figure 426, respectively. The internal combustion engine
technology phase is based on the time from development of internal combustion
technology and fielding in an aircraft. Note that the lowest technology level phase is 50
years, because many UAV and general aviation engines use internal combustion engines
with highly mature technologies.
42
Internal Comustion Engine Technology
Phase vs. Technology Level
0
10
20
30
40
50
0 0.2 0.4 0.6 0.8 1
Design Level
T
e
c
h
n
o
l
o
g
y
P
h
a
s
e
,
Y
e
a
r
s
Figure 425 Internal Combustion Engine Technology Phase
Internal Combustion Engine Technology
Factors vs. Technology Year
0
0.5
1
1.5
2
2.5
1940 1960 1980 2000 2020 2040
Technology Year
T
e
c
h
n
o
l
o
g
y
F
a
c
t
o
r
s
P/W Fact
SFC fact
Figure 426 Internal Combustion Engine Technology Trends
4.3.6 Structures
Two approaches are used for estimating structures technologies  Factor of Safety (FOS)
and direct structural weight reductions. The attributes of these technologies are repeated
in Table 49.
43
S
u
b
s
t
a
n
t
i
a
t
i
o
n
Q
u
a
l
i
t
y
A
g
g
r
e
s
s
i
v
e
n
e
s
s
Structures
Structural Factor 2 3 1
Factor of Safety 3 2 2
Table 49 Structures Technology Attributes
Factor of Safety
The factor of safety for UAV structures was reduced from 1.5 to 1.25 as late as 1994.
This was done to account for the absence of a pilot and the associated safety
considerations. A standard UAV minimum factor of safety is assumed to hold constant
until 2005. This trend is then assumed to become more aggressive into the future, but for
different reasons. By 2025, the factor of safety is assumed to be reduced due to improved
structural analysis methods and improved manufacturing quality control. The 2025 factor
of safety for high technology UAVs is assumed to be 1.15. Figure 427 shows the factor
of safety technology trends.
Factor of Safety vs. Design Year
0
0.5
1
1.5
2
2.5
1920 1940 1960 1980 2000 2020 2040
Design Year
F
a
c
t
o
r
o
f
S
a
f
e
t
y
0
0.5
1
Figure 427 Factor of Safety Technology Trends
Structural Weight Factors
The second structural technology factor relates to structural weight directly. Noor [2000]
provides relevant results of Defense and Aerospace [1995] study, which predicts aircraft
weight reduction due to various technologies from 1995 to 2020. For long range
transonic transports, the anticipated total gross weight reduction due to structural
44
contributions is estimated at 24%. This reduction, however, is not the structural weight
reduction, but the total synthesized aircraft weight reduction. The structural contribution
is found by a simple mass fraction analysis similar to that performed on the subsystems.
A very simple sizing relationship for a given aircraft class is:
Tech Struct Struct Struct E Fuel Payload
TO
F MF MF MF W
W
, ,
1
1
⋅ − − −
=
−
Numbers representative of the Boeing 777 were used as an estimate for the 1995 mass
fraction numbers. The fuel mass fraction is assumed to be 0.4. The mass fraction of the
empty weight less structure is assumed to be 0.25. The mass fraction of the 1995
structure is assumed to be 0.2, when the structural weight multiplication factor is equal to
1. In order to achieve a 24% reduction in takeoff gross weight due to structural
contributions, the structural weight multiplication factor must equal 0.765. The 1960
structural weight multiplication factor is assumed to be 1.1, or 10% greater than the 1995
structural weight, since advanced finite element structural design was introduced between
these dates. This technology trend is assumed to be applicable to UAVs. The structural
weight multiplication factor and factor of safety are applied independently. The
structural technology phase and technology multiplication factors are found in Figure
428 and Figure 429, respectively.
Structural Technology Phase vs. Technolgy
Level
0
1
2
3
4
5
6
0 0.2 0.4 0.6 0.8 1
Technology Level
T
e
c
h
n
o
l
o
g
y
P
h
a
s
e
,
Y
e
a
r
s
Figure 428 Structural Technology Phase
45
Structural Weight Multiplication Factor vs.
Design Year
0.6
0.8
1
1.2
1950 1960 1970 1980 1990 2000 2010 2020 2030
Design Year
S
t
r
u
c
t
u
r
a
l
W
e
i
g
h
t
M
u
l
t
i
p
l
i
c
a
t
i
o
n
F
a
c
t
o
r
Figure 429 Structural Weight Multiplication Factor Technology Trend
46
Chapter 5 Geometry
5.1 Geometry Overview
Several geometry definitions are required for various analyses within the AV design
code. The aerodynamics, structures/weights, and graphical generators all require
specialized geometry evaluations. A parametric geometry model is employed to facilitate
the optimization process.
5.2 Geometry Definition
The geometry definition is used to define aircraft geometry in detail. The complex
geometry definition takes general and specialized forms. The general form will be
discussed first. The specialized forms are essentially customized paneling that supports
aerodynamics and structures analysis.
5.2.1 Wings and Tails
To streamline the code, the wing, horizontal tail, and vertical tail definitions are
combined, since all are aerodynamic surfaces with almost identical geometric properties.
The wing geometry equations presented in Appendix A apply to all three surface types
unless specified otherwise.
The airfoil coordinates may be input directly from a designated airfoil coordinate file or
generated parametrically. The airfoil coordinate files take the form of the UIUC airfoil
coordinate database data files to increase utility. The airfoil coordinate file approach is
applied only to the graphics file generator, and not the analysis procedures. The
parametric airfoil generation uses the NACA 4 and 5 series airfoil equations. The NACA
parametric airfoil method is available to all applicable subroutines that require airfoil
geometric information.
It is necessary to know the wing span, root chord, area and aspect ratio. However, these
parameters are not independent. If any two are known, then the remaining two can be
determined through a set of equations. The user selects which two are input. This
approach permits the greatest flexibility.
5.2.2 Fuselage
A robust yet simple fuselage geometry formulation is applied. The goal is to minimize
the quantity of data needed to describe a wide range of fuselage geometries. The selected
method employs conic sections applied across multiple fuselage stations. Conic sections
47
are extremely useful for metal aircraft, because conic sections describe shapes that 2
dimensional sheets may attain through simple wrapping. A conic curve may be described
by three points and a shape parameter. Composite aircraft often have complex curvature
that does not require conic sections, yet the flexibility in geometry description still makes
conic sections a useful geometry analysis tool.
Each fuselage file data contains geometry descriptions in the following format:
x/L z/Lm z/Lt z/Lb y/Ls rhot rhob
0.000 0.000 0.000 0.000 0.000 0.414 0.414
0.005 0.000 0.003 0.003 0.003 0.414 0.414
……. …… ……. ……. …… ……. …….
x/L
f
(j) z/L
m,f
(j) z/L
t,f
(j) z/L
b,f
(j) y/L
f
(j) ρ
t
(j) ρ
b
(j)
The xcoordinate moves from the nose to the tail, the ycoordinate moves out the right
wing, and the zcoordinate is positive in the updirection. All x, y, and z data are
normalized to the fuselage length. Three zcoordinate data points are defined for each x
coordinate. The first is a midline, which defines the widest portion of the fuselage for a
given section. The second is the upper surface of the fuselage, and the third is the lower
surface of the fuselage. The single ycoordinate data point defines the maximum width
of the fuselage. The ρ parameters define the fullness of the fuselage cross section such
that the surface between adjacent fuselage segments is a conic section. A ρ value of 0
corresponds to a straight line (triangle), 1 is a rectangle, and 0.414 is an ellipse. The ρ is
defined separately for the upper and lower surfaces.
Figure 51 shows the Predator nose region. The upper conic parameter ρ for the main
fuselage is 0.35, and the EO/IR ball has a conic parameter of 0.414. The EO/IR ball is
the sphere under the Predator fuselage in Figure 51. Note that lines of intersection are
not shown in these AutoCAD
TM
grid drawings.
Figure 51 Predator Fuselage Nose Region
48
Figure 52 shows the X45A geometry as represented by this geometry definition. The
conic parameter is 0.3 for the entire upper surface.
Figure 52 X45A Geometry
Figure 53 shows the Dragon Eye fuselage representation. The selected conic parameter
is 0.75 for the center fuselage, giving a boxy appearance.
Figure 53 Dragon Eye Fuselage Geometry
5.2.3 Propeller
The propeller input is very similar to that of the wing except there are many segments.
The first input is the name of the propeller geometry file. The file has the following
form:
r/R c/R Beta Sweep nsweep ntwist
0.0 0.08 35. 0. 0.25 0.25
0.05 0.08 35. 0. 0.25 0.25
…. ….. …. … ….. ……
r/R
p
(j) c/R
p
(j) β(j) Λ
p
(j) x/c
Λ,p
(j) x/c
β,p
(j)
All dimensions are normalized to the propeller radius. The radial position r/R
p
goes from
0 to 1. β is the blade twist angle relative to the rotational plane. x/c
Λ,p
is the reference
chordwise location for the blade sweep. x/c
β,p
is the reference chordwise location for the
wing blade twist. The propeller chord and twist distribution geometry used for graphics
generation and the detailed propeller analysis models are not linked. However, this does
49
not affect the results because the propeller radius is consistent throughout the code and
the detailed propeller design methods are not directly utilized here. The only negative
outcome is that the graphical output of the chord and twist distribution do not match.
5.3 Parametric Geometry
The analysis of a known aircraft with a fixed geometry lends itself well to detailed
geometry inputs described above. However, optimization requires that geometry
components, such as wings, tails and bodies, move together in a sensible manner. For
example, the tails and wings should remain attached to the fuselage when so configured.
Simply making the wing geometry and fuselage length design variables does not
guarantee the resulting optimized geometry solution is reasonable. Geometry sizing laws
must be established.
5.3.1 Parametric Fuselage
A parametric fuselage formulation is used for all optimization cases considered. The
geometry selected supports both jet and propeller configurations. For the propeller case,
the propeller is centered at 96% of the fuselage length. Pusher propellers are applied to
the parametric geometry for propellerdriven aircraft, because both the Shadow 200 and
Predator use pusher propellers. For jets, the engine nacelle is located on top of the
fuselage such that the rear of the nacelle is located at 95% of the main fuselage length, for
configuration similarity to the Global Hawk. Figure 54 and Figure 55 show how
propellers and jets are applied to the generic fuselage, respectively.
Figure 54 Propeller Integration on Generic Fuselage
50
Figure 55 Jet Integration on Generic Fuselage
The aerodynamic center of the wing is fixed to 48% of the main fuselage length,
regardless of sweep in order to keep a reasonable tail moment arm for tail sizing. 48%
appears to be a reasonable number based on visual inspection and from experience with
early sizing attempts. The wing root leading edge will shift forward relative to the
fuselage as wing sweep increases. The wing root leading edge is always defined as the
origin. The wing root leading edge rests at the bottom of the fuselage at the point of
intersection.
The parametric fuselage permits blending with other bodies. The mostoutside dimension
at a given point for two bodies is selected as the final outer geometry for the resultant
body. This holds true for upper and lower heights and the width. The maximum conic
parameter of the two bodies is applied for samehalf upper or lower surfaces.
The wing fairing is one example of fuselage blending. A fairing body with a length twice
the wing root chord and centered about the wing is blended with the main fuselage. The
resultant blended fairing geometry can be seen in Figure 56.
Figure 56 Wing Fairing Geometry
Two of the three vehicle classes have radomes to house dish antennas for satellite
communications. The proportions of the radome relative to the fuselage vary, and a
single fuselage geometry that represents the radome geometry impacts is not possible. A
51
parametric radome body is applied, which is parametrically linked to the dish diameter.
Figure 57 shows the radome geometry impact on the fuselage.
Figure 57 Radome Impacts on Fuselage Geometry
If unconstrained, the optimizer tends to minimize the fuselage length such that volume
constraints are satisfied. This raises the possibility of creating an infeasible solution for
tail sizing, as will be discussed shortly. A fuselage minimum length constraint is
necessary. Rather than creating a constraint with a penalty function for the optimizer, the
fuselage length is automatically set to the minimum length if the input fuselage length is
too low. There are two minimum fuselage length definitions, and the fuselage length
must be greater than the greatest of the two. First, the fuselage minimum length is 35%
of the wing span. Second, the fuselage length is:
Wing AC
avg Wing
fuse
L x
c
L
,
,
/ 1
4
−
⋅ =
The first minimum fuselage length protects the fuselage minimum length for high aspect
ratio wing cases, and the second minimum protects the fuselage length for low aspect
ratio wing cases.
An EO/IR ball is attached to, but not blended with, the main fuselage. The diameter of
the ball is found through the EO/IR payload calculations. The ball is fixed at 10% of the
fuselage length and approximately 20% of the ball height is submerged in the main
fuselage. The relative location of the ball is inspired by the Predator configuration and
based on visual inspection. Figure 58 shows the EO/IR ball geometry.
52
Figure 58 EO/IR Ball Geometry
The jet engine nacelle, partially described earlier, is another body that is attached to the
main fuselage but not blended. The nacelle length to diameter ratio is assumed to be 1.1
times greater than the jet engine length to diameter ratio. The nacelle length is assumed
to be 1.8 times greater than the jet engine length. The nacelle length to diameter and
overall length relative to the engine is based on visual inspection of appropriate
proportions.
5.3.2 Parametric Tails
The parametric tail geometry uses a tail volume coefficient methodology applied to a V
tail. Vtails are used here, because all three calibration cases use variants this tail
configuration. Conventional tails or other alternative tail configurations could also be
applied in a similar fashion. A closed form solution to the tail sizing for a fixed tail
trailing edge location is not forthcoming, so a convergence approach is used. First, the
dihedral angle of the Vtail is found by:


.

\

⋅
⋅
= Γ
−
−
w H
w V
Tail V
MAC TVC
b TVC
1
tan
The length between the wing aerodynamic center and the Vtail trailing edge, L
ACVT,TE
, is
found such that the Vtail trailing edge is located at 93% of the fuselage length. The
secant method is employed to solve for tail area such that:
( ) ( )
( ) ( ) ( ) 0 cos sin
cos
,
= Γ + Γ ⋅
Γ ⋅ ⋅ −
⋅
⋅ −
− −
− − −
− Tail V Tail V
Tail V Tail V Func TE VT AC
w w
H Tail V
S AC L
MAC S
TVC S
where AC
Func
is the distance from the Vtail root chord aerodynamic center to the trailing
edge normalized by the square root of with Vtail area.
53
This tail sizing approach permits the tails to automatically scale with the wing and tail
moment arm. The tail root vertical height is fixed to the maximum width location on the
fuselage at the attachment location, which automatically has vertical separation from the
wing. The vertical separation combined with upwardcanted Vtails ensures that the lift
distribution methods will not have artificial interference leading to singularities.
5.4 Wing Lift Distribution Paneling
Three lift distribution methods are employed: vortex lattice method, lifting line theory,
and Trefftz plane analysis. All are described further in the Aerodynamics section. Each
of the methods require specific geometric paneling, but sufficient commonality exists for
a single geometry definition subroutine to handle all three cases.
There are two primary methods of paneling the wing. The first method panels each
segment individually. The second method panels the entire wing by blending segments.
Further discussion of these methods can be found in Appendix A.
First, reference parameters are defined. The lifting line theory analyzes wings
sequentially rather than simultaneously, so the reference wing area and span is for each
individual wing. The reference area for the other methods is the wing area of the main
wing. The total average chord for the Trefftz plane analysis is the ratio of the first wing
planform area to the first wing’s span.
Figure 59 and Figure 510 show the Virginia Tech Forestry UAV and its wing paneling
for lift distribution methods, respectively. The front two faces of the triangles are xyz1
and xyz2. The rear triangle points are xyzm. The first paneling method was used. Note
that the panels on the inboard portion of the main wing are evenly spaced and the
outboard segment has sinusoidal spacing.
Figure 59 Virginia Tech Forestry UAV
54
Figure 510 Wing Aerodynamic Paneling for Virginia Tech Forestry UAV
5.5 Wing Structures Paneling
The code shares many wing paneling methods for aerodynamics and structures. The
wing is broken up into multiple discrete panels. The structural contours for the skins are
then determined using the NACA airfoil generator. The wing box for each panel begins
after a specified leading edge chordwise distance and ends at the control surfaces, if any
are present. Properties at the root and tip of the panel are calculated, and the average
value is used for the panel calculations. The properties under consideration are described
below.
The NACA airfoil coordinate generator calculates the 2D coordinates of the upper and
lower surface of the skin. Each skin is composed of straight line segments connecting
disretized points on the surface. The coordinate system used here is the same as the
aircraft coordinates. The length of each element is found by:
( ) ( )
2 2
) 1 ( ) ( ) 1 ( ) ( ) ( + − + + − = i Z i Z i X i X i L
pt pt pt pt elt
The midpoint of each element is found by:
2
) 1 ( ) (
) (
+ +
=
i X i X
i X
pt pt
mid
2
) 1 ( ) (
) (
+ +
=
i Z i Z
i Z
pt pt
mid
The angle of the element from the xaxis in the xz plane is found by:
55


.

\

− +
− +
=
−
) ( ) 1 (
) ( ) 1 (
) (
1
i X i X
i Z i Z
Tan i
pt pt
pt pt
elt
θ
The total length of the wing box surface excluding the front and back walls is found by:
Bottom Top
nelt
i
elt tot Surf
i L L
,
1
,
) ( 
.

\

=
∑
=
The centroid calculations neglect wing thickness because the model assumes that the
wing skin thickness is uniform. The wing box centroid is found by:
Bottom Top
tot Surf
nelt
i
elt mid
cent
L
i L i X
X
,
,
1
) ( ) (




.

\

⋅
=
∑
=
Bottom Top
tot Surf
nelt
i
elt mid
cent
L
i L i Z
Z
,
,
1
) ( ) (




.

\

⋅
=
∑
=
The X and Z distance from each point is found by:
) ( ) ( i X X i X
mid cent cent
− = ∆
) ( ) ( i Z Z i Z
mid cent cent
− = ∆
The area moments of inertia divided by the skin thickness are found by:
( )
Bottom Top
nelt
i
cent elt elt
elt
t XX
i Z i L i Sin
i L
I
,
1
2 2
3
,
) ( ) ( ) (
12
) (


.

\



.

\

∆ ⋅ + ⋅ =
∑
=
θ
( )
Bottom Top
nelt
i
cent elt elt
elt
t ZZ
i X i L i Cos
i L
I
,
1
2 2
3
,
) ( ) ( ) (
12
) (


.

\



.

\

∆ ⋅ + ⋅ =
∑
=
θ
( ) ( )
Bottom Top
nelt
i
cent cent elt elt elt
elt
t XZ
i Z i X i L i Cos i Sin
i L
I
,
1
3
,
) ( ) ( ) ( ) ( ) (
12
) (


.

\



.

\

∆ ⋅ ∆ ⋅ + ⋅ ⋅ =
∑
=
θ θ
56
The shear flow at the centroid divided by thickness considers only the contribution of the
wing skins. Shear web contributions are neglected for the purposes of the calculations.
The shear flow divided by thickness is found by:
Top
nelt
i
elt t Shear
i Z i L q 
.

\

∆ ⋅ =
∑
=1
,
) ( ) (
The above calculations were performed on both the root and tip sections of the wing
panel. Next, the average panel characteristics must be calculated. The coordinates of the
panel centroid are simply the average of the root and tip coordinates. The skin thickness
corrected area moments of inertia for the panel is the average of the root and tip values.
The skin thickness corrected shear flow is the average or the root and tip values. The
total thickness of the panel wing box is the average of the root and tip thicknesses. For
the geometry calculations, k is the index for the structural panels. The structural sweep
of the panel is found by:


.

\

−
−
= Λ
−
) ( ) (
) ( ) (
) (
, ,
, , 1
k Y k Y
k X k X
Tan k
Root cent Tip cent
Root cent Tip cent
cent
The centroid dihedral of the panel is found by:


.

\

−
−
= Γ
−
) ( ) (
) ( ) (
) (
, ,
, , 1
k Z k Z
k X k X
Tan k
Root cent Tip cent
Root cent Tip cent
cent
The structural span of the panel is found by:
( ) ( ) ( )
2
, ,
2
, ,
2
, ,
) ( ) ( ) ( ) ( ) ( ) ( ) ( k Y k Y k Y k Y k X k X k b
Root Cent Tip cent Root Cent Tip cent Root Cent Tip cent Struct
− + − + − =
The total surface area of the skins for each panel is found by:
( )
) (
2
) ( ) (
) (
, ,
k b
k L k L
k S
Struct
Root Surf Tip Surf
Pan
⋅
+
=
The results of the structural paneling are shown in Figure 511 and Figure 512. The
shear web is not included.
57
Figure 511 Vehicle Structural Paneling Example
Figure 512 Wing Tip Paneling Detail
5.6 Graphics File Generator
The graphics file generator outputs data exchange format (DXF) drawing files that can be
accessed by Computer Aided Drawing (CAD) software. The purpose of the graphics file
generator is to provide nearly photorealistic 3D renderings of the designs. The
graphical outputs provide a means of rapidly evaluating the configurations for
fundamental flaws. The graphics also provide an effective means of communicating the
behavior of the optimization results, where tables and plots become cumbersome. The
graphics file generator automatically outputs the geometry of the vehicle of interest.
The graphics file generator has the ability to evaluate a wide variety of configurations.
Not all of this functionality is utilized in this research. The graphics generating
capabilities are inspired by the work of Libeau [1999]. Figure 513 shows one
unconventional configuration.
58
Figure 513 Example Unconventional Configuration
The wing, nacelle, and propeller rotation options allow the listed components to be
rotated such that the incidence is modified about some point P
rot
. This is necessary to
graphically represent tiltwing V/STOL configurations as shown in Figure 514.
Figure 514 TiltWing Graphics Option
The rotation feature and other capabilities facilitate rapid drawings of new concepts.
Typically, any new aircraft concept can be graphically represented within 2 hours.
59
Chapter 6 Aerodynamics
6.1 Aerodynamics Overview
The aerodynamics code models a large variety of aircraft shapes in the subsonic and low
transonic flight regimes over a wide Reynolds number range. Both simple and complex
methods are employed with varying degrees of accuracy, computational efficiency, and
flexibility. The primary product of the aerodynamics code is a drag polar for use in the
mission analysis subroutine.
The aerodynamics code utilizes a number of methodologies to evaluate the aerodynamic
characteristics of relatively arbitrary vehicles. The various approaches permit different
types of questions to be answered, depending on the analysis objectives. The methods
also vary in robustness and computational efficiency. All methods must operate in
Reynolds numbers ranging from less than 100,000 to very large numbers on the order of
20,000,000 and Mach number between zero and the low transonic region where the drag
rise begins.
Early conceptual design work does not require the level of detail necessary in early
preliminary design. Early conceptual design focuses on simple sizing, and it does not
require exact definition of characteristics such as wing twist, thickness and camber
distribution or tail incidence. Early preliminary design generates much more detailed
information and, therefore, must model the impacts of detailed inputs. This research
focuses on both conceptual and early preliminary design aerodynamics methods.
The problem is decomposed into an incompressible solution with an additive
compressible correction. The incompressible solution is assumed to be equivalent to
Mach zero at a given Reynolds number, with no compressibility effects on lift or drag.
The compressible correction is zero until the critical Mach number is reached, after which
the drag rises with Mach number according to the Korn equation [Grasmeyer 1998]. Lift
is not corrected for compressibility, which reduces the number of aerodynamics tables
that must be generated. There is no consideration for MachReynolds number
interactions.
The incompressible drag is composed of induced drag, aerodynamic surface induced
drag, profile drag, fuselage drag, interference drag, landing gear drag, and miscellaneous
drag. Induced drag contains both traditional drag due to lift and some trim drag. The
profile drag is the additive combination of the friction and pressure drag on aerodynamic
surfaces such as wings and tails. Interference drag consists of drag at intersections of the
wings and bodies, and drag reductions due to reduction of wetted areas at the
intersections. Gear drag is calculated in the extended and retracted positions, and
consists of the tire, strut, door, and cavity contributions, as appropriate.
60
The various methodologies book keep drag in different ways. For example, induced drag
can also include profile drag due to lift and profile trim drag, where these components are
subtracted from the profile drag.
Aerodynamics interacts with propulsion and other systems. Propulsion interactions are
captured as drag penalties in the propulsion section. Propulsion effects on lift and
pitching moment are neglected. Cooling drag is captured as a drag penalty in the
propulsion analysis during mission execution, though these influences will be described
in the aerodynamics chapter.
The philosophy adopted for this research is to place the greatest emphasis on the largest
contributors to the aerodynamic behavior of interest. The lift methods are quite detailed
to provide detailed lift distribution information to determine local stall, usable total lift
coefficient, trim requirements, and induced drag. The lifting surface profile drag is
another significant drag driver that is given much attention.
The inputs for the aerodynamics code include vehicle geometry and paneling,
aerodynamics parameters, flight regime limitations, method selections, drag polar array
size parameters, and technology data.
The primary output is a drag polar and an additive compressibility correction array. The
drag polar consists of angle of attack and Reynolds number dependent lift and drag arrays
and a suitability array. The suitability array determines if the drag array element is usable
by considering elevator deflection and local stall constraints. The compressibility
correction array consists of a Mach array starting with the critical Mach number, and a
drag rise array corresponding with the Mach array. The primary products are shown in
Table 61.
Product Description
C
L
(α,Re) Total lift coefficient array
C
D
(α,Re) Total drag coefficient array
Allow (α,Re) Permissible data point flag array
C
Dw
(α,M) Total wave drag coefficient array
Table 61 Primary aerodynamics product list.
6.2 Aerodynamics Code Architecture
This section describes the aerodynamics code architecture and its components at a high
level. Subsequent sections will discuss the methodologies used in the subroutines in
more detail.
Much information is required prior to running the AEROC aerodynamics analysis routine.
Airfoil data input and consderable geometry definition occurs beforehand. The
MISSPARAM subroutine estimates the minimum and maximum flight Reynolds number
61
per unit length based on the flight envelope boundaries. The CGLOC subroutine
estimates the aerodynamic center based on the geometry and locates the center of gravity
based on the aerodynamic center and static margin. Once the inputs are properly
conditioned, the AEROC subroutine is called by EVAL.
Figure 61 AEROC code Architecture
Figure 61 shows the AEROC code architecture. The first step is to generate the final
paneling used for the lift distribution methods, which was described in the Geometry
section. If the vortex lattice method is used, then the velocity influence coefficients are
calculated up front with subroutine VLMUVW and INFLVPM, respectively. The zero lift
angle of attack for the panels is determined in subroutine AIRFOILI. The positive and
negative stall angles of attack for all panels are found in subroutine CLMINMAX.
If either the VLM or LLT are selected, then the elevator position to trim the vehicle at a
given angle of attack is determined through the secant method by setting the pitching
moment about the center of gravity equal to zero. An elevator position and angle of
attack, among other inputs, are sent to subroutine CM, which then outputs the pitching
moment about the center of gravity. If the Trefftz plane analysis is selected, then the lift
coefficient is input, and the lift distribution, induced drag, and angle of attack are output.
Once trimmed, the profile drag for each panel is calculated by one of three methods in
CDPCALC. The wave drag is calculated for a given lift distribution and Mach number in
WAVEDRAG. The fuselage drag is calculated in subroutine FUSEDRAG. The
interference drag is calculated in subroutine INTDRAG. The retracted and extended
landing gear drag contributions are calculated in subroutine GEARDRAG.
62
Figure 62 Pitching Moment Subroutine Architecture
Figure 62 shows the architecture of the pitching moment subroutine, CM. First, flap
efficiency parameters are calculated in subroutine, FLAPETA. Then, either the LLT or
VLM subroutines are called to determine the lift distribution as a function of angle of
attack and elevator deflection. The airfoil 2D pitching moment subroutine, CMPROF, is
called. The pitching moment caused by the lift moment arm about the center of gravity is
calculated in subroutine CMLIFT. The pitching moment due to flap deflections is
determined in subroutine CMFLAP. Finally, the fuselage pitching moment is calculated
in subroutine CMFUSE. These pitching moments are used to help determine how to trim
the aircraft, as described in the lift methodology section of this chapter.
6.3 Lift Methodology
6.3.1 Lift Methodology Overview
The lift distribution of a configuration must be determined to evaluate the total lift
coefficient, C
L
, at which local stall first occurs and induced drag. The lift distribution
information is also necessary for determination of profile drag and loads for structural
analysis.
6.3.2 CG Location Determination
63
The center of gravity location must be known in order to estimate the trimmed lift
distribution. The key inputs to the center of gravity location are the vehicle geometry and
the static margin.
The geometry definition subroutines calculate the Xcoordinates of the aerodynamic
center of all lifting surfaces, wing mean aerodynamic chord, and tail volume coefficients.
The origin of this coordinate system is the front of the main wing root chord, and X is
positive moving towards the rear of the aircraft.
The end equation for the Xcoordinate of the center of gravity is:
c sm X X
NP CG
⋅ − =
where the neutral point for the aircraft is defined as:
( ) c h X X
n wing AC NP
⋅ − + = 25 . 0
,
The wing aerodynamic center is found though standard multisegmented trapezoidal wing
methods. The aircraft neutral point is found using standard methods found in Etkin
[1996] and other stability and control texts. The derivative of the fuselage pitching
moment with respect to angle of attack is found using methods described in Nelson
[1989]. The fuselage is broken into a number of horizontal strips perpendicular to the X
axis.
6.3.3 Lift Distribution Methods
Three lift distribution methods are employed in this code: Prandtl lifting line theory
(LLT), vortex lattice method (VLM), and a Trefftz plane analysis. The first two methods
take a detailed geometry definition, including camber effects (α
0L
shift and elevator
deflections) and washout. The elevator deflections are adjusted at a given angle of attack
to trim the vehicle. The Trefftz plane analysis calculates the minimum induced drag
trimmed lift distribution for a given planform geometry without determining how the
distribution was generated. The Trefftz plane analysis was applied to generate all results
for this research, as shown in Table 62.
Method Applied Here
1. Prandtl Lifting Line Theory No
2. Vortex Lattice Method No
3. Trefftz Plane Analysis Yes
Table 62 Lift Distribution Methods
64
6.3.3.1 Prandtl Lifting Line Theory
The wing lift distribution is determined by using Prandtl's line theory as presented in
Davenport [1998]. This method allows varied chord, camber and twist distributions
along the span. The method provided in Bertin and Smith [1989] describes only a single
taper ratio directly. In the method used, sweep and dihedral cannot be accounted for and
it is assumed that the wing is tapered about the quarter chord. The matrix formula takes
the form:
( )
∑
∞
=
+
⋅ ⋅
= − +
odd n
n L twist
s
n c
n A
s
c
, 1
0
sin
4
) sin( sin
4
θ
π
θ θ α α α
π
where
) / ( cos
1
s y − =
−
θ
υ is divided up into N segments between 0 and π/2 radians, and n assumes odd integer
values from 1 to 2N1. Note that in effect the integer 2N1 replaces : in the above
equation. The chord at the given υ, c, is found through linear interpolation across the
semispan.
The following matrix formula is solved:
     ) ( ) ( ) , ( i b i x j i A =
where
1 2 − ⋅ = j n
( ) ( )


.

\

+
⋅
⋅ ⋅
⋅ ⋅ = ) ( sin
2
) (
) ( sin ) , ( i
b
n i c
i n j i A
w
θ
π
θ
( ) ( ) ) ( sin ) ( ) (
2
) (
) (
0
i i i
b
i c
i b
L twist
w
θ α α α
π
⋅ − + ⋅
⋅
⋅
=
The coefficients x(i) are determined by the matrix operation:
      ) ( ) , ( ) (
1
i b j i A i x
−
=
The total lift coefficient for the wing is:
) 1 ( x AR C
L
⋅ ⋅ = π
and the induced drag is
65
e AR
C
C
L
Di
⋅ ⋅
=
π
2
where
( )
∑
−
=
⋅ +
=
1 2
, 3
2
) ( / ) ( 1
1
N
odd n
l x n x n
e
The section lift coefficients at υ are:
( )
∑
−
=
⋅ ⋅
⋅
=
1 2
, 1
) ( sin ) (
8
) (
N
odd n
l
i n x
c
s
i C θ
The tip lift coefficient is always zero, so υ = 0 is not included in the A matrix.
The LLT can only analyze a single wing at a time, so downstream downwash effects on
tails can not be evaluated directly. The tail is modeled as a new wing without an
upstream wing, which creates inaccuracies. The downstream surface is always more
affected by the upstream surface than vice versa, because the upstream wing has trailing
vortices. A wing downwash model that creates a new tail effective local angle of attack
distribution would improve the tail lift predictions, though this is not explored further
here. Because trailing vortices from upstream wings are not considered, there is no
possibility for tail calculations to fail due to singularities induced by the other surfaces.
The inherent limitations of the LLT are that the tail effectiveness is overpredicted, and
the tail contribution to induced drag is underpredicted. Although a selectable option in
the code, the LLT method was not used to generate the results for this research.
6.3.3.2 Vortex Lattice Method
The vortex lattice method (VLM) offers the ability to evaluate more complex
configurations than the LLT. Sweep and dihedral can now be modeled. Configurations
with multiple surfaces can be analyzed simultaneously.
The 2D and 3D vortex lattice methods of Bertin and Smith [1989] are used directly in
the code. As in the example provided in Bertin and Smith [1989], only one panel per
chord length is used. The 3D formulation determines the streamwise, spanwise and
vertical velocities (u
m
, v
m
, and w
m
) induced by the circulation by ensuring that there is no
flow passing through each control point m. This can be shown by:
) cos( ) sin( ) cos( ) cos( ) sin( ) cos( ) cos( ) sin( φ δ α δ φ φ δ φ δ ⋅ − ⋅ − = ⋅ ⋅ + ⋅ ⋅ − ⋅ ⋅ −
∞
U w v u
m m m
66
The angle of the control point at zero angle of attack, δ, is replaced by the zero lift angle
of attack minus local incidence, α
0L
α
twist
, because the airfoil geometry is not known for
many cases. This yields some inaccuracies, but a direct relationship between δ and α
0L
cannot be established for both conventional and reflexed airfoils. Also, δ and α
0L
are
intuitively interchangeable on the right hand side of the matrix equation, since a change
of δ in the term (αδ) should have the same effect on local lift as a change in α
0L
α
twist
.
The right hand side of the equation was modified further to account for the fact that the
angle of attack has less direct influence on the lift of an airfoil as the local dihedral
approaches vertical. Additionally, U
:
is not necessary to determine the lift coefficient
distribution, so all velocities are normalized by U
:
. Using the matrix form:
     ) ( ) ( ) , ( i b i x j i A = ,
the right hand side of the equation is modified to:
( ) ( ) ) ( ) ( ) ( cos sin ) (
0
i i i a i b
L twist
α α φ − + ⋅ − =
The x(i) coefficients are determined by the matrix operation:
      ) ( ) , ( ) (
1
i b j i A i x
−
=
The 3D code is likely to yield unrealistic results if the trailing vortex from an upstream
panel approaches a downstream panel control point, so it is important to ensure sufficient
vertical separation between tandem aerodynamic surfaces. The user must control this
through inspection for arbitrary configurations. The parametric geometry model used for
optimization automatically ensures that these singularities do not occur. This is
accomplished through surface vertical separation.
The elements of the 2D and 3D A(i,j) matrix are presented in Bertin and Smith [1989],
and they will not be reproduced here. The matrix represents the influence of the bound
and trailing vortices of all panels on all control points.
The 2D formulation neglects dihedral effects and vertical separation of panels. By
eliminating a dimension, the problem is greatly simplified and computation time is
shortened. The 2D formulation increases the risk of panel interference. The 2D
equation becomes:
) sin( ) cos( δ α δ − ⋅ − = ⋅
∞
U w
m
As with the 3D case, the velocities are made nondimensional by U
:
and δ is replaced by
α
0L
α
twist
.
The local lift coefficient is found by:
67
) (
) ( 2
) (
i c
i x
i C
l
⋅
=
The total vehicle lift coefficient is found by:
∑
=
⋅ ∆ ⋅ ⋅ ⋅
=
N
i w
l
L
S
i i y i c i C
C
1
)) ( cos( ) ( ) ( ) ( 2 φ
where the factor 2 accounts for ∆y being summed over only one semispan by assuming
symmetry in the lift distribution.
The induced drag is found by evaluating the downwash in the Trefftz plane and then
relating it back to the downwash at the wing surface. This is similar to the approach
taken in the Trefftz plane analysis, which is described later. The Trefftz plane can be
thought of as the conditions associated with the ideal flow at a location infinitely
downstream of the lifting surfaces. The downwash on the bound vortex is half that of the
associated downwash infinitely downstream. The induced downwash influence
coefficient matrix A
i
(i,j) is determined for a position 100 main span lengths downstream
of the origin. The equivalent nondimensional velocity in the Trefftz plane is determined
by the matrix b(i).
     ) ( ) ( ) , ( i b i x j i A
i
=
Note that the A
i
(i,j) matrix inverse does not need to be evaluated in this operation, since
x(i) for this system was previously determined.
The downwash angle, ε, is found by:


.

\
 ⋅
=
∑
=
−
N
j
i
j b j i A
i
1
1
2
) ( ) , (
tan ) ( ε
The local induced drag, C
di
(i), is found by:
) ( ) ( ) ( i i C i C
l di
ε ⋅ =
The total induced drag is found by:
∑
=
∆ ⋅ ⋅ ⋅
=
N
i w
di
Di
S
i y i c i C
C
1
) ( ) ( ) ( 2
As before, the factor of 2 represents symmetry.
The VLM analysis is available to the user, but it was not used to generate the results for
the research.
68
6.3.3.3 Trefftz Plane Analysis
The Trefftz plane analysis offers the advantage of automatically determining the optimal
trimmed lift distribution for minimum induced drag at a given C
L
. Unlike LLT and
VLM, no iterations of elevator deflections are necessary. There are, however, several
disadvantages. First, the method does not determine how this optimal lift distribution is
attained or if it is even possible to achieve. Second, the twist and camber distribution (if
it were determined) for one lift condition may not provide the optimal lift distribution at a
second condition. In other words, the method is optimistic by assuming that a single
vehicle can generate optimal lift distributions at more than one lifting condition without
variable geometry. Third, no angle of attack information is provided to or generated by
this method directly.
Despite its limitations, the Trefftz plane analysis is ideal for quickly analyzing and
assessing the optimized results of a lifting system with little definition. This makes the
Trefftz plane approach ideal for conceptual design. Methods like LLT and VLM become
more beneficial when further vehicle definition becomes available.
Grasmeyer [1998] discusses the details of this method in detail. Grasmeyer's general
methodology is employed here. The Trefftz plane analysis is used to generate all results
of this research.
6.3.4 Pitching Moment
Both the LLT and VLM provide lift distributions and induced drag estimations for a
given configuration at a given angle of attack and elevator deflection. However, there is
no guarantee that the configuration is trimmed at the given condition. The trimmed flight
condition must be known for performance estimation. The primary purpose of the
aerodynamics model is to determine the trimmed drag polar for use in the fuel burn
calculations in the performance model.
The trimmed condition occurs only when the summation of the pitching moments about
the vehicle center of gravity is equal to zero. The pitching moment contributors are the
lifting moments, airfoil zerolift pitching moments, flap pitching moments, and pitching
moments due to fuselages and nacelles. Note that the empennage lifting moments for a
conventional configuration, or elevatoraffected lift distribution on a flying wing,
counteract all other contributions to attain the trimmed condition. The means of
determining the trim requirements are handled by the lift distribution methods.
Nac Fuse M Flap M L M CL M M
C C C C C
, , , 0 , ,
+ + + =
The pitching moment due to lift for each panel is determined by:
69
( ) ) ( cos
) ( ) ( ) (
) ( 2
4 /
1
,
i
mac
i x x
S
i y i c
i C C
W
c CG
w
N
i
L CL M
φ ⋅
−
⋅
∆ ⋅
⋅ ⋅ =
∑
=
The zerolift pitching moment of the airfoil sections is determined by:
( ) ) ( cos
) ( ) (
) ( 2
1
0 , 0 ,
i
S
i y i c
i C C
w
N
i
L M L M
φ ⋅
∆ ⋅
⋅ ⋅ =
∑
=
The flap pitching moment is estimated by a series of empirical equations developed from
data of several flap types found in McCormick [1995]. The flap/elevator pitching
moment increment is estimated by:
2
,
) ( 141 . 0 ) ( 40 . 0 259 . 0 ) (


.

\

⋅ −


.

\

⋅ + − =
∂
∂
i
c
c
i
c
c
i
C
C
flap flap
L
flap M
The flap/elevator 2D lift increment due to deflection is:
) ( ) ( 2 ) (
,
i i i C
f f flap l
τ η δ π ⋅ ⋅ ⋅ ⋅ = ∆
where η
f
is found through interpolation of data found in McCormick [1995], and τ is
determined by an empirical equation from the same source.
The profile pitching moment contribution due to flap deflection is:
( ) ) ( cos
) ( ) (
) ( ) ( 2
1
,
,
,
i
S
i y i c
i C i
C
C
C
N
i
w
flap l
l
flap M
flap M
φ ⋅
∆ ⋅
⋅ ∆ ⋅
∂
∂
⋅ =
∑
=
The pitching moment due to fuselages and nacelles is determined by methods found in
Nelson [1989]. Note that fuselage and nacelle pitching moment contribution methods are
treated identically in this code, so the typical fuselage and nacelle subscript is covered by
the fuselage subscript alone. The pitching moment is:
∑


.

\

⋅
∂
∂
+ =
Fuselages
fuse M
fuse M fuse M
C
C C α
α
,
, 0 ,
The derivative of the fuselage pitching moment with respect to angle of attack was
discussed previously. The reference line used to determine fuselage angle of attack is the
zeroincidence fuselage reference. The zerolift fuselage pitching moment estimation
approach is the element method used by Nelson [1989]. Due to the order in which the
aerodynamics subroutines are called, the aircraft zero lift angle of attack is unknown
when this fuselage pitching moment contributor is invoked, so the zero lift angle of attack
is assumed to be zero.
70
For a given angle of attack, the only means of trimming the vehicle is through flap/
elevator deflections. Though deflections create profile pitching moments directly, their
greatest contribution comes from redistributing the lift on the surfaces. This creates a
pitching moment due to lift. The elevator deflection that creates a zero pitching moment
about the CG is found via the secant method. If the required elevator deflection falls
outside the permissible bounds, the solution is infeasible, and an error flag is set to
indicate the condition. Similarly, if the local C
l
is greater than C
l,max
, an error flag is set
to indicate the local stall condition. No poststall evaluations are performed, so the
maximum lift coefficient occurs when the first wing panel is stalled.
The Trefftz plane analysis requires a pitching moment input prior to calculating the
trimmed lift distribution. The zerolift profile pitching moment and fuselage pitching
moment coefficients are input into the method, and the trimmed lift distribution solution
ensures that the total pitching moment about the center of gravity is zero. No flap
pitching moments are included.
6.4 Drag Methodology
6.4.1 Drag Methodology Overview
A traditional conceptual design wing drag polar equation takes the following parabolic
form:
e AR
C
C C
L
D D
⋅ ⋅
+ =
π
2
0
where
∑
⋅ ⋅ ⋅ =
seg seg seg seg d
ref
D
S Q FF C
S
C
0 0
1
Recall that FF
seg
is the shape form factor, Q
seg
is the interference factor, and S
seg
is the
wetted area of the body or aerodynamic surface of interest.
Note that wave drag contributions are not under consideration for the purposes of this
discussion. This formulation requires that all liftdependent drag sources are included in
the second term on the right hand side of the first equation. The span efficiency factor, e,
is usually considered to be independent of lift, but it may vary with geometry. C
D0
accounts for the profile drag, which is usually held constant at the average Reynolds
number. C
D0
is also considered to be independent of lift.
The above formulation works reasonably well for aircraft with allturbulent or fixed
transition boundary layers, but it is not suitable for many practical applications. A more
suitable model is as follows:
71
e AR
C
C C
L
Dprof D
⋅ ⋅
+ =
π
2
where
∑
⋅ ⋅ =
seg seg seg prof d
ref
Dprof
S Q C
S
C
, ,
1
and
press d frict d prof d
C C C
, , ,
+ =
The profile drag is readily obtainable from airfoil wind tunnel test results.
6.4.2 Profile Drag Overview
The profile drag is composed of both friction and pressure drag contributions. Significant
attention is given to profile drag, because it is a major drag contributor, especially at low
Reynolds numbers. Using a realistic profile drag will yield an aircraft drag polar that
generally does not have a parabolic form at all Reynolds numbers. The upper and lower
transition locations are a strong function of lift coefficient and Reynolds number. The
transition location strongly affects the friction drag. The pressure drag coefficient is also
a function of lift coefficient and Reynolds number.
Initially, an Excel spreadsheet was developed to help formulate a series of parametric
drag polar functions. The desired equation would take the following form:
( ) ( ) ( )
n
drag l l drag l l d d d
C C b C C a C c t C C
min ,
2
min , 0 0
/ Re, − ⋅ + − ⋅ + ∆ + =
where equations for ∆C
d0
, a, b, and n are found from regression analysis and may be
dependent upon Reynolds number, camber and thickness. This equation worked well for
high Reynolds numbers (Re>3,000,000). At low Reynolds numbers (Re<300,000), the
drag polar behavior becomes erratic and no suitable formulation could be derived. At
very low Reynolds numbers (Re<100,000), no simple formulation with acceptable
accuracy is possible for most airfoils.
Figure 63 shows data for the NACA 0009 airfoil at very low, low and high Reynolds
numbers. Note that the parabolic features of the drag polar become much more
pronounced as the Reynolds number decreases. Finally, at a Reynolds number of 60,000,
the parabolic nature collapses. The NACA 0009 is a very predictable airfoil compared to
most that were investigated. The data is based on Abbot and Von Doenhoff [1959] and
the UIUC low Reynolds number wind tunnel tests [Selig 1995].
72
NACA 0009 Wind Tunnel Data
0
0.01
0.02
0.03
0.04
0.05
0 0.5 1 1.5
CL
C
D
Re=60K
Re=100K
Re=200K
Re=300K
Re=3M
Re=9M
Figure 63 NACA 0009 at Several Reynolds Numbers
Figure 64 is more representative of most airfoils in the lower Reynolds number range.
This airfoil, the SeligDonovan 7032, is relatively predictable at Reynolds numbers above
200,000. The shape of the drag polar at lower Reynolds numbers differs greatly between
airfoils. The data is based on the UIUC low Reynolds number wind tunnel tests [Selig
1995].
SD 7032 Wind Tunnel Data
0
0.01
0.02
0.03
0.04
0.05
0.5 0 0.5 1 1.5
CL
C
D
Re=60K
Re=100K
Re=200K
Re=300K
Figure 64. Selig Donnovan SD7032 at Low Reynolds Numbers
After expending considerable effort pursuing a parametric function, it became clear that a
different approach was necessary. The subroutine must work for all Reynolds numbers
73
encountered by several classes of UAV. Sacrificing fidelity at small UAV Reynolds
numbers is not acceptable.
Three methods of determining the profile drag were developed for this research, as shown
in Table 63. The direct airfoil data method interpolates from databases of airfoil data.
The parametric airfoil method applies parametric relationships for the pressure drag and
friction drag of several airfoils. Finally, the simplified airfoil method uses simple
relationships to relate profile drag to equivalent flat plate skin friction drag. The airfoil
interpolation and extrapolation methods can be applied to the direct airfoil data method
and the parametric airfoil method.
Profile Drag Method Applied Here
1. Direct Airfoil Data Method No
2. Parametric Airfoil Method Yes
3. Simplified Airfoil Method No
Table 63 Profile Drag Methods
6.4.2.1 Airfoil Interpolation
Unfortunately, aerodynamic data is not always available for an airfoil with the desired
thickness, camber and pitching moment. A method was developed to determine the drag
polar characteristics for an arbitrary airfoil given data from a variety of other airfoils.
Ideally, characteristics of the arbitrary airfoil should be bounded by those of the reference
airfoils. The data is interpolated for a fixed Reynolds number and incompressible flow.
Compressibility effects are described separately. Figure 65 shows a graphical
representation of the problem.
t/c
camber
C
Mc/4
Figure 65 Airfoil 3D Interpolation Space
Traditional multidimensional interpolation would require all but one of the independent
parameters to remain constant while the dependent variable is found for the nonconstant
independent parameter. Assume that, for a given Reynolds number and lift coefficient,
the drag coefficient must be determined for an airfoil of a certain thickness, camber and
74
pitching moment. The last three characteristics are independent parameters, and their
values for the target airfoil must be bounded by the reference airfoils surrounding it. An
interpolation in one parameter requires two reference data points. Using regular
parameter increments, a twoparameter interpolation would require four data points.
Assuming regular parameter increments, the number of required reference airfoils is 2
n
,
where n is the number of independent variables. Thus, there must be exactly eight
reference airfoils for three independent parameters if regular parameter increments are
assumed.
The volume of airfoil data required to perform such an interpolation for all Reynolds
numbers of interest would be difficult to obtain within the scope of this research. The
goal of the methodology applied here is to find a way of interpolating between relatively
few reference airfoils.
First, a reference 'radius' is determined between the characteristics of the target airfoil and
those of the reference airfoils. For the above example, the airfoil characteristics are
transformed into weighted lengths as follows:
( )
max
, arg
/
/ /
/
c t
c t c t
c t x
i ref et t
fact i
−
⋅ =
( )
max
, arg
camb
camb camb
camb y
i ref et t
fact i
−
⋅ =
( )
max ,
,
,
arg
,
,
m
i ref
m
et t
m
fact
m i
C
C C
C z
−
⋅ =
These lengths are then converted into equivalent radii:
2 2 2
i i i i
z y x r + + =
The algorithm takes a series of sequential weighted factors and penalties based on the
equivalent radii and the number of preceding cases. If the radius between a reference and
target airfoil is zero, then the two airfoils are identical. The factor for this reference
airfoil is equal to 1 and all others are zero. Assume that there are four airfoils, and the
target airfoil is equidistant from all, then the factor for all four is 0.25. The sum of the
factors is always equal to unity.
For a given airfoil drag polar, the drag coefficient for that airfoil is found by interpolating
first by lift coefficient (for two bounding Reynolds numbers) to get two drag coefficients.
These two drag values are then interpolated between the two bounding Reynolds
numbers.
The zerolift angle of attack, maximum lift coefficient and drag coefficient for each
airfoil are found via the airfoil lookup function. The target airfoil maximum lift
coefficient is the sum of the products of the factors and reference airfoil maximum lift
75
coefficients. The lookup lift coefficient is scaled from each reference/target airfoil
maximum lift coefficient ratio. The target airfoil drag coefficient at a given lift coefficient
is the sum of the products of the factors and reference airfoil drag coefficients. Figure
66 shows the interpolation plane. For this case, the factor for polar 1 and 2 is about 0.25
and 0.75, respectively. This procedure may be performed for any number of reference
airfoils greater than 1.
Polar 1
Polar 2
Polar Target
C
L
C
D
C
lmax
1
C
lmax
2
Figure 66 Drag Polar Interpolation Diagram for Two Airfoils
The procedure for determining the interpolation factors is relatively straightforward.
First, the initial weighting factors for airfoils 1 and 2 relative to one another are assessed:
) 2 ( ) 1 (
) 1 (
1 ) 1 (
r r
r
fact
beg
+
− = ,
) 2 ( ) 1 (
) 2 (
1 ) 2 (
r r
r
fact
beg
+
− =
Next, the initial weighting factors and initial penalties for the remaining airfoils are
assessed:
) 1 ( ) ( ) 1 (
) ( ) 1 (
1 ) (
− + ⋅ −
⋅ −
− =
i r i r i
i r i
i fact
eq
beg
) 1 ( ) ( ) 1 (
) 1 (
1 ) 1 (
− + ⋅ −
−
− = −
i r i r i
i r
i pen
eq
eq
beg
Where r
eq
(i) is the smallest equivalent radius up to reference airfoil i. Next, the factor
penalty is assessed:
∏
=
=
i
j
beg
j pen i factpen
2
) ( ) (
Finally, the factor for each airfoil is:
76
) ( ) ( ) ( i fact i factpen i fact
beg
⋅ =
The summation of all the factors always equals unity.
The airfoil database has a small population of airfoils. These airfoils span the range of
thickness, camber and pitching moment. Much of this data comes from UIUC lowspeed
wind tunnel test databases [Selig 1989 & 1995]. Other data is obtained from offline X
Foil [Drela 2000] airfoil design and analysis software outputs, which is described later.
The current approach is to use data for airfoils designed to operate at a given Reynolds
number rather than use a single airfoil over the entire Reynolds number range.
6.4.2.2 Airfoil Extrapolation
The airfoil interpolation subroutines focus on positive lift coefficients below stall.
However, most airfoils considered have a substantial nonstalled lifting region at negative
lift coefficients for which no data is available. It has been determined from analysis of
wind tunnel data that the negative stall occurs at an angle of attack equal in magnitude
and opposite in sign to that of the positive stall. This effect is independent of camber,
which merely shifts the lift curve up and down but does not impact the stall angles.
The parametric airfoil interpolation subroutine permits lift and drag extrapolation at
negative lift coefficients without modification. The predicted transition locations and
pressure drag terms exhibit no undesirable behavior under extrapolated conditions. The
stall angle of attack is assumed to be symmetric about the α=0
o
axis.
The directdata airfoil interpolation subroutine presents more challenges than the
parametric case. Again, the stall angle is assumed to be symmetrical. Assuming that the
behavior of the lift curve is symmetrical as well, with the change in slope opposite in sign
on either side of the axis, the maximum negative lift coefficient is:
( )
max max,
0 2
l
o
l neg l
C C C − = ⋅ = α
and the prestall extrapolated lift coefficient is:
( ) ( ) ( )  
dat l l dat l l
C C C C
min min
α α α α α α − = − − = − = =
where α
mindat
is less than 0
o
and may correspond to a positive lift coefficient.
The drag bucket is not assumed to be symmetric about the α=0
o
axis, but the minimum
drag coefficient is assumed to lie near α=0
o
. This assumption is quite suitable in practice.
The extrapolation formula is as follows:
( ) ( ) ( )  
dat d d dat d d
C C C C
min min
α α α α α α − = − − = + = =
77
The lift and drag airfoil data behavior as functions of angle of attack are shown in Figure
67 (AB), respectively.
α
α
min,dat α
α
min,dat
C
L
α
α
min,dat
α
α
min,dat
C
D
(A) (B)
Figure 67 Angle of Attack Symmetry Diagram
A report by Sheldahl and Klimas [1981] provides the lift, drag, and pitching moment data
of symmetric airfoil sections through 0 < α < 180
o
at low Reynolds numbers. By
symmetry, the results can be applied to a full 360
o
angle of attack range. Only lift and
drag are considered here.
The numerical data was downloaded and analyzed. It was found that the poststall
behavior may be modeled as modified trigonometric functions. Unlike the prestall
angles of attack, the poststall lift and drag are functions of angle of attack alone, and are
insensitive to Reynolds number and thickness. A regression analysis was used to help
determine a curvefit to the poststall data. The results are shown in Figure 68 (AB).
PostStall Lift Behavior
1.5
1
0.5
0
0.5
1
1.5
0 90 180 270 360
Angle of Attack, Degrees
L
i
f
t
C
o
e
f
f
i
c
i
e
n
t
PostStall Drag Behavior
0.5
0
0.5
1
1.5
2
0 100 200 300 400
Angle of Attack, Deg
D
r
a
g
C
o
e
f
f
i
c
i
e
n
t
(A) (B)
Figure 68 Curve Fit for 360Degree Angle of Attack Sweep
78
The drag coefficient is:
( ) ( )
6729 . 1
sin 8332 . 1 α ⋅ =
d
C
If sin(2α) is greater than or equal to 0, then the lift coefficient is:
( ) ( )
3548 . 1
2 sin 0617 . 1 α ⋅ ⋅ =
l
C
Otherwise
( ) ( )
3548 . 1
2 sin 0617 . 1 α ⋅ ⋅ − =
l
C
After the onset of stall, the lift coefficients decrease and the drag coefficients increase
relative to their prestall values. To have the airfoil prediction code jump from the pre to
poststall values without any blending would create a discontinuity that may prove
troublesome for some design problems. It is therefore desirable to create some
smoothing function in a transition region. A sinusoidal function was selected for a
transition region of 5
o
angle of attack. The transition function for the upper stall is as
follows:
( ) ( )
2
2 5
sin 1
5
, ,

.

\

− ⋅
−
+
⋅ − + = + =
π
π
α α
α α
o
stall
stall l
o
stall l stall l l
C C C C
The lower stall function takes the same form. It is recognized that some airfoils have an
abrupt stall and that a transition function may not be the best representation for those
cases. However, the flexibility afforded by this methodology is considered ample
justification for any modeling inaccuracies of a phenomenon that is inherently rife with
uncertainty. The poststall data covers only symmetrical airfoils, and the effects of
camber are not known at this time.
The drag transition function takes the same form as the lift transition function.
The poststall subroutine is applied to both the directdata and parametric airfoil
interpolation subroutines. Results for the directdata interpolation subroutine are
presented in Figure 69 (AB). Note that the drag is on a logarithmic scale.
79
Drag Coefficient vs. Angle of Attack
0.001
0.01
0.1
1
40 20 0 20 40
Angle of Attack, Deg
D
r
a
g
C
o
e
f
f
i
c
i
e
n
t
Lift Coefficient vs. Angle of Attack
2
1.5
1
0.5
0
0.5
1
1.5
2
2.5
60 40 20 0 20 40
Angle of Attack, Deg
L
i
f
t
C
o
e
f
f
i
c
i
e
n
t
Figure 69 Example Drag polar and Lift Curve of an Airfoil Through a Large
Angle of Attack Sweep
6.4.2.3 Direct Airfoil Data Method
The direct airfoil data method interpolates airfoil lift and drag data taken from an airfoil
database. The database contains airfoils designed for a specific Reynolds number. In
other words, the airfoil set changes with reference Reynolds number. The current airfoil
database contents are listed in Table 64 and Table 65.
Reynolds
Number
Airfoil t/c
(%)
Camber
(%)
C
m0
Source
100,000 S1223 11.93 8.67 0.290 XFOIL
SD8020 10.10 0.00 0.000 Selig
S823 21.00 2.49 0.150 Selig
SD7032 9.95 3.66 0.099 Selig
MH46 11.35 2.85 0.0100 XFOIL
MH64 8.61 1.91 0.005 XFOIL
200,000 S1223 11.93 8.67 0.290 XFOIL
SD8020 10.10 0.00 0.000 Selig
S823 21.00 2.49 0.150 Selig
SD7032 9.95 3.66 0.099 Selig
MH46 11.35 2.85 0.0100 XFOIL
80
MH64 8.61 1.91 0.005 XFOIL
300,000 S1223 11.93 8.67 0.290 XFOIL
SD8020 10.10 0.00 0.000 Selig
S823 21.00 2.49 0.150 Selig
SD7032 9.95 3.66 0.099 Selig
MH46 11.35 2.85 0.0100 XFOIL
MH64 8.61 1.91 0.005 XFOIL
500,000 S823 21.00 2.49 0.150 Selig
MH46 11.35 2.85 0.010 XFOIL
MH64 8.61 1.91 0.005 XFOIL
MH78 14.4 2.6 0.050 XFOIL
MH20 9.02 2.1 0.010 XFOIL
NACA
2412
12.00 2. 0.0575 XFOIL
NACA
4412
12.00 4. 0.1065 XFOIL
Table 64 Low Reynolds number airfoil database.
81
Reynolds
Number
Airfoil t/c
(%)
Camber
(%)
C
m0
Source
1,000,000 MH46 11.35 2.85 0.010 XFOIL
MH64 8.61 1.91 0.005 XFOIL
MH78 14.4 2.6 0.050 XFOIL
MH20 9.02 2.1 0.010 XFOIL
NACA 0012 12.00 0.00 0.00 XFOIL
NACA 2412 12.00 2. 0.0575 XFOIL
NACA 4412 12.00 4. 0.1065 XFOIL
3,000,000 NACA 0012 12.00 0.00 0.00 XFOIL
NACA 2412 12.00 2. 0.0575 XFOIL
NACA 4412 12.00 4. 0.1065 XFOIL
9,000,000 NACA 0012 12.00 0.00 0.00 XFOIL
NACA 2412 12.00 2. 0.0575 XFOIL
NACA 4412 12.00 4. 0.1065 XFOIL
20,000,000 NACA 0012 12.00 0.00 0.00 XFOIL
NACA 2412 12.00 2. 0.0575 XFOIL
NACA 4412 12.00 4. 0.1065 XFOIL
Table 65 High Reynolds number airfoil database.
The accuracy of the direct airfoil data interpolation is only as good as the number and
type of airfoils in the database. Some of the provided airfoils represent different
technologies, such as standard versus natural laminar flow airfoils. The variety of airfoils
is important to ensure that the target airfoil is bound within the parameter space. The
direct data airfoil method was not used to generate the results of this research.
6.4.2.4 Parametric Airfoil Method
The parametric airfoil method uses empirical equations to generate smooth and adaptable
airfoil lift, angle of attack, and drag data. This method divides the drag into skin friction
and profile drag components, which facilitates the application of technology corrections
to each component. All equations are based upon XFOIL [Drela 2000] outputs over a
Reynolds number range of 60,000 to 20,000,000. Equations were generated for airfoils
of symmetrical, single camber, and reflexed camber sections.
This method differs from the direct airfoil data interpolation in another regard. The first
method uses the data for airfoils that may be designed for the Reynolds numbers of
interest. The empirical equation method models the behavior of a given airfoil over a
very wide range of Reynolds numbers. Since airfoils are designed for a relatively small
Reynolds range, most of the Reynolds numbers in the second method are not at the
design condition. Despite this limitation, the empirical equations are very useful for
describing the behavior of airfoils as a function of Reynolds number. The list of airfoils
for which empirical equations are written is provided in Table 66.
82
Airfoil t/c
(%)
Camber
(%)
C
m0
SD7032 9.95 3.66 0.099
MH18 11.14 3.25 0.015
NACA 0012 12.00 0.00 0.000
NACA 2412 12.00 2.00 0.0525
NACA 4412 12.00 4.00 0.103
Table 66 Empirical equation airfoil database.
Quite unexpectedly, a simple set of equations having a similar form can be used to model
the behavior of a wide range of airfoil parameters for many airfoil sections. For a given
airfoil, first level equations are functions of Reynolds number only. Second level
equations are functions of the first level equations and the lift coefficient.
The second level equations are as follows:
top tr
top l top l
top l l
top tr top tr
n
c c
c c
c x c x
,
, 1 , 2
, 1
max , , ,
2
cos / /


.

\

⋅
−
−
⋅ =
π
Upper surface transition location
bot tr
bot l bot l
bot l l
bot tr bot tr
n
c c
c c
c x c x
,
, 1 , 2
, 1
max , , ,
2
sin / /


.

\

⋅
−
−
⋅ =
π
Lower surface transition location
( ) ( )
db d l l dp d l l dp dp dp
c c c c c c c c c + − ⋅ ∆ + − ⋅ ∆ + =
6
min , 6
2
min , 2 0
Pressure drag
where
2
2
2 sin 1
1 2
1
0


.

\

− ⋅ ⋅
−
−
+
⋅ =
π
π
lb lb
lb l
db db
c c
c c
c c Laminar separation bubble drag
The factors and exponents given symbols on the right hand side of the second level
equations are functions of only Reynolds number for a given airfoil. The factors and
exponents described by first level equations are determined through an optimization
routine in Microsoft Excel, where the difference between airfoil data and the calculated
data is minimized up to stall. The forms of the first and second level equations were
selected from visual inspection of the behavior of the data.
The first level equations for the upper surface transition are as follows:
83
max , , , /
max , , 0 max , , max , ,
2 Re
Re
sin / / /
top tr c x
ref
top tr top tr top tr
n
c x c x c x


.

\

⋅ ⋅ ∆ + =
π
top cl
ref
top l top l top l
n
c c c
, 1
, 1 0 , 1 , 1
2 Re
Re
sin


.

\

⋅ ⋅ ∆ + =
π
max
2 Re
Re
sin
max 0 max max
l
c
ref
l l l
n
c c c


.

\

⋅ ⋅ ∆ + =
π
2 max
2 max 0 , 2 max 2 max
2 Re
Re
sin
r l
c
ref
r l r l r l
n
c c c


.

\

⋅ ⋅ ∆ + =
π
max 2 max , 2 l r l bot l
c c c ⋅ =
top tr
n
ref
top tr top tr top tr
n
n n n
,
2 Re
Re
sin
, 0 , ,


.

\

⋅ ⋅ ∆ + =
π
Excluding the maximum lift coefficient, which is evaluated in a separate optimization,
the optimization of the second level upper surface transition location formulation
involves twelve independent variables.
The first level equations for the lower surface transition are as follows:
0 . 1 /
max , ,
=
bot tr
c x
bot cl
ref
bot l bot l bot l
n
c c c
, 1
, 1 0 , 1 , 1
2 Re
Re
sin


.

\

⋅ ⋅ ∆ + =
π
bot cl
ref
bot l bot l bot l
n
c c c
, 2
, 2 0 , 2 , 2
2 Re
Re
sin


.

\

⋅ ⋅ ∆ + =
π
bot tr
n
ref
bot tr bot tr bot tr
n
n n n
,
, 0 , ,
2 Re
Re
sin


.

\

⋅ ⋅ ∆ + =
π
The lower surface transition location formulation optimization involves nine independent
variables, since the maximum transition location is set to 1.0 for all Reynolds numbers.
The first level equations for the pressure drag are as follows:
84

.

\

⋅ ∆ +


.

\

⋅ ⋅ ∆ + =
Re
000 , 100
2 Re
Re
sin
Re ,
0
0 00 0 dp
cdp
ref
dp dp dp
c
n
c c c
π
d cl
ref
d l d l d l
n
c c c
min
min , 0 min , min ,
2 Re
Re
sin


.

\

⋅ ⋅ ∆ + =
π
2
2 20 2
2 Re
Re
sin
cdp
ref
dp dp dp
n
c c c
∆


.

\

⋅ ⋅ ∆∆ + ∆ = ∆
π
6
6 60 6
2 Re
Re
sin
cdp
ref
dp dp dp
n
c c c
∆


.

\

⋅ ⋅ ∆∆ + ∆ = ∆
π
The optimization of the pressure drag formulation, excluding the bubble drag, involves
thirteen independent variables.
The first level equations for the laminar separation bubble drag are as follows:
0
,
0 00 0
2 Re
Re
sin
cdb
bub ref
db db db
n
c c c


.

\

⋅ ⋅ ∆ + =
π
1
,
1 10 1
2 Re
Re
sin
clb
bub ref
lb lb lb
n
c c c


.

\

⋅ ⋅ ∆ + =
π
2
,
2 20 2
2 Re
Re
sin
clb
bub ref
lb lb lb
n
c c c


.

\

⋅ ⋅ ∆ + =
π
Typically,
1
2 1
= =
clb clb
n n
and
00 0 db db
c c − = ∆
All parameters for the bubble drag are selected via manual inputs, since the optimizer
will otherwise exploit the parameters in areas where the bubble drag is not relevant. The
bubble drag term does not fully capture the bubble drag, but merely large deviations from
the polynomial drag growth characteristics of the pressure drag equation.
85
The above first level equations are applicable for a Reynolds number range of 60,000 to
the reference Reynolds number. After the reference Reynolds number is exceeded, the
values are held fixed at:
param param param ∆ + = 0
The reference Reynolds number for all first level equations except for the bubble drag is
20 million. The bubble drag reference Reynolds number is dependent upon the
characteristics of the airfoil.
Due to the nonlinear nature of the first level equations, the airfoil interpolation subroutine
interpolates the first level lefthandside (LHS) parameters versus the more numerous
righthandside (RHS) parameters. The following pseudocode shows the interpolation
process:
Param = 0.
Do i=1,nAirfoils
Param = Param+LHS(i)*Fact(i)
End Do
Where
∑
=
=
nAirfoils
i
i Fact
1
1 ) (
The interpolated LHS of the first level equations are used to evaluate the second level
equations.
The total drag of the airfoil is given by:
dp dfrict d
c c c + =
where the pressure drag is found directly from the above second level equation. The
friction drag is found using standard laminar and turbulent flat plate skin friction
equations found in Raymer [1992]. The overall skin friction coefficient for a surface that
transitions from laminar to turbulent flow is:
( )
fturb tr lam tr f
c c x cf c x c ⋅ − + ⋅ = / 1 /
The airfoil skin friction coefficient is the summation of the upper and lower skin friction
coefficients.
86
NACA 2412 Upper Transition
0.2
0
0.2
0.4
0.6
0.8
1
1.2
0.5 0 0.5 1 1.5
CL
x
/
c
T
r
a
n
s
i
t
i
o
n Re
NACA 2412 Lower Transition
0
0.2
0.4
0.6
0.8
1
1.2
0.5 0 0.5 1 1.5
CL
x
/
c
T
r
a
n
s
i
t
i
o
n
Re
(A) (B)
Figure 610 Upper and Lower Transition Location vs. Lift Coefficient Comparison
Between XFOIL and Empirical Equation Matching for the NACA 2412 Airfoil
Figure 610 (AB) show the correlation between the XFOIL and parametric boundary
layer transition locations as a function of Reynolds number and lift coefficient. The lines
without markers represent the curve fit, and the lines with markers represent XFOIL data.
Agreement is quite good for most Reynolds numbers.
Figure 611 The Pressure Drag Comparison Between XFOIL and the Empirical
Equation Matching for the NACA 2412 Airfoil
Figure 611 shows the correlation between XFOIL and parametric pressure drag
coefficients over a large range of Reynolds numbers and lift coefficients. The laminar
separation bubble drag occurs at lower Reynolds numbers. Correlation is quite good for
this case as well.
87
Direct Airfoil Data Interpolation
0.000
0.005
0.010
0.015
0.020
0.025
0.030
0.035
0.040
0 0.5 1 1.5 2
CL
C
D
Parametric Airfoil Interpolation
0.000
0.005
0.010
0.015
0.020
0.025
0.030
0.035
0.040
0 0.5 1 1.5 2
CL
C
D
(A) (B)
Figure 612 Comparison of Drag Polars Between the Direct Data and Parametric
Airfoil Interpolation Methods
Figure 612 (AB) shows the difference in drag polars for a general airfoil between the
direct airfoil data interpolation (method 1) and the parametric airfoil interpolation
(method 2). The magnitude of the minimum drag is in close agreement, but the profile
drag grows more slowly with method 2. This is caused by three factors. First, the
database for method 1 is presently populated by more airfoils at low Reynolds numbers.
Second, method 1 uses wind tunnel data at low Reynolds numbers that has a stronger
pressure drag rise than XFOIL data. Third, the interpolation of first level equation results
in method 2 has some nonlinear effects in the second level equations.
(A) (B)
Figure 613 Comparison Between XFOIL and Wind Tunnel Data.
88
Figure 613 (AB) show the correlation between XFOIL and wind tunnel data for
cambered and symmetrical airfoils. Figure 613 (A) shows the correlation for a Selig
Donnovan SD 7032, and Figure 613 (B) shows results for a NACA 0009. All wind
tunnel data for Reynolds numbers of 300,000 and below come from Selig's low speed
wind tunnel tests at University of Illinois or Princeton [Selig 1989]. The high Reynolds
number runs for the NACA 0009 are from the NACA tests as reported in Abbot and Von
Doenhoff [1959]. Both series of tests describe the tunnel conditions as 'low turbulence'.
According to the XFOIL manual [Drela 2000], the e
N
turbulence model corresponds to an
average turbulence wind tunnel when N
crit
= 9.
In general, the XFOIL results tend to underestimate the drag measured in a wind tunnel.
Correlation suffers with increasing camber, as can be seen on the SD 7032 graph in
Figure 613 (A). XFOIL does capture the laminar separation bubble effects on pressure
drag, though the behavior does differ.
Decreasing the value of N
crit
does improve the correlation with the available wind tunnel
data, but the wind tunnel data does not necessarily correspond to the conditions
encountered during flight. A N
crit
value of 9 is low compared to the flight conditions
encountered by some aircraft types. Due to these considerations, N
crit
is set to 9 for all
XFOIL runs, even if the results do not match the wind tunnel data most closely.
The parametric airfoil method was used to generate all results for this research effort.
6.4.2.5 Simplified Airfoil Method
The two detailed airfoil methods both require a database and a large number of
calculations. In order to reduce memory requirements and to speed computation time, a
simplified airfoil method was developed. This is close to the more traditional methods
employed in conceptual design. The airfoil profile drag takes the form:
( )
2
0 l l dfrict d
C C K FF C C − ⋅ + ⋅ =
Special functions were created for the minimum drag lift coefficient, C
l0
, and the
parabolic parameter, K, using curve fits to airfoil data and XFOIL results. While the
detailed airfoil methods were under development, the simplified airfoil method was used
to satisfy the profile drag needs. This method is available within the code, but it was not
used to generate any results for this research.
6.4.2.6 Laminar Flow Technology
The impacts of 2D laminar flow due to design level and laminar flow influence on
transition are covered. The 2D transition multiplication factor, TR
2D,Fact
, is found
using:
89
( ) 5 . 0 2 1
, 2
− ⋅ + =
− Lam Fact D
Tech TR
The TR
2D,Fact
transition multiplication factor behavior described by this equation is
assumed in the absence of other models. Future research could investigate laminar flow
improvements relative to a baseline airfoil through experimental or CFD efforts to further
develop this function. The transition factor is multiplied by the upper and lower x/c
transition locations for the parametric airfoil model. For other airfoils, the ratio of the flat
plate skin friction drag for an assumed fixed transition location multiplied by the
transition multiplication factor and the flat plate skin friction drag for an assumed fixed
transition location is multiplied by the profile drag.
The effect of sweep on transition is modeled by a curve fit to the laminar flow versus
sweep technology curve used by Grasmeyer [1998]. The slope of the curve is:
( ) ( )
Lam
Tech Slope − ⋅ + ⋅ = 5 . 0 4 . 0 1
15
9
The transition multiplication factor due to sweep is:
π
60
1
4 / ,
⋅ Λ ⋅ − =
c Fact Sweep
Slope TR
This factor is applied to the profile drag just like the 2D factor.
6.4.3 Interference Drag
The interference drag model is built upon that used by Grasmeyer [1998] on the SBW
project, which is a modified version of the Hoerner [1990] method. The interference
drag is broken down into several components that can be summed. The total interference
drag is:
( )
∑
+ + + ⋅ =
incline D sweep D lift D wall D FACT D D
C C C C C C
int, int, int, int, int int
C
Dint,FACT
is a multiplication factor to account for improvements possible with a good
design. Hoerner [1990] suggested that 0.1 is a suitable value.
The interference drag of a straight wing perpendicularly intersecting a flat wall with no
lift is estimated by Grasmeyer [1998] as:
( )
ref
wall D
S
c
c t C
2
3
int,
0003 . 0 / 8 . 0 ⋅ − ⋅ =
The interference drag increment due to lift is estimated by Grasmeyer [1998] as:
90
ref
l lift D
S
c
C C
2
2
int,
1 . 0 ⋅ ⋅ =
The interference drag increment due to wing sweep relative to the flat wall is estimated
by Grasmeyer [1998] as:
( )
ref
c c sweep D
S
c
C
2
4 /
2
4 / int,
0009 . 0 00018 . 0 ⋅ Λ ⋅ + Λ ⋅ − =
The interference drag increment due to inclining the wing relative to the flat wall is
estimated by Grasmeyer [1998] as:
( )
ref
incline incline incline D
S
c
C
2
2
int,
0015 . 0 00006 . 0 ⋅ ⋅ + ⋅ = θ θ
The Hoerner [1990] formulation as modified by Grasmeyer [1998] has limitations. The
flat wall assumption may not hold for fuselages of small diameter. Fuselage overall size
never enters into the formulation. In other words, a fuselage length equal to the chord
length will produce the same interference drag as a fuselage a hundred times this length.
Detailed geometry inputs are required for this formulation, but the overall accuracy is
questionable.
The sweep interference drag formulation is suspect because it produces negative drag
coefficients at moderate sweep angles.
The interference drag of two intersecting wings is not included in this formulation.
θ
incline
θ
incline
Figure 614 WingFuselage Interference Geometry
Determining the interference geometry is not trivial. Figure 614 shows the basic wing
fuselage interference geometry. An algorithm determines which wing segments
intersected each fuselage. Once the intersection is confirmed, the spanwise point on the
wing panel that intersects the fuselage and the height along the centerline of the fuselage
are determined. The height of the intersection along the fuselage is translated sideways
91
to the right and left, but the wing dihedral angle is preserved. A number of fuselage
points are defined within the conic section formulation, and straight lines between these
points are assumed to represent the fuselage adequately. The difference between the
wing dihedral angle and the slopes of the fuselage walls is assumed to be the inclination
angle. If the wing ends in the fuselage, then the inclination angle is taken only for the
relevant side. Otherwise, if the wing passes through the fuselage, then the average of the
right and lefthand side inclination angles is used.
The above geometry calculations are performed on the leading and trailing edges. The
amount of span covered by the fuselage for the leading and trailing edges are averaged
and multiplied by the chord to estimate the amount of covered wing planform area. This
area is divided by the total planform area of the given surface to provide a profile drag
multiplication factor. This factor is always less than unity, so it can be considered a
constructive interference drag.
6.4.4 Fuselage Drag
The fuselage drag subroutine calculates the drag of generic bodies ranging from slender
fuselages to spheres to flat plates normal to the freestream. The fuselage cross section
may be circular or have squared edges at the higher fineness ratios. The sphere and
slender fuselage formulations are smoothly blended together.
The general equation for fuselage drag is:
( )
ref
ref
press D frict D Fuse D
S
S
C C C
1
, , ,
⋅ + =
For a slender fuselage this formulation becomes:
( )
ref
Fuse wet
Fuse frict D Fuse D
S
S
Penalty FF C C
,
, ,
⋅ ⋅ ⋅ =
∑
where the pressure drag is accounted for by the form factor and the penalty factors.
The form factor equation presented in Raymer [1992] for circular cross sections is
applied.
92
Fuselage Drag Cross Section Penalty
0.6
0.8
1
1.2
1.4
1.6
0 0.2 0.4 0.6 0.8 1
Rho
P
e
n
a
l
t
y
F
a
c
t
o
r
Figure 615 Fuselage Drag Cross Section Penalty
Not all fuselage sections of interest are circular in cross sections. Ritchie [1968],
provided drag penalties for cross sections ranging from ovals to rectangles. This data
was modified to correspond with the conic parameter, ρ (rho). Figure 615 shows the
penalty factors. The ρ value associated with a circular cross section is 0.414. The
volumeaveraged ρ value for the fuselage is used to find the penalty through interpolation
of the data.
Fuselage Drag Maximum Diameter
Location Penalty
0
1
2
3
4
5
6
7
8
9
0 0.2 0.4 0.6 0.8 1
Maximum Diameter Location, X/L
P
e
n
a
l
t
y
F
a
c
t
o
r
Figure 616 Fuselage Drag Maximum Diameter Location Penalty
Ritchie [1968] provides drag penalty data associated with the location of the maximum
equivalent diameter location of the fuselage. This data is presented in Figure 616. The
penalty can be quite high for fuselages with the maximum equivalent diameter located
too far forward or too far aft.
93
Unfortunately, the thickness location function was exploited by the optimizer to
artificially drive sweep. Because the wing was affixed to the fuselage by a parametrically
blended wing fairing, the wing root would shift forward to align the fairing with the
radome of some configurations. The aerodynamic center is fixed on the fuselage in the
parametric geometry, so the wing would sweep. The wave drag is the realworld driver
for wing sweep. The artificial wing sweep justified setting the factor equal to unity at all
times.
The drag coefficient for a sphere is found by:
ref
frontal
sphere D fuse D
S
S
C C ⋅ =
, ,
where the friction and pressure drag are accounted for in C
D,sphere
. Note that the sphere
drag coefficient is referenced to the sphere frontal area, so the ratio of the sphere frontal
area to the wing reference are is applied to be consistent with other drag contributions.
The sphere drag coefficient is 0.41 when below a Reynolds number of 400,000 and 0.15
when above.
A flat plate is assumed to have a drag coefficient of 1.18 referenced to the frontal area
regardless of shape.
The three cases are blended together as a function of length to diameter ratios. Above a
L/D
Fuse
value of 2, the slender fuselage formulation is used. Between 2 and 1, the
solution is linearly blended between the slender fuselage and sphere. Between 1 and 0,
the sphere and flat plate are blended together.
6.4.5 Wave Drag
The wave drag formulation is the same method used by Grasmeyer [1998], which is
based on the Korn equation. The drag divergent Mach number is found for a number of
panels through the following equation:
( ) ( ) ( ) ) ( cos 10
) (
) ( cos
) ( /
) ( cos
) (
2 /
3
2 /
2
2 /
i
i Cl
i
i c t
i
i M
c c c
a
DD
Λ ⋅
−
Λ
−
Λ
=
κ
where the airfoil technology factor, κa, is estimated by Grasmeyer [1998] as 0.87 for a
NACA 6Series section and 0.95 for a supercritical section. The selection of the airfoil
technology factor is described in the technology modeling chapter. The critical Mach
number is defined found by:
3 / 1
80
1 . 0
) ( ) ( 
.

\

− = i M i M
DD crit
94
If the flight Mach number is above the critical Mach number, then the local wave drag
coefficient is:
( )
3
,
20 ) (
crit wave d
M M i C − ⋅ =
and the total wave drag coefficient is:
∑
⋅ =
ref
segment
wave d wave D
S
S
i C C ) (
, ,
If the flight Mach number is below the critical Mach number, then the wave drag is zero.
The wing wave drag is assumed to become a major driver at a lower Mach number than
the fuselage. Since the wave drag increases so rapidly with Mach number, the impacts of
fuselage wave drag in the lower transonic region are considered to be unimportant in the
determination of cruise Mach number. Therefore, the fuselage wave drag is not
incorporated into the formulation.
6.4.6 Landing Gear Drag
The landing gear drag is broken into two phases of flight: cruise and landing/takeoff.
The cruise condition has all the gear in the full retraction position, if the configuration has
retractable gear. The landing/takeoff condition has the gear fully extended.
The gear drag is composed of combinations of tires, struts, streamwise doors, and body
cavities. This geometry is passed to the gear drag subroutine prior from the complex
geometry definition subroutines. The landing gear drag coefficient is defined as:
( )
∑
+ + + =
cavity D door D strut D tire D gear D
C C C C C
, , , , ,
The tire drag is estimated for the nose and main gear by:
ref
tire tire
strut strut tire tire D
S
W D
N N C
⋅
⋅ ⋅ ⋅ =
/ ,
25 . 0
where the size of the nose and main gear may be different. Note that 0.25 is a typical
frontal area based drag coefficient. Hoerner [1990] shows a variation in drag coefficient
between 0.12 and 0.35, depending on tire type. The above drag coefficient is then
modified by a multiplication factor for rear fairing, forward fairing, or wheel pants cases,
based upon factors derived from Hoerner [1990].
95
The strut drag is treated as a circular cylinder perpendicular to the freestream or as a
faired strut. The circular cylinder drag coefficient varies with Reynolds number,
assuming natural transition, with the critical Reynolds number centered about 3x10
6
.
The faired strut assumes a t/c of 0.33, and the simple airfoil method is employed for
calculating the drag.
If retractable landing gear is used, then gear door drag is calculated for the extended gear
case. The gear doors are assumed aligned with the flow. The equivalent flatplate
turbulent skin friction on the door is calculated on both sides to yield the drag coefficient.
For retractable landing gear, the gear bay cavity drag is calculated for the extended gear
case. The drag of the open gear bay cavity is found in Hoerner [1990]. The drag
coefficient is:
ref
door door
strut cavity D
S
L W
N C
⋅
⋅ ⋅ = 0083 . 0
,
where the value for the nose and main gear well cavities may be different.
The total drag for the up and down position depends on the gear configuration. If a gear
is retractable, then the retracted drag coefficient is zero and the extended drag includes
the tire, strut, door, and cavity. If the gear is permanently extended, then both drag
values include the strut and tire, but not the gear well cavity or the gear door. The
retraction scheme can be different for the nose and main gears.
6.4.7 Cooling Drag
The cooling drag associated with the engine and systems is book kept as in the propulsion
section. The cooling can be accomplished from an environmental control system (ECS)
driven by bleed air from the engine or by a ram air ECS system. Regardless of the cool
air source, the engine and systems require a cooling mass flow.
For ram air ECS systems, the total drag associated with cooling is found by:


.

\

− ⋅ ⋅ =
∞
∞
U
U
U mdot D
e
Ram Ram
1
The ram air mass flow rate (kg/hr) [Sloan 1985] is found by:
air comp
cool
Ram
T T
Q
mdot
−
⋅ = 59 . 3
96
where the temperatures are in degrees Celsius, and the heat load is in Watts. To put the
ram cooling equation in convenient units, the following is attained:


.

\

− ⋅ ⋅
−
⋅ =
∞
∞
U
U
1 U
T T
Q
0.0012686 D
e
air comp
cool
Ram
where the drag is in pounds, the heat load is in Watts, the temperatures are in degrees
Fahrenheit, and the velocity is in knots. The velocity recovery, U
e
/U
∞
, is assumed to be
only 0.2 due to pressure losses in the ECS system.
The cooling drag for an air cycle machine ECS system is assumed to be the same as a
ram air ECS system for the purposes of this code. A fuelcooled ECS system, like the
one used on Global Hawk, is assumed to have no drag penalties.
6.4.8 Miscellaneous Drag
There are inevitably many drag contributors that are not accounted for in the above
formulations. A userinput miscellaneous drag factor can be applied to scale the drag
polar. This factor is a direct multiplier, so it scales the minimum drag value and the drag
due to lift by the same ratio. In other words, the factor acts as a simultaneous minimum
drag shift and increases the parabolic behavior of the drag polar. A 5% miscellaneous
drag is used for all cases.
97
Chapter 7 Propulsion
7.1 Propulsion Overview
The propulsion code evaluates the performance of turbofans, turboprops, and piston
props. Each engine type has specialized analysis methods. The primary output of the
propulsion code is an engine deck, which is used by the performance analysis.
The engine deck has a common format, because the performance code must adapt to
multiple different engine types without modifying its formulation. The burden falls on
the propulsion methods to generate this single format deck. The engine deck consists of
the following arrays:
Altitude(alt
min
:alt
max
)
Velocity(alt
min
:alt
max
, vel
min
:vel
max
)
Thrust(alt
min
:alt
max
,vel
min
:vel
max
, throt
min
:throt
max
)
TSFC(alt
min
:alt
max
, vel
min
:vel
max
, throt
min
:throt
max
)
Allow(alt
min
:alt
max
, vel
min
:vel
max
, throt
min
:throt
max
)
The onedimensional altitude array varies from the minimum altitude, typically sea level,
to a specified maximum altitude for the engine deck. The twodimensional velocity array
varies with altitude and velocity setting. The threedimensional thrust, Thrust Specific
Fuel Consumption (TSFC) and allowable arrays vary with altitude, velocity, and throttle
setting. The thrust specific fuel consumption is in jet engine format. The allowable array
indicates whether or not a given point in the engine deck violates a propulsion constraint,
such as being outside the permissible altitude, velocity, or thrust bounds.
The engine is scaled as a rubber engine, meaning that many engine characteristics are
independent of scale. The jet thrust to weight and specific fuel consumption are assumed
to stay constant. The piston engine power to weight ratio and brake specific fuel
consumption remain constant. The optimizer freely scales the engines to meet the
demands of a given solution. Scaling engines is a bold assumption. This implies that the
appropriate engine is either available or can be produced in time to meet the UAV
development schedule. This may not always be reasonable due the high cost of engine
development and the typically fastpaced UAV development schedule. Despite the real
world limitations, the rubber engine assumption is applied here.
7.2 Code Architecture
The propulsion code is broken into multiple subroutines based upon propulsion type.
Figure 71 shows the top level propulsion code architecture. The initial propulsion data
is retrieved from PROPDATIN which is called by INFILES. ACMAIN calls INFILES
prior to any optimizations on analyses. ACMAIN calls EVAL, which then calls
98
PROPULSION. PROPULSION can call the propulsion category subroutines: GASPROP
for piston and turboprops, and TURBOFAN for turbojets and turbofans.
Figure 71 Propulsion Code Architecture
GASPROP combines pistonpropellers and turboprops because of a high degree of
commonality. The first step is to calculate the propeller diameter in PROPDIAM using
empirical formulas for a specified design condition. If a complex propeller design
method is chosen, then the propeller is designed in PROPDES. The tabular engine deck
is then calculated. A loop is entered where the altitude, velocity, and thrust levels
(throttle) vary. The propulsion system is evaluated at each point. The atmospheric
conditions are found by standard atmosphere routine, STDATM. The engine power and
fuel consumption characteristics are found by PISTALT for piston engines and by
TURBORATIO for turboprops. The propeller efficiency is found by PROPEFF if the
disk actuator model is used or by PTHRUST if the complex propeller analysis method is
used.
The TURBOFAN subroutine creates engine decks for turbojets and turbofan propulsion
systems. The engine deck loop is then entered. The jet engine thrust and fuel
consumption characteristics at each altitude, velocity, and throttle are found via the
TURBORATIO subroutine. The JETON detailed jet ondesign performance subroutine is
integrated but was not used here.
7.3 Propeller Methods
Common propeller design and evaluation methods are used, because pistonprop,
electricprop, and turboprop propulsion systems all utilize propellers. Two propeller
analysis methods are employed. A simple disk actuator method provides corrected
99
theoretical efficiencies with little computation time. A detailed blade element model
designs propellers and provides highly fidelity propeller analysis outputs. The propulsion
code has the ability to size propeller diameters based on empirical formulas.
Unfortunately, the blade element model is computationally expensive relative to the disk
actuator model. Therefore, it is only utilized for determining supportable correction
factors for the faster disk actuator model. The blade element model is fully integrated
into the code, and is available for use. Appendix B describes the blade element
methodology in detail.
7.3.1 Propeller Diameter Sizing
Raymer [1992] provides empirical equations for sizing propellers as a function of
maximum engine power and the number of blades. These equations are intended for
manned aircraft with two or three blades. UAV designs frequently make more extensive
use of gearing than manned aircraft to achieve large propeller diameters with high
relative efficiency. The Predator and Shadow 200 designs appear to fit within the
application of the diameter estimation. Both of these cases use twoblade propellers.
Despite the limitations, the Raymer methods are used when the propeller diameter is not
specified. Figure 72 shows results of the Raymer empirical equations.
Propeller Diameter vs. Power
0
2
4
6
8
10
12
0.01 0.1 1 10 100 1000
Engine Power, Horsepower
P
r
o
p
e
l
l
e
r
D
i
a
m
e
t
e
r
,
F
e
e
t
nb=2
nb=3
Data Pts
Figure 72 Raymer Propeller Sizing Equation Results
The second method for sizing the propellers is scaling from known propellerengine
combination. The Raymer [1993] formulation is used to aid in the scaling.
( )
( )
Ref func
func
f
BHP D
BHP D
D D ⋅ =
Re
100
7.3.2 Disk Actuator Model
Two propeller performance models were developed for this effort: 1) a disk actuator
model, and 2) a blade element model. Only the disk actuator model was applied here to
generate the results. Details of the blade element model can be found in Appendix B.
The disk actuator model is a theoretical method of predicting the ideal efficiency of a
propeller. The disk swept out by the propeller is assumed to act uniformly on the air that
passes through it. Details of the method are well documented [Mattingly 1987] and will
not be reproduced here. Important limitations of the model are that viscosity, non
uniform flow characteristics, propeller hubs, and other relevant aspects of propellers are
neglected. Importantly, the disk actuator model cannot predict static thrust without
significant modifications. The static case is not considered in this research. Corrections
were applied to the theoretical model to provide more realistic predictions.
The relevant thrust equation is:
( ) V mdot V V mdot T
e
∆ ⋅ = − ⋅ =
For a propeller, the thrust equation may be written in terms of the propeller disk area, A
p
:
( ) V V V A T
p
∆ + ⋅ ⋅ ∆ ⋅ ⋅ ⋅ = 2
2
1
ρ
Solving for the velocity difference yields:
2
8
4 2
2
p
A
T
V V
V
⋅
⋅
+ ⋅ + ⋅ −
= ∆
ρ
The exit velocity, V
e
, is simply the sum V+∆V. The theoretical efficiency is:
1
2
1 2
,
+
⋅
∆
=
−
⋅
=
V
V
V V
V
e
ideal p
η
With the above equations, the ideal efficiency can be found for a given propeller diameter
at any altitude, airspeed (except static case), or thrust condition. This powerful result is
achieved at little computational expense. However, inescapable practical inefficiencies
are neglected in the disk actuator model. To remedy this deficiency, a nonideal
efficiency correction is included. The total propeller efficiency of a nonideal propeller
is:
nonideal p ideal p p , ,
η η η ⋅ =
101
where the nonideal efficiency ranges from approximately 8595% based on comparisons
with the blade element model. A nominal value of 90% is applied.
The propeller thrust can be calculated as a function of shaft horsepower:
V
P
T
p shaft
η ⋅
=
The thrust is a function of the propeller efficiency. The propeller efficiency is a function
of the nonideal propeller efficiency corrections and the ideal efficiency from the disk
actuator model. Therefore, the thrust is a function of itself. The ideal efficiency is found
iteratively. The initial value of ideal efficiency is set to unity for the starting conditions.
The engine shaft power to produce a given amount of thrust is:
gear p
shaft
V T
P
η η ⋅
⋅
=
When the flight velocity is zero, the propeller efficiency is zero, and the power absorbed
by the propeller cannot be realistically determined by the disk actuator theory.
A comparison is made in Figure 73 between the blade element analysis and the ideal
propeller using the disk actuator model. The ideal efficiency plotted below corresponds
to the thrust output at each given advance ratio. The ideal efficiency increases as the
thrust level decreases over the constant propeller disk area. In the sample case, the blade
element design code comes within 710% of the ideal theoretical efficiency. Note that
the ideal efficiency neglects viscosity and other considerations. A 90% efficiency factor
is applied to the disk actuator model based on these results.
102
Ideal and Predicted Efficiency vs.
Advance Ratio
0.6
0.65
0.7
0.75
0.8
0.85
0.9
0.95
1
0.2 0.4 0.6 0.8 1
Advance Ratio J
E
f
f
i
c
i
e
n
c
y
Predicted
Ideal
Figure 73 Comparison Between Predicted and Ideal Propeller Efficiencies
7.4 PistonProp Engines
The pistonprop method generates an engine deck for piston engines that drive a
propeller. Although the word piston is used, the method applies to rotary engines as well.
A normally aspirated engine relies upon ambient pressure and suction to bring air into the
cylinders. This type of engine generates its maximum power at sea level, and
performance degrades as altitude increases. The GaggFerrar [Stinton 1995] equation
describes this behavior:
( )
55 . 7
1
) (
0
σ
σ σ
−
− = =
BHP
BHP
F
h
BHP
where the subscript h indicates the conditions at flight altitude and 0 indicates sea level
conditions. The density ratio is defined as:
0
ρ
ρ
σ
h
=
The GaggFerrar equation is used for all rotary and reciprocating engines in this code.
Engine power can be restored by means of turbo or supercharging, where the pressure of
the intake air is increased by mechanical means. This helps aircraft that fly in hot, humid
103
conditions or at high altitudes. Neither Predator nor Shadow 200 are known to include
turbo or superchargers, so the potential improvements made possible by these devices
were not investigated here.
The specific fuel consumption for a piston engine is defined as the ratio of fuel flow to
the power, and is usually given in the units lb/HPhr. Mises [1945] wrote that the typical
engine specific fuel consumption ranged from 0.450.55lb/HPhr. It can be assumed that
he was speaking of fourstroke engines for manned aircraft applications. The situation
has hardly improved since 1945 due to the domination of jet engines. The applied
technology trends for reciprocating engines are found in the technology chapter.
The density altitude function F
BSFC
(σ) is estimated by:
( )
( ) C
C
F
n
BSFC
−
− ⋅
=
σ
σ
σ
σ
1
) (
where C is about 0.065 and nσ is 1.117. The density altitude function F
BSFC
(σ) may be
applied directly to the sea level BSFC of a known engine:
0
) ( BSFC F BSFC
BSFC h
⋅ = σ
The above equation is used for both reciprocating and rotary engines in this code.
Figure 74 shows the specific fuel consumption as a function of density ratio for a typical
naturally aspirated internal combustion engine.
Figure 74 Density Ratio Impacts on Brake Specific Fuel Consumption
104
7.4.1 PistonProp Engine Deck
The throttle setting on a piston engine regulates the engine power directly, not the thrust.
The power available to the propeller is:
( ) ( )
gear BHP
F P P throt P P η σ ⋅ ⋅ − ⋅ + = ) (
min , 0 max , 0 min , 0
The power available to the propeller is assumed to be equal to the power absorbed by the
propeller. This power is passed to either the disk actuator or blade element analysis
codes to find the associated thrust and overall propeller efficiency. The engine power is
not stored in the engine deck, only the thrust value.
The Brake specific fuel consumption is:
gear p
BSFC
h
F BSFC
BSFC
η η
σ
⋅
⋅
=
) (
0
The engine deck for fueled engines must be written in terms of thrust specific fuel
consumption and not brake specific fuel consumption. The conversion for piston engines
is as follows:
550 3600
6080
⋅
⋅ ⋅ = V BSFC TSFC
h
where V is in knots, and BSFC
h
is in pounds mass per pounds force per hour.
7.5 Turbojets and Turbofans
Turbopropulsion covers turbojets, turbofans, and turboprops. Complex design methods
and simplified offdesign performance estimation codes were written for each engine
type.
The objective of this research is to estimate the performance of UAVs and to predict the
impacts of technology trends on future UAV designs. Unfortunately, technology trends
are not available for many jet engine components, only the jet engines as a whole.
Because of this, it is fruitless to try to surmise how the technology improvements are
divided up into multiple components. The sensible approach is to apply the gross
technology trends to simple methods. Simple offdesign relationships were established to
evaluate how gross engine characteristics vary with operating conditions. The ondesign
analysis is available to the code, but it was not used here.
The end result is an engine deck composed of specific fuel consumption versus thrust
level, altitude and Mach number.
105
7.5.1 Turbojet/Turbofan Engine Decks
The looping structure for the turbojet/turbofan engine deck model is similar to the piston
prop. There are altitude, velocity, and throttle loops. The throttle loop regulates the
thrust. First, the maximum thrust at the given altitude and Mach number is determined
using modified Mattingly [1987] equations:
( )
6 . 0
3
2 . 1 25 . 0 568 . 0 σ ⋅ − ⋅ + = M a for high bypass ratio (assumed bypass = 7)
( )
7 . 0
4 . 1
4 . 0 245 . 0 88 . 0 σ ⋅ − ⋅ + = M a for low bypass ratio (assumed bypass = 2)
( )
7 . 0
5 . 1
5 . 0 262 . 0 907 . 0 σ ⋅ − ⋅ + = M a for high bypass ratio (assumed bypass = 0)
The actual thrust ratio is found through a linear interpolation between the thrust ratios
based on bypass ratio. The maximum thrust at a given altitude and velocity is found by:
a T T
SLS
⋅ =
max, max
The maximum sea level static installed thrust is provided in an input file or is calculated
within the code based upon a thrust to weight ratio or other parameter. The minimum
installed thrust in flight is assumed to be 5% of the maximum thrust. The throttled thrust
is found by:
( ) ( )
min max min
T T throt T T − ⋅ + =
The thrust specific fuel consumption is found using an equation modified from
Grasmeyer’s [1998] SBW code. This equation is calibrated to work well for high bypass
ratio jet engines, though it also works well for low bypass ratio engines and turbojets:
( )
ref
ref
SL
rat
TSFC
M TSFC
Temp
Temp
TSFC
⋅ +
⋅


.

\

=
4021 . 0
4704 . 0
The reference thrust specific fuel consumption is 0.5 for a bypass ratio of 7.0, 0.65 for a
bypass ratio of 2.0, and 0.85 for a bypass ratio of 0.0. The reference value used in the
above equation is interpolated within these values by bypass ratio. The bypass ratio at a
given altitude and velocity is found by:
SLS rat
TSFC TSFC TSFC ⋅ =
106
The best installed sea level static thrust specific fuel consumption is provided in an input
file.
107
Chapter 8 Size, Weight, and Power
8.1 Size, Weight, and Power Overview
A systems approach to AV design is essential, since Size, Weight, and Power (SWaP) of
all components comprising the aircraft have significant impacts on the overall sizing.
The size, as considered here, refers to the component volume. Weight considers both
component weights and installation weights. Power contains both power consumed by
the component and the cooling power that must be dissipated. These considerations must
be addressed for every component on the aircraft whether it falls under structures,
avionics, subsystems, propulsion, payload, or fuel categories. A flexible means of
evaluating SWaP is provided.
8.2 Generic Systems
8.2.1 Generic Systems Overview
The systems approach to aircraft design is essential for UAV applications. There simply
is no way to contend with such a wide range of vehicle sizes and missions using
empirical methods alone. The systems method is a core strength of the design code.
Consideration of SWaP must be given to every component that enters the air vehicle.
Component and type attributes have been assigned to every component that enters the
aircraft. The categories and components comprising the categories are shown in Table
81.
Avionics Subsystems Structures Propulsion Payload
Autopilot
INS
GPS
Processor
Camera
ADS
Recorder
ATC
FCS
ECS
Communications
Electrical System
Fuel System
APU
Generic Struct.
Wing
Tails
Fuselage
Nacelles
Landing Gear
Generic Prop.
Recip. Engines
Jet Engines
Generic Payload
EO/IR
SAR
Table 81 Component Categories and Types
Multiple options are provided for SWaP. In general, the user may input values directly or
select from multiple calculation options. Many component types have specialty input
subroutines, and typespecific methods can be entered if available. Though complex, this
approach is intended to maximize flexibility over a wide range of problems where
108
information is often difficult to obtain. Default information for all component categories
are input through the defaults file, which may be edited by the user.
8.2.2 Generic Systems Weights
The design code is intended to offer multiple means of calculating weights to enhance
flexibility. Weight is divided into two categories: component weight and component
installation weight.
Standard methods for calculating the component weights include direct weight input,
weight as a fraction of the empty weight, and weight as a fraction of the takeoff gross
weight. When specialty methods are available and selected, the code calls the specialized
weight subroutines for the components. Default component weight categories for all
component types are set in the code, but these may be edited by the user.
Standard methods of calculating the installation weights include direct input, a default
fraction of the uninstalled weight, and input fraction of the uninstalled weight. The
default installation weight factor is a function of the component category and level of
design. Default component installation weight categories for all component types are
input through the defaults file, which may be edited by the user. The default installation
weight factor is found by:
( )
Max Install Wt Min Install Wt Design Max Install Wt Install Wt
F F F F F
, , , , , , ,
− ⋅ + =
where the installation factor is set by component category. Typically, the minimum
installation weight factor is 0.15 and the maximum is 0.30 in this research. SAWE
[1996] indicates that the installation weight factor usually ranges from 0.250.5, but this
range was modified as stated to attain sensible calibration results. The installation weight
factor is independent of technology year. Installation weight does not apply to fuel, since
the fuel system handles all fuel installation considerations. The default structural
installation weight factor is set to zero. The installation weight is found by:
Install Wt d Uninstalle Install
F Wt Wt
,
⋅ =
An important aspect of the component weights estimation is determining the location of
the weight. In particular, weight in the wing will provide inertia relief that affects wing
weight if the piecewise linear beam model is utilized. The location also affects volume
margin, as described below. The component location is determined by the user while
setting up the input files. A subroutine guides the inputs to ensure the user selects only
existing wings, tails, fuselages, and nacelles.
109
8.2.3 Generic Systems Size
Size, represented here as volume, considerations are very important for many UAV
applications. As with manned aircraft, many components on an AV compete for similar
locations on the vehicle. A vehicle sized as a point mass may be smaller and lighter than
a vehicle that is designed taking volumetric details into account. Volumetric
considerations constrain the design in a realistic manner.
There are multiple, userselectable means of calculating component volumes. The user
may input volume directly, calculate volume from an input density, default density, or
enter dimensions of various shapes. The density methods allow the user to input
component density directly or for default densities to be used.
In limited cases, there are component specialized volume calculations performed in the
component weights procedures. These specialized volumes will override densitybased
volume calculations, but not dimensional inputs or direct volume inputs. The specialized
volume calculations are described along with the weight methodology in the weights
chapter.
The design code has options for full or partial external volumes. External volumes may
occur for antennas and payloads, though almost all components may be mounted
externally. The ratio of internal to external volume is input by the user.
Structures are exempt from standard volume calculations, since the volume of the
structural outer mold line is calculated separately.
The dimensional inputs for volume calculations permit multiple components and multiple
shape types to be input by the user. This is necessary because a single functional item
may consist of multiple individual elements. For example, most EO/IR ball payloads
consist of the sphericalcylindrical EO/IR ball and separate rectangular driver electronics.
Shapes used for calculations include rectangular boxes, spheres, hemispheres, and
cylinders.
The default installation volume is a function of component category and level of design.
Currently, all default category volume installation factors are a minimum of 0.1 and a
maximum of 1.0 of the uninstalled volume. The number used is a linear interpolation
between these values with the design level as the interpolation variable. Installation
volume factors are independent of technology year. The recommended method is direct
user input of installation factors. For the default case, the volume installation factor is
found by:
( )
Max Install Vol Min Install Vol Design Max Install Vol Install Vol
F F F F F
, , , , , , ,
− ⋅ + =
The default installation volume is found by:
110
d Uninstalle Install Vol Install
V F V ⋅ =
,
The code calculates the volume margin for various elements such as wings and bodies.
The outer mold line volume is calculated in the geometry subroutines. All bodies and
wings have volume calculated. Volumetric decrements to the outer mold line volume due
to structures thickness, substructures, routing, and other considerations are estimated.
The uninstalled and installation volumes of all components allocated to a given wing or
body are summed. The difference between the available volume and the volume of the
contents is the volume margin. Fuel may be applied to this margin if fuel is allocated to
the element. Violated volume constraints are output, as described in the Optimization
section.
8.2.4 Generic Systems Power
The power of all components are divided into mission phases. This enables power
sequencing for flight performance fuel burn calculations and sizing of electrical and ECS
subsystems. The user also has the option of specifying that the component is powered at
all times.
The minimum, nominal, and maximum power are either input by the user or calculated.
The power level is selected either for each phase of flight or for the total mission. The
calculated power is based on power per weight of the component.
The design code has the ability to calculate minimum, nominal, and maximum power for
some component types. These methods are described for the applicable components in
the weights chapter.
The performance impacts of power consumption are covered in the mission performance
chapter. The nominal power is applied to all mission segments here.
8.2.5 Generic Systems Thermal Management
Component cooling is covered by the code, but component heating is not evaluated. The
required cooling of a component is based on the power consumed. This varies for
radiating and nonradiating types. Radiating components include RF communications
antennas. For nonradiating types, the cooling required is equal to the power consumed.
For radiating components the cooling required is calculated by:
( )
Rad
P C η − ⋅ = 1
where η
Rad
is the radiating efficiency of the component.
111
The design code has the ability to calculate minimum, nominal, and maximum cooling
loads for some components. These methods are described for the applicable components
in the weights chapter.
The performance impacts of cooling are covered in the Performance and Aerodynamics
sections.
8.3 Specialized Size, Weight, and Power Methods
The specialized weights methods are selected by the user when building the system files.
In addition to the standard weight methods, the user has the ability to use specialized
weights methods when available. Often multiple methods per component type are
available. This chapter describes all available methods.
8.3.1 Subsystems
8.3.1.1 Flight Control System (FCS)
The flight control system includes the flight control actuators, and equipment required to
support them. The flight control surfaces are not included in this category. Multiple
flight controls weight methods are integrated into the design code. The flight control
methods and associated functional dependencies are shown in Table 82 and Table 83,
respectively.
Method Applied
Here
1. Roskam[1990] – Cessna General Aviation FCS Weight Equaiton No
2. Roskam[1990] – General Dynamics Commercial Transport FCS Weight
Equation
No
3. Roskam[1990] – General Dynamics Bomber RCS Weight Equation No
4. Raymer[1992] Fighter FCS Weight Equation No
5. Raymer[1992] Transport FCS Weight Equation No
6. Raymer[1992] General Aviation FCS Weight Equation No
7. Custom Method Yes
Table 82 Flight Control Systems Weight Methods
112
Meth. N
Crew
N
FCS
N
Mech
N
Surf
M
Max
q
Max
V
eq,Max
S
Cont
L
Overall
W
Overall
W
DG
N
Ult
I
yy,mass
1 Yes Yes   Yes   Yes     
2  Yes Yes Yes    Yes     Yes
3  Yes       Yes Yes Yes Yes 
4  Yes         Yes  
5  Yes    Yes     Yes  
6  Yes    Yes     Yes  
7  Yes     Yes Yes     
Table 83 Flight Control System Weight Method Functional Dependencies
Method 1 has a severe limitation in that it requires a crew to be present. If no crew is
present, then the equation outputs zero weight. In order to preserve some functionality,
the number of crew, N
crew
, is set to one. This is justified because most of the FCS weight
is related to crew interfaces.
Method 2 accounts for mechanical functions beyond flight control surfaces. For the
purposes of this code, the number of mechanical functions, N
Mech
, is set to zero. This
method is not applied to the cases considered. If integrated later, the mechanical
functions of retractable landing gear could be linked.
Method 3 requires that the overall length and width is calculated. The overall length is
the distance from the farthest forward point on the vehicle to the farthest aft. The
maximum width is typically the wing span.
Method 4 has potential for broad applicability, because it is a function of design gross
weight and number of flight control systems only. Many of the listed methods have very
complex interactions among many variables that limit applicability beyond their intended
use. There is no airspeed dependency, which is largely the reason for elimination here.
Method 7 is based on rough calculations performed on model aircraft electric actuators.
The basis of the equation is the assumption that the actuators and other FCS weight
contributions are proportional to the dynamic pressure and the total control surface area.
The custom flight control system equation developed is:
FCS Cont eq FCS
N S V W ⋅ ⋅ ⋅ =
2
max ,
00002 . 0
8.3.1.2 Environmental Control System (ECS)
The environmental control system weight includes all equipment necessary to thermally
manage the aircraft. Unlike manned aircraft, there is no need for cabin environmental
control. The equations have been appropriately modified to reflect this consideration.
The environmental control system weight methods and associated functional
dependencies are shown in Table 84 and Table 84, respectively.
113
Method Applied
Here
1. Roskam[1990] – General Dynamics Fighter ECS Weight Equation No
2. Roskam[1990] – General Dynamics LowSubsonic Aircraft ECS Weight
Equation
Yes
Table 84 Environmental Control System Weight Methods
Method W
AntiIce
W
Avionics
1  Yes
2 Disabled Yes
Table 85 ECS System Weight Method Functional Dependencies
The antiicing factors in Method 2 have not been applied, because none of the UAVs
considered include antiicing systems.
8.3.1.3 Communications System
The communications system SWaP varies significantly across different applications. For
many cases, these values are entered directly by the user through the generalized methods
described earlier. One specialized method is applied for dish antennas typically used for
satellite communications.
The assumed dish weight equation, which includes the dish, gimbal, and actuators, is:
2
3
Dish Dish
D W ⋅ =
The factor of 3 was found through an analysis of terrestrial dishes. The correlation was
poor, but the factor of 3 represents highly transportable terrestrial satellite
communications dishes. A rough approximation of the volume occupied by the dish is
based on estimations from drawings of Global Hawk, which has a 4foot dish diameter.
The assumed dish volume equation is:
3
4
120 
.

\

⋅ =
Dish
Dish
D
Vol
For a given frequency, the antenna gain, and therefore signal to noise ratio, improves
with increasing diameter. The requirements for power to the antenna decrease with
improved gain for a given data rate and frequency, because the effective radiated power
in the desired direction improves. No link analysis is performed to evaluate these
impacts. Furthermore, the frequency and desired data rates are not uniform across
currently fielded UAV satellite communication solutions. It is assumed that the selected
antenna power input is proportional to the dish diameter, and the data rates and
114
frequencies are selected as appropriate to meet the aperture size. The assumed dish
power equation, based on bestjudgment, is:
Dish Dish
D P ⋅ = 250
Further refinement of satellite communications systems SWaP would be beneficial to
future UAV research efforts.
8.3.1.4 Electrical System
The electrical system includes all elements of the electrical system including, but not
limited to, generators, batteries, converters, and the power distribution elements.
Multiple empirical equations have been included in the code. The electrical system
weights methods and associated functional dependencies are shown in Table 86 and
Table 87, respectively.
Method Applied
Here
1. Roskam[1990]Cessna General Aviation Electrical System Weight
Equation
No
2. Roskam[1990]USAF Electrical System Weight Equation No
3. Roskam[1990]General Dynamics Transport Electrical System Weight
Equation
No
4. Roskam[1990]General Dynamics Bomber Electrical System Weight
Equation
No
5. Roskam [1990] – General Dynamics Navy Fighter Electrical System
Weight Equation
No
6. Raymer [1992] Transport Electrical System Weight Equation Yes
7. Raymer [1992] General Aviation Electrical System Weight Equation No
Table 86 Electrical System Weight Methods
Method W
DG
W
Fuels
W
Avion
P
Max
L
Elec
N
Gen
1 Yes     
2  Yes Yes   
3  Yes Yes   
4  Yes Yes   
5  Yes Yes   
6    Yes Yes Yes
7  Yes Yes   
Table 87 Electrical System Weight Method Functional Dependencies
The Raymer [1992] transport electrical system weight equation, Method 2, requires an
estimation of the overall length of the electrical system. This value, L
Elec
, is assumed to
be 50% of the overall vehicle length. The maximum power is calculated for all avionics
115
and most subsystems prior to the electrical system subroutine call. This method is used
because it is a function of system power, not avionics and fuel system weights.
8.3.1.5 Fuel System
The fuel system contains the fuel tanks, fuel pumps, fuel lines, and other components.
Several fuel system weight empirical equations have been integrated into the code. The
fuel system weight methods and associated functional dependencies are shown in Table
88 and Table 89, respectively.
Most of the methods are adaptable to external fuel tanks, but this was disabled since none
of the UAVs under consideration have external fuel tanks. The parameter
Vol
Internal
/Vol
Tot
is set to one.
The tank volume is simply the fuel weight divided by the fuel density for the purposes of
these calculations. Fuel volume constraint checking is handled in other subroutines.
The total number of tanks, N
Tanks
, is assumed to be two for the purposes of this code.
Method Applied
Here
1. Raymer [1992] Fighter Fuel System Weight Equation No
2. Raymer [1992] Transport Fuel System Weight Equation Yes
3. Raymer [1992] General Aviation Fuel System Weight Equation No
Table 88 Fuel System Weights Methods
Method Vol
Tank
Vol
Internal
/Vol
Tot
N
Tanks
T
max
TSFC N
Eng
1 Yes Yes Yes Yes Yes Yes
2 Yes Yes Yes   
3 Yes Yes Yes   Yes
Table 89 Fuel System Weight Method Functional Dependencies
The Raymer [1992] fighter fuel system weight equation originally contained provisions
for protected fuel tanks. Factors for this consideration were disabled.
8.3.1.6 Auxiliary Power Unit (APU)
The user may either enter in the APU weight directly or use the APU weight estimation
method. The estimation method is an attempt to relate APU weight from the power
required of it.
116
It is assumed that the power required of the APU is 50% of the maximum power
experienced by the aircraft. The APU is assumed to weight 1.5 pounds per horsepower,
which is between typical values of 2stroke and 4stroke engines, and is heavy for a
turboshaft. None of the cases considered are believed to contain an APU, so this method
was not utilized.
8.3.2 Avionics
Avionics are given substantial consideration in this research, because they can be a
dominant weight driver. For example, a small component like a GPS antenna can weigh
as much as an optical payload on a very small UAV. The weight of a GPS antenna is
within typical weight calculation error levels for a manned aircraft.
A general methodology for avionics weight was attempted. Avionics suites are highly
dependent on the mission, vehicle class, design philosophy, and technology. General
avionics methods were attempted for autopilots, INS, GPS, processors, cameras,
recorders, and ATC avionics. Though providing some utility, these methods were not
employed in the final evaluations of the cases of interest. The avionics suites of the
vehicles under consideration are simply too divergent and casespecific to be captured in
generalized formulations. With the exception of the applied ADS method, all avionics
were input directly. The avionics methods are shown in Appendix C for completeness,
though the methods are not mature. It is unlikely that an effective, generalized avionics
methodology is possible.
8.3.2.1 General Avionics Functions
The weight, volume and power functions all use a common form. The common equation
is:
n
o
X a Y Y ⋅ + =
X
o
(=0) and Y
o
is input. The mid values X
1
(=0.5) and Y
1
(=1) are input. The maximum
values X
2
(=1) and Y
2
are also input. The values of n and a are found by:


.

\

−
−


.

\

−
−
=
0 1
0 2
0 1
0 2
ln
ln
X X
X X
Y Y
Y Y
n
( )
n
X X
Y Y
a
0 1
0 1
−
−
=
117
If Y
0
is greater or equal to Y
1
, or if Y
1
is greater than or equal to Y
2
, then linear
interpolation is used.
8.3.2.2 Air Data System (ADS)
Air data systems are necessary to provide the autopilot with angle of attack, angle of
sideslip, pressure altitude, and dynamic pressure, depending on the autopilot. The ADS
weight equations used here are based upon bestjudgment assumptions. The ADS
method is intended to make the ADS estimated weight of the correct order of magnitude
for a wide range of UAV AVs. Future research could adjust the factors of these
equations based on regression analysis of a custom ADS database. The ADS weights
resulting from these methods appear to be reasonable for the purposes of this research,
considering that there are no alternative methods available.
Two types of air data systems are considered: ADS booms and integral ADS. The typical
ADS boom weight is assumed to be:
max ,
05 . 0
eq ADS
V W ⋅ =
The typical weight of an integral ADS system is assumed to be:
max ,
02 . 0
eq ADS
V W ⋅ =
If the maximum flight altitude is greater than 15,000 feet, the ADS is assumed to have a
heater. The weight of the ADS with heater is assumed to be:
ADS ADS
W W ⋅ = 2 . 1
The ADS weight is corrected for ruggedness, performance, and miniaturization. The
factors are shown in Table 810.
Driver Minimum (0) Mid (0.5) Maximum (1)
Ruggedness 0.8 1.0 1.2
Performance 0.8 1.0 1.2
Miniaturization 0.6 1.0 1.4
Table 810 Air Data System Factors
The weight and volume are initially calculated by using the boards and boxes formulas.
Only 25% of the ADS volume is internal. The weight and volume installation factors are
the same as for boards and boxes.
118
8.3.3 Structures
An attempt has been made to provide multiple options for the various structural weights.
Numerous manned aircraft equations have been utilized, but other customized options
have been introduced. The new piecewise linear beam structures methods do not have
reference UAV cases to draw upon, so fidelity may be low. However, this is often better
than using equations for manned aircraft outside their applicable range.
8.3.3.1 Wing
Multiple methods for calculating wing weights have been incorporated into the design
code. There are options for three parametric weights and one semianalytic method. The
parametric weights are modified versions of Raymer’s [1992] fighter, transport, and
general aviation wing weights. The semianalytic method is a piecewise linear beam
model. All wing weight methods are available, but the piecewise linear beam model is
believed to have the most general applicability to small AV designs. The piecewise
linear beam model was applied to all cases in this work. The wing weight methods and
associated functional dependencies are shown in Table 811 and Table 812, respectively.
The code can handle multiple wings performing lifting functions, so canards, tandem
wings and other types may be evaluated. The level flight loading at the design gross
weight condition of each wing is found by:
tot W
W DG
Wing
S
i S W
i Load
,
) (
) (
⋅
=
Method Applied
Here
1. Raymer [1992] Fighter Wing Weight Equation No
2. Raymer [1992] Transport Wing Weight Equation No
3. Raymer [1992] General Aviation Wing Weight Equation No
4. Piecewise Linear Beam Yes
Table 811 Wing Weight Methods
Meth. Load
W
N
Ult
S
w
AR
w
t/c
Avg
λ
eq
Λ
c/4
S
c
/S
w
q W
Fuel,wing
Tech
1 Yes Yes Yes Yes Yes Yes Yes Yes   Yes
2 Yes Yes Yes Yes Yes Yes Yes Yes   Yes
3 Yes Yes Yes Yes Yes Yes Yes  Yes Yes Yes
Table 812 Wing Weight Method Functional Dependencies
8.3.3.1.1 Piecewise Linear Beam Model
119
The wing and tail structure is analyzed using the piecewise linear beam model. The
methods were inspired by publications from the StrutBraced Wing (SBW) research at
Virginia Tech [NaghshinehPour 1998]. The method applied here is more limited in
scope, since only cantilever wings are considered.
Piecewise Linear Beam Load Estimation
The load allocated to each wing is assumed to be proportional to the ratio of the area of
the wing to the total wing area. The total load carried by the wing is found by:
Tot W
W
DG ult Wing
S
S
W N Load
,
⋅ ⋅ =
The design code currently does not calculate the wing load distribution based on the lift
distribution methodologies used in the aerodynamics. However, these methods could be
adapted in future versions of the code. The lift load distribution method assumes an
elliptical distribution. The lift load for each panel is found by:
Tot Struct
Run
Tot Struct
Struct
Wing
b
k Y
b
k b
Load k Lift
, ,
) ( 2
1
4 ) (
) (
⋅
− ⋅ ⋅ ⋅ =
π
where Y
Run
is the distance of the midpanel centroid from the root along the path of
centroids, and b
Struct,Tot
is the total structural span of the wing along the path of centroids.
The wing may carry nonstructural components such as avionics and subsystems, which
provide inertia relief. These are allocated to the wing by the standardized systems
subroutines. The total weight of these nonstructural components, W
NonStruct,Tot
, is
allocated externally. The load distribution is assumed to be triangular, with the
maximum weight at the root. The nonstructural weight distribution is found by:


.

\

⋅
− ⋅
⋅
⋅ =
− −
Tot Struct
Run
Tot Struct
Struct
Tot Struct Non Struct Non
b
k Y
b
k b
Wt k W
, ,
,
) ( 2
1
) ( 2
) (
The overall shear load and bending moment due to lift and nonstructural weight are
found by numerical integration from the tip to the root. The wing weight is calculated
from tip to root. The shear and moment are corrected for the structural wing weight of
outboard panels.
8.3.3.1.2 Piecewise Linear Beam Wing/Tail Weight Estimation
The thickness of the bending material for the panel is found by:
120
ult t xx
Box
Bend
Bend
k I
k T
k Mom
k t
σ ⋅
⋅
=
) (
2
) (
) (
) (
,
If this value is less than the minimum gage value, then the thickness is set to the
minimum gage thickness. The total bending material weight of the panel is found by:
) ( ) ( ) ( k S k t k W
Pan Bend Matl Bend
⋅ ⋅ = ρ
The total panel area moment of inertia is found by:
) ( ) ( ) (
,
k t k I k I
Bend t xx xx
⋅ =
The shear material weight assumes that the majority of the structural material is
contained in the wing skins, so the contribution of shear material to the area moment of
inertia and the shear flow is small. The area moment of inertia for shear material
calculations is identical to that found through the wing bending material calculations.
The shear web is assumed to be a single web which occupies the entire maximum
thickness of the panel. It is assumed that if multiple webs are used, the total shear
material weight will be comparable. The shear flow is found by:
) ( ) ( ) (
,
k t k q k q
Bend t Shear Shear
⋅ =
The shear material thickness is found by:
ult xx
Shear
Shear
k I
k q k Shear
k t
τ ⋅
⋅
=
) (
) ( ) (
) (
If the shear material thickness is greater than the minimum gage thickness, then the shear
material thickness is set to the minimum gage value. The shear material weight of the
panel is found by:
) ( ) ( ) ( ) ( k b k T k t k W
Struct Pan Shear Shear
⋅ ⋅ =
The bending and shear material weights are summed across the wing to provide the total
wing weight:
∑
=
⋅ =
Nseg
k
Bend Tot Bend
k W W
1
,
) ( 2
∑
=
⋅ =
Nseg
k
Shear Tot Shear
k W W
1
,
) ( 2
121
The wing box weight is found by summing the total bending and total shear weights:
Tot Shear Tot Bend Box
W W W
, ,
+ =
The wing box is only part of the wing weight. Other weights components include:
• Attachments for engines, undercarriage, and miscellaneous items, W
Attach
• Leading and trailing edge control surfaces and secondary structure, W
CS,Secondary
• Rib Weight, W
Ribs
Empirical equations for these components are difficult to find. Detailed weight
statements for aircraft, manned or unmanned, are much more scarce. Macci [1995]
developed a semianalytical method for estimating wing weight. While this method is
not directly employed here, the results of the analysis are helpful in determining the
weight of wing components. Figure 81 shows the resulting multiplication factors
applied to the wing box for other wing structural elements.
Wing box mulitplication factors
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0 10,000 20,000 30,000
Wing box weight
W
i
n
g
b
o
x
m
u
l
t
i
p
l
i
c
a
t
i
o
n
f
a
c
t
o
r
s
Attach fact
CS Second fact
Rib fact
Figure 81 Macci Wing Factors
The average wing box multiplication factor value for each weight type was determined,
to give the following weight estimation equations:
Box Attach
W W ⋅ = 303 . 0
Box Secondary CS
W W ⋅ = 500 . 0
,
Box Ribs
W W ⋅ = 114 . 0
The overall wing weight is:
122
Ribs Secondary CS Attach Box Wing
W W W W W + + + =
,
8.3.3.1.3 Piecewise linear beam model considerations
The piecewise linear beam model contains both conservative and optimistic assumptions.
While these considerations may partially cancel one another, this will vary from case to
case. No single parametric wing weight equation is applicable across the wide range of
vehicles considered in this research. The hope is that, despite the beam model
limitations, it will provide a reasonable estimate across all vehicles where no applicable
competing methods exist.
One conservative assumption made in the model is that the wing skin bending material
and shear material thicknesses are uniform for a given segment. The thicknesses are
calculated at the points of maximum bending and shear stresses for a given wing
segment, which means that all other points have more thickness than required. Although
conservative, this may reflect reality because of limitations of processes to manufacture
variable thickness skins at a given spanwise location.
This model does not account for wing torsion, drag loads, drag moments, raillaunch
shock loads, landing loads, offnormal lift load vectors, or axial loads. Also, aerolastic
effects, especially flutter, can be a major driver in high aspect ratio wings. No direct
consideration of aerolastic effects are considered here.
All of the wings are assumed to be designed for strength rather than for stiffness.
Additional weight penalties will be incurred if a stiffness based wing design is necessary.
These penalties are not applied in this code.
8.3.3.2 Tails
The tail weight accounting includes all tail structures as a single weight, which simplifies
the calculation of multiple tail arrangements. The piecewise linear beam model and
Raymer [1992] transport and general aviation tail weight methods are included. The
piecewise linear beam model is applicable to horizontal tails and Vtails. All the
configurations under consideration have Vtails, so the piecewise linear beam model was
employed. The tail weight methods are shown in Table 813. The horizontal tail weight
methods functional dependencies are shown in Table 814. The vertical tail weight
methods functional dependencies are shown in Table 815.
123
Method Applied Here
1. Raymer [1992] Transport Tail Equations No
2. Raymer [1992] General Aviation Tail Equations No
3. Piecewise Linear Beam Yes
Table 813 Tail Weight Methods
Meth. W
DG
N
Ult
S
H
AR
H
L
H
λ
H
Λ
c/4,H
S
c
/S
H
q Tech
1 Yes Yes Yes Yes Yes Yes Yes Yes  Yes
2 Yes Yes Yes Yes  Yes Yes  Yes Yes
Table 814 Horizontal Tail Weight Method Functional Dependencies
Meth. W
DG
N
Ult
S
V
AR
V
L
V
λ
V
Λ
c/4,V
q Tech
1 Yes Yes Yes Yes Yes  Yes  Yes
2 Yes Yes Yes Yes  Yes Yes Yes Yes
Table 815 Vertical Tail Weight Method Functional Dependencies
For Vtails (also known as butterfly tails), the tail weight is calculated as a combination
of horizontal and vertical tail weights when using parametric equations. The Vtail
weight is calculated without these manipulations when applying the piecewise linear
beam model. All other geometry parameters such as taper ratio, sweep, and thickness to
chord ratio are identical. The Vtail weight is found by:
( ) ( )
2 2
Γ ⋅ + Γ ⋅ =
−
Sin W Cos W W
VT HT Tail V
The aspect ratio for the vertical tail part of the calculation is assumed to be half of the
input aspect ratio.
8.3.3.3 Fuselage
Three fuselage methods have been adopted from Raymer [1992]. For many of the
smaller UAVs considered, these manned aircraft equations are not working within the
intended applicable bounds. The fuselage weight methods and associated functional
dependencies are shown in Table 816 and Table 817, respectively.
Method Applied
Here
1. Raymer [1992] Fighter Fuselage No
2. Raymer [1992] Transport Fuselage No
3. Raymer [1992] General Aviation Fuselage Yes
Table 816 Fuselage Weight Methods
124
Meth. W
DG
N
Ult
S
wet,Fuse
L
Tail
L/D
Fuse
L
Fuse
q Wing
Geom
Tech
1 Yes Yes   Yes* Yes   Yes
2 Yes Yes Yes  Yes Yes  Yes Yes
3 Yes Yes Yes Yes Yes  Yes  Yes
Table 817 Fuselage Weight Method Functional Dependencies
The Raymer [1992] General Aviation fuselage equation was used for all cases. This
method requires special information on the nominal cruise dynamic pressure, tail location
on the fuselage, and cabin pressurization. The nominal cruise dynamic pressure, q, is
assumed to be 75% of the maximum dynamic pressure. The tail geometry imposes
limitations on the design, since many UAVs may not have tails attached to the fuselage or
even have tails at all. Nevertheless, the length of the tail from the wing, L
Tail
, is
approximated as 60% of the fuselage length. The fuselage is assumed to be
unpressurized, so no pressurization weight penalties are incurred.
8.3.3.4 Nacelles
The Raymer [1992] transport engine nacelle weight was applicable to both pylon and
other arrangements. The pylon mounted option is not used, so the associated correction
factor, K
NG
, is set to unity. The equation is dependent upon the engine controller weight,
W
Engcont
, but this is not always included in the nacelle. The engine controller weight was
set to 10% of the engine weight for the purposes of these calculations, as an assumption.
8.3.3.5 Launch and Recovery Gear
Multiple empirical methods exist for estimation of landing gear weight [Raymer 1992,
Roskam 1990], but these methods typically have landing gear geometry functional
dependencies. Raymer [1992] landing transport landing gear weights equations were
applied to the Global Hawk, Predator, and Shadow 200 in a spreadsheet. The average
mass fraction of the landing gear relative to the design gross weight as evaluated was
4.28%. The minimum landing gear mass fraction was 3.26%, and the maximum was
5.16%. Given the limitations of the analysis and large spread of results, 4% was selected
as the basic landing gear mass fraction applied to all conventional landing gear in this
research.
Recoveryonly gear can take the form of nonrecoverable (crash landing), skid landing, or
parachute recovery. The parachute recovery is considered here. Parachute recovery gear
is assumed to weigh 2% of the design gross weight, as a weight index to the conventional
landing gear method. The parachute mass fraction method is a bestjudgment
assumption. The Shadow 200 can use a parachute recovery system. The Global Hawk
and Predator use conventional landing gear only.
125
8.3.4 Propulsion
8.3.4.1 Reciprocating Engines
The engine calculation method multiplies the specific power and the engine power to
obtain the engine weight. This is calculated by:
Eng Specific Eng
P P W ⋅ =
The overall engine dimensions are found by Roskam [1990] 6cylinder engine formulas
modified to have application beyond general aviation engines:
15 . 0
56 . 1
Eng Eng
P L ⋅ =
Eng
Eng
Eng
P
P
Wid ⋅ + ⋅ = 00134 . 0
200
5 . 1
Eng
Eng
Eng
P
P
H ⋅ + ⋅ = 00156 . 0
200
7 . 1
The engine volume is found by:
Eng Eng Eng Eng
H Wid L Vol ⋅ ⋅ =
The power draw and cooling of the engine are set to zero for the standard component
accounting. These are calculated under the mission performance.
8.3.4.2 Jet Engines
A simple thrust to weight method and a parametric method are used to determine the
engine weight. A specialized means of calculating the jet engine volume was employed.
Jet engine weight – thrust to weight method
The thrust to weight jet weight calculation method is:
Jet
Jet
Jet
W T
T
W
/
=
Jet engine weight – Roskam method
126
The Roskam [1990] jet weight calculation method is available in the code. This method
relates the jet engine weight to the thrust level and bypass ratio. This method was not
used to generate the results of this effort.
Jet engine weight – default rubber engine method
This method uses a default reference engine to calculate the engine weight, length, and
diameter. The thrust scale factor is found by:
Ref Jet,
Jet
T
T
SF =
The jet engine weight is found by:
SF W W
Ref Jet, Jet
⋅ =
Jet engine volume calculations
For the thrust to weight and Roskam [1990] engine weight methods, the engine diameter
is found by using a reference thrust/diameter
2
parameter. The default rubber method uses
the length and diameter methods described earlier.
A limited engine database for turbojets and turbofans was used to determine the
length/diameter
2
parameter. Unfortunately, there is little correlation between bypass ratio
and the parameter, and there is significant scatter. A nominal value of 500 lb/ft
2
was used
for the purposes of these simple calculations. The jet engine diameter equation is:
( )
Ref Jet
Jet
Jet
D T
T
D
2
/
=
The engine length is found by a length/diameter parameter. The engine database was
used to select an appropriate value. Similar to the thrust/diameter
2
parameter, there was
little correlation with aspect ratio, and the data has significant scatter. The jet engine
length equation is:
( )
Ref Jet, Jet Jet
D L D L / ⋅ =
The jet engine volume is then found by:
2
4
Jet Jet Jet
D L Vol ⋅ =
π
Jet engine power and cooling
127
The jet power consumption and cooling are set to zero in the component calculations.
The impacts of engine cooling are calculated in the performance model. These are
covered in the mission performance chapter.
128
Chapter 9 Flight Performance
9.1 Overview
The performance model predicts aircraft fuel burn and optimizes flight trajectories. The
mission analysis evaluates a wide array of mission types with an emphasis on flexibility
and robustness. Simple performance methods are combined in multiple ways to solve
complex performance problems without the need for external flight trajectory
optimization.
The performance code is designed to evaluate the fuel burn of the AV as it flies the
mission profile and to evaluate associated constraints. Several options exist for
evaluating the mission, ranging from direct evaluations of a specified mission to limited
mission optimization.
The code evaluates multiple mission segments. These segments can represent climb,
cruise, descent, loiter or combinations thereof. The individual segments are further
broken into subsegments. During each segment of flight, several constraints are
evaluated. These include stall, specified Mach limits, specified altitude limits, specified
dynamic pressure limits, insufficient thrust for level flight, insufficient thrust for
prescribed rate of climb, and engine operating limits.
The performance code has the ability to optimize flight velocities. The figures of merit
include range efficiency, endurance efficiency, climb efficiency, and maximum rate of
climb. These velocities are optimized such that no constraints are violated.
Two possibilities exist for the evaluation of fuel burn. First, standard or arbitrary mission
profiles can be evaluated either forwards or backwards to determine the percentage of the
mission that could be completed. Second, a loiter segment can be established in the
middle of a mission profile with sufficient endurance to use all of the available fuel. The
first possibility lends itself well to a mission completion constraint, and the second is
suitable for an objective function contributor.
129
9.2 Performance Code Architecture
Figure 91 Performance Code Architecture
Figure 91 shows the performance code architecture. The PERFORM subroutine drives
the performance evaluation. Prior to calling PERFORM, the mission parameters must be
input, the aerodynamics and propulsion tables must be generated, and the weights must
be determined. Once this is done, PERFORM calls the necessary subroutines to evaluate
the fuel burn and mission constraints.
ACMAIN initially calls PERFORMIN to read and interpret the mission input data files.
ACMAIN then calls EVAL directly or through the optimizer. EVAL calls the propulsion,
geometry, weights, and aerodynamics subroutines, along with others, to generate the data
necessary to drive PERFORM.
PERFORM, the main performance subroutine, first calls MISSOPT to organize the
mission. MISSOPT has the ability to drive the evaluation of a normal onedirection
mission or an optimized centered mission. With either option, MISSOPT calls PERFSEG
for every mission segment. PERFSEG subdivides the mission segment into sub
segments and calls either PSEGSEG for prescribed velocities or PSEGSEGO to
determine optimized velocities. PSEGSEGO calls PSEGSEG multiple times to determine
optimal velocities that do not violate constraints. PSEGSEG calls WTFINSEG multiple
times to determine the fuel burn or performance and any remaining unknowns.
The header of the mission input file has the following format:
130
Test performance input file
2 miss_type (1=nonopt,2=R spec)
0.99 ff_taxi
4 nsegseg
8 nseg
The toggle flag miss_type specifies the type of mission to be flown. A value of 1
indicates a normal onedirection mission, and a value of 2 indicates that the mission is
centered and that all the fuel will be depleted during the midmission loiter.
ff_taxi is the fuel fraction that is used during warmup, taxi, and takeoff. This does not
include the climb to altitude.
nsegseg is the number of subsegments each segment will be divided into, and nseg is the
total number segments in the mission.
The remainder of the mission input file represents the individual mission segments. Each
segment has the following format:
3 segment number
1 fwd_flag (1=fwd, 2=reverse, 3=center)
1 ra_en_flag (1=range inpt, 2=endur inpt, 3=climb)
20. rangeseg (if ra_en_flag = 1), nmi
1. endurseg (if ra_en_flag = 2), hrs
3 vm_flag  velocityMach flag (1=V,2=M,3=R,4=E,5=h/t,6=h/f)
150. vel (if vm_flag = 1), kts
0.3 mach (if vm_flag = 2)
1 contig_flag (1=contiguous, 2 = not contig)
0. alt1  initial altitude (if contig_flag=2), ft
1 altroc_flag (1=fin alt,2=ROC,3=cruise,4=const)
30000. alt2 (if altroc_flag = 1), ft
500. roc (if altroc_flag = 2), ft/min
The selection flag fwd_flag indicates whether the mission is flown forward from takeoff
(at design gross weight) to landing, backwards from landing (at the empty weight) to
takeoff, or from a centered mission. A centered mission must follow a centered mission
profile described later.
The toggle flag ra_en_flag indicates whether range, endurance, or climb is the primary
objective of the mission segment. Range, R
seg
, is input if range is the primary objective.
Endurance, E
seg
, is input if endurance is the primary objective. If climb is the primary
objective, then climb between two specified altitudes will be achieved either with
maximum rate of climb or maximum fuel efficiency.
131
The selection flag vm_flag indicates how the flight velocity will be determined. The
flight velocity or the Mach number may be input directly, where constraints may be
violated. The flight velocity may be optimized such that no constraints are violated. The
options for optimized flight velocities include the velocities associated with best range,
endurance, rate of climb, or most efficient climb.
The toggle flag contig_flag indicates whether the mission segment is contiguous with an
adjacent mission segment. Either the initial or final altitude may be set to the appropriate
altitude of the adjacent segment when contiguous, depending on the direction in which
the mission is evaluated. The altitudes must be input directly when the segment is not
contiguous.
The selection flag altroc_flag indicates whether the final altitude or constant rate of climb
is input.
9.3 Performance Methodology
9.3.1 Performance Equations
The heart of the performance modeling for fueled aircraft begins with the Breguet range
equation:


.

\

⋅
⋅
=
f
i
W
W
TSFC
D L V
R ln
/
This form of the Breguet equation requires that the product V*L/D/TSFC remains
constant, which necessitates a cruiseclimb. Many missions are flown at constant altitude
or velocity, creating inaccuracies in the Breguet range equation estimation. Climbs
differing from the cruise climb can also introduce inaccuracies. The product may not
hold constant well if the altitude change is significant. The prediction accuracy can be
improved by breaking each mission segment into subsegments and introducing
corrections.
∑
=
seg
R R
The equations that follow may be calculated in several different orders, depending on
which information is known a priori.
The Breguet range equation is modified to account for thrust values differing from the
ideal.
132


.

\

⋅
⋅

.

\

⋅
=
f
i
mult
seg
W
W
TSFC T
D
L
V
R ln
where the thrust multiplication factor, T
mult
, is found by:
D
T
T
req
mult
=
The required thrust, T
req
, is found by:
W
D L V
ROC
T
req
⋅ 
.

\

+ =
/
1
The lift to drag ratio, L/D, is replaced by the ratio of weight over drag. The required
thrust becomes:
D W
V
ROC
T
req
+ ⋅ =
The lift coefficient, C
L
, is simply:
w
L
S V
W
C
⋅ ⋅ ⋅
=
2
2 / 1 ρ
The air density, ρ, is found via a standard atmosphere subroutine. The drag coefficient is
found by the aerodynamics table lookup subroutine:
) / , , ( L R M C C C
e L D D
=
where,
a
V
M = or a M V ⋅ =
depending on whether Mach number or velocity is known beforehand. The speed of
sound, a, is found via the standard atmosphere subroutine. The Reynolds number per unit
length, R
e
/L, is found by:
u
ρ V
L R
e
⋅
= /
where the viscosity and density are found via the standard atmosphere subroutine.
133
The specific fuel consumption, TSFC, is found via the engine deck lookup subroutine. In
general,
) , , (
req
T M alt TSFC TSFC =
The minimum and maximum thrust values at the given altitude, T
min
and T
max
,
respectively, are also output from the engine deck lookup subroutine.
The time taken to complete the subsegment range is:
V
R
E
seg
seg
=
The performance modeling must also capture the impacts of cooling drag and power
extraction. The cooling drag estimation is described in the aerodynamics chapter. The
relationship between drag and power, assuming thrust is equal to drag, is:
V D Pow ⋅ =
If a generator supplies this power, then there will be power generation efficiency, η
Gen
.
When this relationship is rearranged and corrected for units, power extraction equivalent
drag is found by:
Gen
Pow
V
Pow
D
η ⋅
⋅ = 43687 . 0
where D
Pow
is in pounds, Pow is in Watts, and V is in knots. This method is a simple
approximation for the equivalent drag resulting from power extraction. The approach
lends itself well to reciprocating engines, and is assumed to be applicable to jet engines as
well.
The overall drag including all considerations becomes:
Cool Pow
L
D
D D W
C
C
D + + ⋅ =
The weight calculations can occur in different orders, depending on the available inputs
and required outputs. For example, the segment range and initial weight can be input and
the final weight output. Or, the initial and final weights can be input and the endurance
output. The mission performance code is structured such that multiple combinations of
inputs and outputs may be accommodated.
For a forwardcalculated mission segment, the final weight is found by:
134





.

\

⋅
⋅ ⋅
=
D
L
seg mult
C
C
V
TSFC R T
i
f
e
W
W
For a reversecalculated mission segment, the initial weight is found by:





.

\

⋅
⋅ ⋅
⋅ =
D
L
seg mult
C
C
V
TSFC R T
f i
e W W
9.3.2 Performance Constraints
There are multiple flight envelope constraints:
1)
stall
V V < Stall constraint
2)
min
ROC ROC
avail
< Minimum rate of climb constraint
3)
max
T D > Insufficient thrust for level flight constraint
4)
SL
eq
V
V
ρ ρ /
max ,
> Dynamic pressure constraint
5) a M V ⋅ >
max
Mach constraint
6)
min
T T
req
< Minimum available thrust constraint (not set yet)
7)
max
alt alt > Maximum altitude constraint
Figure 92 shows potential flight envelope constraints.
Figure 92 Flight Envelope Boundaries
135
9.3.3 Flight Velocity Optimization
To optimize the velocities, there are several figures of merit:
f i
seg
W W
R
FOM
−
= Range efficiency
f i
seg
W W
E
FOM
−
= Endurance efficiency
avail
ROC FOM = Maximum climb rate
f i
seg
W W
E ROC
FOM
−
⋅
= Climb efficiency
A search algorithm finds the optimal velocity for the given figure of merit in the
constrained flight envelope. In a trade between speed and robustness, a bisection search
algorithm is used.
Figure 93 shows how the search is initiated. The initial minimum velocity is the stall
speed. The initial maximum velocity is the minimum of the dynamic pressure limit and
the Mach limit. In this case, the minimum and maximum velocities fall outside the
constraint boundaries. Depending on which constraints are violated, the minimum
velocity is incrementally increased or the maximum velocity is incrementally decreased
until the boundaries fall within the feasible region.
Figure 93 Velocity Search Initialization
Once all the velocities fall within the feasible region, the search begins. If the point with
best figure of merit does not lie on the end, then the point with the best figure of merit
becomes the point mid. The case where the best figure of merit lies on the end will be
discussed later. The right and left adjacent points become points min and max,
respectively. Point 1 is created halfway between points min and mid, and point 2 is
created halfway between points mid and max. The process continues through several
136
iterations. The velocity associated with the best figure of merit at the end of the search is
the final velocity selected.
The process is different when the best figure of merit is an end point. In the case where
the best figure of merit is point min, point mid becomes point max and point 1 becomes
point mid. Points 1 and 2 are generated halfway between these points as described
above. The process continues with one or more of the processes discussed until the
maximum number of iterations is reached.
9.3.4 Climb Performance
The cruise climb altitude change is assumed equal to the altitude change required to keep
the lift coefficient constant. The final subsegment altitude is found through linear
interpolation or extrapolation. The reference altitude change is 10,000 feet. The cruise
climb altitude change is found by:
( )
( ) ( )


.

\

− ⋅
− +
⋅ + = 1
1 000 , 10 1
1
000 , 10 1 2
i
f
W
W
alt alt
alt
alt alt
ρ ρ
ρ
The rate of climb associated with the cruise climb is estimated. If the cruise climb rate of
climb exceeds the 80% of the available rate of climb at the given velocity, the cruise
climb rate of climb is set to 80% of the available rate of climb. The 80% factor is applied
to account for typical reductions in available rate of climb as altitude is increased. When
a cruise climb profile is selected, the fuel burn is first calculated without any rate of
climb. The difference in fuel burn for the level flight and cruise climb conditions is
found by first estimating the difference in thrust required for cruise climb:
W
V
ROC
T
Climb
⋅ = ∆
Next, the difference in fuel burn for the cruise climb is estimated by:
Seg Climb Climb Fuel,
E SFC T W ⋅ ⋅ ∆ = ∆
Note that E
Seg
is the endurance of the subsegment. The optimal cruise velocity is found
for each new subsegment. The initial cruise altitude for the second through final sub
segments is simply the altitude associated with the best cruise condition of the
predecessor segment.
For forwardcalculated cruise climb segments, the altitude used for calculations is the
starting altitude. Similarly, reversecalculated cruise climb segments use the final
altitude. The average altitude is output.
137
A constant rate of climb may be specified for a mission segment. The flight velocity may
be optimized for any figure of merit desired. For a forward evaluation:
60 1 2 ⋅ ⋅ + =
seg seg
E ROC alt alt
For a backwards evaluation:
60 2 1 ⋅ ⋅ − =
seg seg
E ROC alt alt
If the initial and final altitudes are specified, then the rate of climb is found by:
( )
60
1 2
⋅
−
=
seg
seg
E
alt alt
ROC
If the final altitude, alt2, exceeds the maximum altitude, then the final altitude is set to
the maximum altitude. Unlike the flight velocities, the altitude calculations do not
optimize altitudes such that no constraints are violated. The performance code will
output the violated constraint value if the required rate of climb is not attainable.
Even if a rate of climb is input, it may not be achievable. The maximum available rate of
climb is:
( )
60
6080
max
max,
⋅ ⋅
−
= V
W
D T
ROC
avail
For descent, the minimum rate of climb (positive rate of descent) is:
( )
60
6080
min
min,
⋅ ⋅
−
= V
W
D T
ROC
avail
The rate of climb may be reset for missions segments that permit flexibility in the rate of
climb. If the required rate of climb exceeds the maximum rate of climb, then the rate of
climb will be set to the maximum rate of climb. Similarly, if the required rate of descent
exceeds the minimum rate of climb, then the rate of climb is set to the minimum rate of
climb.
The mission analysis code can evaluate the climb performance without a prescribed range
or endurance. The initial and final altitudes must be input, and the range and endurance
will be output. The two desired climb modes are maximum rate of climb and most fuel
efficient climb. The associated velocities are found via the outer loop figure of merit
search procedure described earlier. For the maximum rate of climb, maximum throttle is
used. For most fuel efficient climb, a search procedure is initiated with the following
figure of merit:
138
f i
seg
W W
ROC
FOM
−
=
The initial and final weights are evaluated with the Breguet range equation with the
segment range missing, thus eliminating the range and endurance dependency. The
endurance and range are then found by:
60
1 2
⋅
−
=
seg
seg
ROC
alt alt
E and V E R
seg seg
⋅ =
The initial or final subsegment weight, depending on evaluation direction, is then found
via the modified Breguet range equation.
139
9.3.5 Mission Evaluation
Mission profiles can be either standard or arbitrary. The standard mission profile is
useful because it permits an evaluation of available time on station for a given fuel load.
Arbitrary missions are much more flexible. These can contain up to 20 segments and the
compiler integer limit for subsegments per segment.
Figure 94 Standard Mission Profile
Figure 94 shows the standard mission profile used for longrange missions. The boxes
identify the mission segment name and the flight order. The arrows indicate the direction
of flight. The underlined numbers show the order of calculation for a centered mission.
The order of calculation is the same as the order of segments for forward calculations or
in the opposite order for backwards calculations.
9.3.6 Mission Completion Criterion
The mission completion criterion is a metric for determining if the air vehicle with the
available fuel load can complete the desired mission. The mission completion criterion is
handled differently for centeredstandard and arbitrary mission profiles.
The centeredstandard mission completion criteria depends on the available fuel at the
loiter segment. If the loiter segment’s final weight is greater than the initial weight, then
the time on station will be negative. The mission completion criterion is:
i
i f
W
W W
misscomp
−
− = 1
140
Otherwise, if the loiter segment initial weight is greater than the final weight, the mission
completion criteria is unity. The time on station is a direct result of the available fuel,
and this may be an objective function.
The arbitrary mission completion criterion evaluation method depends on if the AV
completes the mission. If the AV completes the mission from either the forward or
backward analysis direction, the mission completion criterion is found by:
fuelb
fuel
W
W
misscomp =
A different procedure is used when there is insufficient fuel to complete the arbitrary
mission. The mission analysis code keeps track of three different types of mission
segment: range, endurance, and climb. Although all three parameters are known for each
segment, only one per segment is tracked for the mission completion criteria. First, the
total range, endurance, and climb are calculated:
∑
=
seg tot
R R For all rangetype segments
∑
=
seg tot
E E For all endurancetype segments
∑
∆ = ∆
seg tot
alt alt For all climbtype segments
Next, the relative weighting of each metric must be assessed. The range, endurance, and
climb are all put in terms of distance. Then each distancenormalized parameter is
weighted. First, the common denominator is found:
max ,
min
75 . 0
60
eq
tot
tot tot
V
ROC
alt
E R Denom ⋅ ⋅


.

\

⋅
∆
+ + =
Second, the relative weightings are found:
Denom
R
R
tot
wt
=
Denom
V E
E
eq tot
wt
max ,
75 . 0 ⋅ ⋅
=
Denom
V
ROC
alt
alt
eq
tot
wt
max ,
min
75 . 0
60
⋅ ⋅
⋅
∆
=
If the air vehicle depletes the fuel during any mission segment, then an error flag is
output along with the amount of the segment that is completed. All segments that follow
141
are not completed. There is a running summation of the completed and partially
completed segments. The mission completion parameter for noncompleted arbitrary
missions is:
tot
comp
wt
tot
comp
wt
tot
comp
wt
alt
alt
alt
E
E
E
R
R
R misscomp
∆
∆
⋅ + ⋅ + ⋅ =
142
Chapter 10 Calibration
10.1 Calibration Overview
Design code calibration is essential to ensuring the validity of a new design analysis. It is
highly unlikely that a design algorithm will provide sensible results without adequate
tailoring of inputs and methods. The present design code was calibrated for the three
UAV classes under consideration. Once calibrated, the technology design impacts may
be assessed with a higher degree of confidence.
10.2 Calibration Methodology
Design code calibration is a challenging endeavor. It is possible to reach a point
calibration through many different approaches. In this effort, there are more code inputs
than available calibration point references.
The primary outputs that must be matched are empty weight, design gross weight,
volumetric closure, and mission completion. The empty weight is affected by the
avionics and subsystems, structural loading, flight envelope, fuel capacity, payloads, and
numerous other inputs. The design gross weight is a function of empty weight and the
fuel weight. Volumetric closure is a function of component volumes, fuel volume and
available volume. The mission completion is a function of weights, aerodynamics
performance, propulsion performance, and the selfoptimized mission profile.
The calibration process starts with the following inputs:
1. Vehicle Geometry
2. Loads and flight envelope parameters
3. Component SWaP or SWaP method inputs
4. Fuel fraction input
5. Structural materials characteristics
6. Subsystems parameters and characteristics
7. Mission profile
The vehicle design code is run with all available inputs. If the empty weight is higher
than the reference value, then several items may be adjusted:
1. Factor of safety design level (if not specified)
2. Weight growth design level
3. Installation weight design level
4. Avionics and subsystems characteristics (if appropriate)
5. Structural materials characteristics (if appropriate)
143
Once the empty weight at the reference fuel mass fraction is calibrated, the performance
calibration begins. The fuel mass fraction is held constant at the reference value. The
following items may be adjusted to ensure that the design mission profile is completed:
1. Uninstalled engine characteristics (if unknown)
2. Propulsion installation losses
3. Aerodynamics design levels
If volumetric constraints are violated, then the volumetric design level must be adjusted.
If this is not sufficient, then the external geometry of the vehicle and component volume
estimations must be scrutinized for validity.
The parameters adjusted during the calibration process and any unique methods
employed are identified for each of the three cases evaluated here. The calibration cases
utilized the calibration process outlined above.
10.3 Calibration Cases
10.3.1 Global Hawk
10.3.1.1 Global Hawk Description
The Global Hawk is the first and only operational strategic high altitude UAV. This
system began development in 1994. Global Hawk started as an Advanced Concept
Technology Demonstrator (ACTD) with many goals, but the only firm requirements was
a fixed Unit Flyaway Price (UFP). Many modifications have occurred to improve the
system, and it has experienced operational use in wartime. Therefore, the available
performance numbers represent the estimated performance of the vehicle as built, not
necessarily as designed. As with nearly any aircraft program, the performance changes
over time due to weight growth, system modifications, and other considerations. An
attempt is made to calibrate the code against a representative Global Hawk.
10.3.1.2 Geometry Characteristics
A rendering of Global Hawk is shown in Figure 101, and major geometry characteristics
are described in Table 101.
144
Figure 101 Global Hawk
Item Value Source
Wing span 116.2 ft Janes [1999]
Aspect ratio 25 Janes [1999]
Sweep (quarter chord) 5
o
Janes [1999]
Fuselage Length 44.4 ft Janes [1999]
Table 101 Global Hawk Geometry
Detailed geometry characteristics were found through scaling of 3view drawings. The
results were integrated into the detailed geometry input files.
10.3.1.3 Propulsion
The Global Hawk engine is the RollsRoyce 3007H. Major engine characteristics are
shown in Table 102.
Item Value Source
Thrust (TO S/L) 8,290 lbs Janes [1999]
TSFC 0.33 lbm/lbh Janes [1999]
Weight (Dry) 1,581 lbs Janes [1999]
Length 8.88 ft Janes [1999]
Diameter 3.63 ft Janes [1999]
Table 102 Global Hawk Propulsion
In addition to the engine, an additional 50 pounds of propulsion weight was added to
account for the engine control electronics and actuators, as an assumption.
145
The overall propulsion installation thrust installation factor is assumed to be 0.8, and the
thrust specific fuel consumption factor is assumed to be 0.65. These calibration factors
were selected through the performance matching process.
The calculated propulsion performance is shown in Figure 102 and Figure 103.
Global Hawk
Installed Thrust Performance
0
1000
2000
3000
4000
5000
6000
7000
0 100 200 300 400
Velocity, kts
M
a
x
i
m
u
m
T
h
r
u
s
t
,
l
b
s
Alt = 0 kft
Alt = 16kft
Alt = 33 kft
Alt = 49 kft
Alt = 65 kft
Figure 102. Global Hawk Engine Thrust Performance
Figure 103 Global Hawk Engine Specific Fuel Consumption Performance
10.3.1.4 Avionics
Global Hawk is known to have an extensive electronics suite. Weights for all of the
components are not available. Details of some avionics components, such as INS and
data recorders, are found in Global Hawk literature and vendor data sheets. The assumed
avionics weights use a fragmentary Master Equipment List (MEL), developed from
information generated from Altmann [2002] and Janes [1999], as guidance. Figure 104
shows the avionics weights determined for the calibration case.
146
Global Hawk Avionics Weight
0
50
100
150
200
250
300
350
General
Avionics
Autopilot INS GPS ADS ATC Recorder
W
e
i
g
h
t
,
l
b
s
Figure 104. Global Hawk Avionics Weights Summary
General Avionics
The general avionics category is intended to capture all avionics components not covered
by specific avionics categories. 300 pounds of general avionics was assumed. This
weight was the bestjudgment assessment of the weight of all avionics for which detailed
weight information is not available.
Autopilot
The autopilot weight consists of two Integrated Mission Management Computers
(IMMCs). No weights information is readily available on these units. The weight is 40
pounds, as a bestjudgment assumption.
INS
The INS weight consists of two LN100 INS units (21.6 pounds each), and two Kearfott
KN4072 units (10 pounds each).
GPS
The GPS weight consists of the GPS antenna and a GPS processor. The GPS must
support Differential GPS (DGPS) functionality. The GPS weight was set to 5 pounds,
which is a bestjudgment assumption.
Air Data System
The air data system consists of two air data booms on the tails. The custom air data
system weight equation was used. The equation is set to have the largest size, most
rugged design, and highest performance settings.
Air Traffic Control
147
Global Hawk has a robust air traffic control system. Global Hawk is the first UAV to
gain a type certification, enabling a file and fly capability. Global Hawk is known to
have VHF/UHF ATC radios, and Identify Friend or Foe (IFF) avionics [Altmann 2002].
The ATC weight was set to 50 pounds, which is a bestjudgment assumption.
Recorder
Global Hawk has a data recorder capability. L3 claims that its S/TAR family of data
recorders is installed on the Global Hawk [L3 2003]. These recorders have an
advertised weight range of 1933 pounds, and a weight of 25 pounds is selected for the
Global Hawk calibration. It is assumed that two of these units are installed, given the
significant redundancy on the aircraft.
10.3.1.5 Subsystems
Global Hawk has a complex set of subsystems. A list of known subsystems identified by
Altmann and Janes is captured in the simple MEL. Unfortunately, no weights data is
available for the subsystems. Therefore, no actual weights were used, only assumed
subsystem weights and parametric methods. The resulting subsystems weights are shown
in Figure 105.
Global Hawk Subsystems Weight
0
50
100
150
200
250
300
350
400
450
FCS ECS Electrical System Fuel System Communications
W
e
i
g
h
t
,
l
b
s
Figure 105 Global Hawk Subsystems Weights Summary
Flight Control System (FCS)
Global Hawk has a redundant flight control system. Per vehicle side, Global Hawk has
two ailerons, two spoilers, and two ruddervators. Since the flight control system includes
the actuators, the weight was assumed to be twice the parametric flight control system
weight to account for the redundancy. The custom FCS weight estimation equation
(Flight Control Method 7) was used to estimate this weight.
148
Electrical System
Global Hawk has a redundant electrical power system. There are two flight critical
buses, two start buses, three batteries, an AC and a DC generator, power converters, and
interfaces to ground power. The Raymer [1992] transport electrical system equation
(Electrical System Method 6) was used, with two generators input.
Fuel System
Global Hawk has a sophisticated fuel system that interacts with the environmental control
system. The aircraft has two high capacity fuel tanks and one sump tank in the fuselage,
and three wing tanks. In addition to feeding the engine, the fuel is routed through various
avionics, payload, and other bays to provide thermal management. The fuel is actively
pumped through the system. The Raymer [1992] transport fuel system weight equation
(Fuel System Method 2) was employed.
Environmental Control System (ECS)
The environmental control system is linked to the fuel system. Fuel cools the avionics
and other components via heat exchangers and other means. The Roskam [1990]
General Dynamics lowsubsonic aircraft ECS weight method (ECS Method 2) was
employed. The performance model takes no drag penalty or propulsion penalty for the
ECS system, since there is no ram air or engine bleed air used for thermal management.
Communications System
The Global Hawk communication system consists of multiple communications links and
associated equipment. There is a high bandwidth satellite communications link using a
48inch parabolic dish for payload data downlink. There are two low bandwidth satellite
communications links for command and control. Global hawk has air traffic control links
that are accounted for under ATC equipment weights. There is a lineofsight
communications link between the aircraft and the launch and recovery element to support
launch and recovery operations.
The custom dish weight equation is used to estimate the high bandwidth link parabolic
dish weight. Additionally, 100 pounds of communications equipment weight was input
to estimate the weight of all other communications equipment, which is a bestjudgment
assumption used in the absence of better information.
10.3.1.6 Structures
The Global Hawk structure consists of the main wing, tails, fuselage, nacelle, landing
gear and installation weight. No direct weights data is available for the structure.
However, Altmann [2002] provides useful information to describe the structural design
drivers and philosophy. The factor of safety for the structure is 1.25. Altmann provides a
149
VN diagram that indicates that the light weight vertical load is approximately 3.6 G, and
the heavy weight vertical load is approximately 2 G. Because the wing weight is
calculated at gross weight, the vertical load is assumed to be 2 G. The Global Hawk
structural weight is presented in Figure 106.
Global Hawk Structural Weight
0
500
1,000
1,500
2,000
2,500
Wing Tail Fuselage Nacelle Gear Installation
W
e
i
g
h
t
,
l
b
s
Figure 106 Global Hawk Structural Weight Summary
Wing
The wing is made of Carbon Fiber Reinforced Plastic (CFRP) [Jane’s 1999]. The
piecewise linear beam model is used to estimate the wing weights. A “black aluminum”
formulation was used. The base material is the 2017 Aluminum alloy with a minimum
gage of 0.02 inches. A composites weight reduction factor of 0.85 was used, as
suggested by Raymer [1992].
Tails
The tail consists of a single Vtail made of unspecified composites [Jane’s 1999]. The
piecewise linear beam model is employed. As with the wing, 2017 aluminum with a
minimum gage of 0.02 inches was assumed for the reference material. A composites
weight reduction factor of 0.83 was used for the tails, as suggested by Raymer [1992].
Fuselage
The fuselage is essentially an aluminum alloy box structure [Jane’s 1999]. The Raymer
[1992] General Aviation method was used. No credit was provided for composites.
Nacelle
A single nacelle houses the Rolls Royce 3007H engine. The Raymer [1992] transport
nacelle equation was used for weight estimation. The resulting nacelle weight may be
low, though there is no available weights information for comparison.
Landing Gear
150
The Global Hawk landing gear consists of a retractable nose gear mounted in the fuselage
and a twostrut retractable main gear mounted in the wing root. The custom gear weight
equation was used.
10.3.1.7 Payloads
Global Hawk payloads consist of a Synthetic Aperture Radar (SAR), an ElectroOptical/
Infrared (EO/IR) payload, and the supporting electronics. The supporting electronics
include an integrated sensor processor, a receiver/exciter/controller unit, transmitter (for
SAR, presumably), and a sensor electronics unit. It is unclear if the elements of the
communications architecture, INS, or structure are included in the advertised payload
weight of 1,900 pounds [Jane’s 1999]. The summation of the listed components comes to
797 pounds [Jane’s 1999]. There is no available source that clarified this discrepancy.
To satisfy the sizing mission profile, 1,900 pounds was assumed for the total payload
weight, with an even weight division between SAR and EO/IR.
10.3.1.8 Overall Weights
The available information in Jane’s [1999] on Global Hawk weights does not contain any
noticeable internal discrepancies. According to Jane’s [1999], the empty weight is 9,200
pounds, the maximum fuel weight is 14,500 pounds, the payload weight is 1,900 pounds,
and the maximum takeoff weight is 25,600 pounds. The sum of the components
comprising the maximum takeoff weight equals the maximum takeoff weight.
There is conflicting information from other sources on Global Hawk weights. Northrop
Grumman [2003] lists the gross takeoff weight at 26,700 pounds. Due to typical weight
growth of aircraft, both weights may be true for the associated reporting time. The Jane’s
[1999] weight is assumed as correct for the purposes of sizing.
The payload weight of 1,900 pounds, and the fuel fraction associated with Jane’s [1999]
fuel weight and maximum takeoff weight were input. The target design gross weight is
assumed to be equal to the maximum takeoff weight. The calibration process sought to
match the empty weight and design gross weight. The empty weight matches within
0.2%, and the design gross weight matches within 0.2%.
The Global Hawk weight summary is shown in Figure 107.
151
Global Hawk Weight Summary
1
10
100
1,000
10,000
100,000
A
v
io
n
ic
s
S
u
b
s
y
s
te
m
s
P
r
o
p
u
ls
io
n
S
tr
u
c
t
u
r
e
s
P
a
y
lo
a
d
F
u
e
l
G
r
o
w
t
h
E
m
p
t
y
D
e
s
ig
n
G
r
o
s
s
W
e
i
g
h
t
,
l
b
s
Figure 107 Global Hawk Weights Summary
The mass fractions are shown in Figure 108.
Global Hawk Mass Fraction Summary
0%
10%
20%
30%
40%
50%
60%
A
v
i
o
n
ic
s
S
u
b
s
y
s
te
m
s
P
r
o
p
u
ls
io
n
S
t
r
u
c
t
u
r
e
s
P
a
y
lo
a
d
F
u
e
l
G
r
o
w
t
h
E
m
p
t
y
M
a
s
s
F
r
a
c
t
i
o
n
Figure 108 Global Hawk Mass Fraction Summary
10.3.1.9 Aerodynamics
The predicted Global Hawk incompressible trimmed Lift to Drag (L/D) over the range of
Reynolds number per unit length is shown in Figure 109.
152
Global Hawk
Incompressible Aerodynamic Performance
0
5
10
15
20
25
30
35
0 0.2 0.4 0.6 0.8 1 1.2
Lift Coefficient
L
i
f
t
t
o
D
r
a
g
R
a
t
i
o
Re/L = 73,547
Re/L = 87,133
Re/L = 124,250
Re/L = 174,952
Figure 109 Global Hawk Trimmed Incompressible Aerodynamic Performance
The Global Hawk drag contributors by drag type at Mach 0.7 at the maximum bounding
Reynolds number are detailed in Figure 1010.
Global Hawk
Drag Contributors, Mach = 0.7
0.000
0.002
0.004
0.006
0.008
0.010
0.012
0.014
0.016
0 0.2 0.4 0.6 0.8 1 1.2
Lift Coefficient
D
r
a
g
C
o
e
f
f
i
c
i
e
n
t
Cdprof
CD ind
CD fuse
CD int
CD Wave
Figure 1010 Global Hawk Trimmed Drag Contributors
Several aerodynamics methods were used to generate the drag polar. The Trefftz plane
analysis was used to determine the trimmed lift distribution and induced drag. The
parametric airfoil method was used to determine the profile drag. A static margin of 5%
was assumed. A 5% miscellaneous drag factor was applied. All other aerodynamics
methods are standard.
153
10.3.1.10 Performance
Altmann [2002] provides useful information on the Global Hawk performance and flight
envelope limitations. The maximum equivalent airspeed is 175 Keas, and the maximum
Mach is approximately Mach 0.7.
Northrop Grumman advertises the Global Hawk Performance as 24 hours time on station
at 1,200 nautical miles radius [Northrop 2003]. This performance estimate was adopted
for sizing. Range credit was assumed to be 100 nautical miles for the initial climb to
50,000 feet, and 200 nautical miles from the end of cruise to the final loiter altitude. A
halfhour loiter at 5,000 feet was assumed for airfield operations. Altitude and Mach
characteristics are shown in Figure 1011, Figure 1012, and Figure 1013. Note that
Figure 1013 shows the Mach and altitude points flown in the optimized mission profile,
not the boundaries of the flight envelope. Figure 1013 shows the same data points as
Figure 1011 and Figure 1012. Sudden changes in the Machaltitude profile occur
during transitions from one mission segment to another, such as changing from loiter to
egress.
Global Hawk Altitude Profile
0
10,000
20,000
30,000
40,000
50,000
60,000
70,000
0 5 10 15 20 25 30 35
Time, hrs
A
l
t
i
t
u
d
e
,
f
t
Figure 1011 Global Hawk Altitude Profile
Global Hawk Mach Profile
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0 5 10 15 20 25 30 35
Time, hrs
M
a
c
h
Figure 1012 Global Hawk Mach Profile
154
Global Hawk MachVelocity Envelope Used
0
10,000
20,000
30,000
40,000
50,000
60,000
70,000
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Mach
A
l
t
i
t
u
d
e
,
f
t
Figure 1013 Global Hawk MachAltitude Envelope
10.3.1.11 Design Technology
The weights and performance calibration process resulted design technology levels are
shown in Table 103.
Design Item Tech Level (01)
Volume Efficiency 0.5
Induced Drag 0.31
Interference Drag 1.0
Wave Drag 0.31
Laminar Flow 0.31
Factor Of Safety 1.0
Weight Growth 0.35
Installation Weight 0.5
Table 103 Global Hawk Design Technology Levels
155
10.3.2 Predator
10.3.2.1 Predator Description
Predator is a MediumAltitude Endurance (MAE) UAV designed to provide battlefield
surveillance with a beyond line of sight communications capability. This aircraft is an
evolution from the General Atomics Gnat UAV. The Predator program began in 1994 as
an ACTD. The program transitioned to operational use very early in development.
10.3.2.2 Geometry Characteristics
The Predator key geometry characteristics are shown graphically in Figure 1014, and
numerically in Table 104.
Figure 1014 Predator
Item Value Source
Wing span 48.7 ft Jane’s [1999]
Aspect ratio 19.25 Jane’s [1999]
Sweep (quarter chord) 0
o
Jane’s [1999]
Fuselage Length 26.7 ft Jane’s [1999]
Table 104 Predator Geometry Characteristics
156
10.3.2.3 Propulsion
Predator uses the Rotax 914 reciprocating engine to drive a pusher propeller. Major
engine characteristics are presented in Table 105.
Item Value Source
Maximum Power (S/L) 105 HP Jane’s [1999]
BSFC 0.5 lbm/HPhr Assumed
Weight 150.4 lbs Jane’s [1999]
Table 105 Predator Propulsion Characteristics
The predicted installed predator engine performance is shown in Figure 1015 and Figure
1016.
Predator
Installed Thrust Performance
0
50
100
150
200
250
300
350
0 50 100 150
Velocity, kts
M
a
x
i
m
u
m
T
h
r
u
s
t
,
l
b
s
Alt = 0 kft
Alt = 6.25 kft
Alt = 12.5 kft
Alt = 18.75 kft
Alt = 25 kft
Figure 1015 Predator Installed Thrust Performance
157
Predator
Installed Thrust Specific Fuel Consumption
0
0.1
0.2
0.3
0.4
0.5
0 50 100 150
Velocity, kts
T
S
F
C
,
l
b
m
/
l
b

h
r
Alt = 0 kft
Alt = 6.25 kft
Alt = 12.5 kft
Alt = 18.75 kft
Alt = 25 kft
Figure 1016 Predator Installed Specific Fuel Consumption Performance
10.3.2.4 Avionics
Predator has a relatively simple avionics suite compared to Global Hawk. Predator is
largely a singlesting system with little redundancy. A summary of the Predator avionics
weights is presented in Figure 1017.
Predator Avionics Weights
0
5
10
15
20
25
General Autopilot INS GPS ADS ATC Recorder
W
e
i
g
h
t
,
l
b
s
Figure 1017 Predator Avionics Weights Summary
General Avionics
The Predator general avionics category is intended to capture nonallocated avionics
weights. The general avionics weight allocation was 5 pounds, which was a best
judgment assumption used in the absence of better information.
158
Autopilot
Details of the autopilot physical characteristics are not available. The Predator autopilot
must support both autonomous flight and pilot commanded maneuvering. The autopilot
weight allocation was 10 pounds, which was a bestjudgment assumption used in the
absence of better information.
INS
Predator has a single Litton LN100G INS, weighing 21.6 pounds.
GPS
Predator has a GPS capability, and 1 pound was allocated to the receiver. Unlike Global
Hawk, Predator does not land autonomously with the aid of DGPS.
Air Data System
The air data system (ADS) was calculated via the custom method. The ADS is assumed
to have mediumlow ruggedness, a mediumhigh level of miniaturization, and medium
low performance. The resulting ADS weight is 4.28 pounds.
Air Traffic Control
Predator has a VHF/UHF ATC radio and an IFF capability [Jane’s 1999]. The ATC
weight was assumed to be 5 pounds, based on the weight class of simple general aviation
aircraft ATC equipment. The ATC weight is assumed to be avionics weight, though it is
possible that it may be considered payload weight. The ATC equipment may not be
permanent avionics.
Recorder
The data recording capability of the Predator is unknown. The avionics data recorder
weight allocation was 1 pound, which was a bestjudgment assumption. Any additional
data recorder weight is assumed to be covered under payload weights.
10.3.2.5 Subsystems
A summary of the Predator subsystems weights is presented in Figure 1018.
159
Predator Subsystems Weights
0
5
10
15
20
25
30
35
40
45
FCS ECS Electrical Fuel Sys Comms
W
e
i
g
h
t
,
l
b
s
Figure 1018 Predator Subsystems Weights Summary
Flight Control System
The Predator flight control system consists of inboard and outboard wing trailing edge
control surfaces and allflying ruddervators. All control surfaces are electrically actuated.
The custom flight control system method (FCS Method 7) was employed to estimate this
weight.
Electrical System
The electrical system consists, at minimum, of a generator and power distribution wiring.
The Raymer [1992] transport electrical system weights equation (Electrical System
Method 6) was employed. The calculated nominal power load is 2.81 kW.
Fuel System
Little is known from available literature of the Predator fuel system. The Raymer [1992]
transport fuel system weight equation (Fuel System Method 2) was used.
Environmental Control System
The Predator ECS system is assumed to be a ramair system. The performance model
includes the impacts of the ram air on aerodynamic efficiency. The Roskam [1990] 
General Dynamics lowsubsonic aircraft ECS weight equation (ECS Method 2) was used.
Communications
Predator has a sophisticated communications system. The Predator has a satellite
communications capability for command and control and imagery downlink. It also has
the ability to communicate within lineofsight range to a launch and recovery ground
station.
160
Predator uses a 30inch parabolic dish for satellite communications. The custom dish
weight estimation method is used to calculate the weight, volumetric, and power impacts.
Predator includes communications equipment. All or part of the equipment is produced
by Loral. This equipment is assumed to include the Limited Capability Common Data
Link, which weighs 18.3 pounds.
10.3.2.6 Structures
The structure is largely made of carbon/epoxy composites [Jane’s 1999]. The smaller
Gnat UAV in the Predator family is stressed for 6 G maneuvers at an unspecified weight.
Absent of further information, the 6 G loading was applied to the Predator. A summary
of the Predator structural weights is presented in Figure 1019.
Predator Structural Weight
0
20
40
60
80
100
120
140
160
180
Wing Tail Fuselage Gear Installation
W
e
i
g
h
t
,
l
b
s
Figure 1019 Predator Structural Weight Summary
Wing
The piecewise linear beam model was applied to the wing structure. The wing was
assumed to weigh 75% of the equivalent aluminum wing with a minimum gage of 0.02
inches. The 75% composites factor was attained through the calibration process, since
the empty weight could not be sufficiently reduced through other parameters. The 75%
composites multiplication factor applied to the wing, tails, and fuselage was estimated
through the calibration process.
Tails
The tail weight was estimated with the piecewise linear beam model. The tail was
assumed to weight 75% of the equivalent aluminum tail with a minimum gage of 0.02
inches. The factor of 75% was matched to the wing factor, since the wing and tails were
probably designed by General Atomics with similar methods.
161
Fuselage
The fuselage weight was estimated with the Raymer General Aviation fuselage weight
equation. A factor of 75% was applied to account for the extensive use of composites in
the fuselage. Again, the 75% factor was identical to that of the wings and tails due to the
assumed similarities in the design approach.
Landing Gear
The landing gear weight was estimated with the custom tricycle landing gear weight
method.
Payloads
Predator is known to carry both EO/IR and SAR payloads. The advertised payload
capacity is 450 pounds. The weights of the actual payloads appear to be much less than
the maximum payload weight. The Versatron Skyball EO/IR payload with supporting
electronics weighs 66 pounds. The Lynx SAR payload weighs 115 pounds. These two
payloads were incorporated for the purposes of the sizing mission.
10.3.2.7 Overall Weights
Jane’s [1999] lists the Predator Empty weight at 772 pounds, a fuel weight of 650
pounds, and a maximum payload weight of 450 pounds. The sum of these weights comes
to 1,872 pounds. Jane’s lists the maximum takeoff weight at 2,250 pounds. Jane’s
[1999] does not state what payload is carried for the mission. The empty weight and fuel
weight from Jane’s are assumed to be valid, and these values were applied directly. The
payload weight incorporated for the mission profile and the weights sizing is the sum of
the known EO/IR and SAR payload weights, which totals 181 pounds. The target design
gross weight is 1,603 pounds. The calibrated Predator analysis design gross weight is
within 0.5% of this value. The calibrated empty weight is within 0.6% of the target
value. The predator weight and mass fraction summaries are presented in Figure 1020
and Figure 1021, respectively.
162
Predator Weight Summary
1
10
100
1,000
10,000
A
v
io
n
ic
s
S
u
b
s
y
s
t
e
m
s
P
r
o
p
u
l
s
i
o
n
S
t
r
u
c
t
u
r
e
s
P
a
y
lo
a
d
F
u
e
l
G
r
o
w
t
h
E
m
p
t
y
D
e
s
ig
n
G
r
o
s
s
W
e
i
g
h
t
,
l
b
s
Figure 1020 Predator Weights Summary
Predator Mass Fractions
0%
10%
20%
30%
40%
50%
60%
A
v
io
n
ic
s
S
u
b
s
y
s
t
e
m
s
P
r
o
p
u
l
s
i
o
n
S
t
r
u
c
t
u
r
e
s
P
a
y
lo
a
d
F
u
e
l
G
r
o
w
t
h
E
m
p
t
y
M
a
s
s
F
r
a
c
t
i
o
n
Figure 1021 Predator Mass Fraction Summary
10.3.2.8 Aerodynamics
The predicted aerodynamics characteristics of the Predator are shown in Figure 1022 and
Figure 1023.
163
Predator
Aerodynamic Performance
0
5
10
15
20
25
30
0 0.2 0.4 0.6 0.8 1 1.2
Lift Coefficient
L
i
f
t
t
o
D
r
a
g
R
a
t
i
o
Re/L = 167K
Re/L = 227K
Re/L = 391K
Re/L = 615K
Figure 1022 Predator Trimmed Aerodynamic Performance
The Predator drag contributors at the maximum bounding Reynolds number are shown in
Figure 1023. Note that wave drag is not shown here, since this is not a consideration at
these flight Mach numbers.
Predator
Drag Contributors
0.000
0.005
0.010
0.015
0.020
0.025
0 0.2 0.4 0.6 0.8 1 1.2
Lift Coefficient
D
r
a
g
C
o
e
f
f
i
c
i
e
n
t
CDprof
CDind
CDfuse
CDint
Figure 1023 Predator Trimmed Drag Contributors
10.3.2.9 Performance
The mission profile for Predator is 24 hours time on station at a 500 nautical mile radius,
according to Jane’s [1999]. The 2003 General Atomics Predator brochure indicates that
the performance is 24 hours time on station at a 400 nautical mile radius. The Jane’s
[1999] mission profile was used here. The EO/IRSAR payload combined weight of 181
pounds was used, not the maximum payload capacity. An additional onehour loiter at
sea level is added to account for recovery operations. A ceiling of 25,000 feet was
imposed on the mission performance calculation. A climb from sea level to 20,000 feet
was included in the ingress segment. The descent from the final loiter point to sea level
164
was included in the egress segment. The Predator altitude and velocity performance is
shown in Figure 1024, Figure 1025, and Figure 1026. Note that Figure 1026 is a
combination of the data points shown in Figure 1024 and Figure 1025. Figure 1026
does not represent the limits of the flight envelope, only velocityaltitude points flown in
the optimized mission.
Predator Altitude Profile
0
5,000
10,000
15,000
20,000
25,000
30,000
0 10 20 30 40
Time, hrs
A
l
t
i
t
u
d
e
Figure 1024 Predator Altitude Profile
Predator Velocity Profile
0
20
40
60
80
100
120
0 10 20 30 40
Time, hrs
V
e
l
o
c
i
t
y
,
k
t
s
Figure 1025 Predator Velocity Profile
165
Predator VelocityAltitude Envelope Used
0
5000
10000
15000
20000
25000
30000
40 50 60 70 80 90 100 110
Velocity, kts
A
l
t
i
t
u
d
e
,
f
t
Figure 1026 Predator VelocityAltitude Envelope
10.3.2.10 Design Technology
The weights and performance calibration process resulted design technology levels
shown in Table 106.
Design Item Tech Level (01)
Volume Efficiency 0.5
Induced Drag 0.4
Interference Drag 1.
Wave Drag 1 (No compressibility impacts)
Laminar Flow 0.4
Factor Of Safety 1.
Weight Growth 0.75
Installation Weight 1.
Table 106 Predator Design Technology Levels
166
10.3.3 Shadow 200
10.3.3.1 Shadow 200 Description
Shadow 200 is a small tactical UAV designed to support lineofsight battlefield
surveillance missions. Initial development began in 1990. However, the technology year
was assumed to be 2000 due to the extended development time, significant design
evolution, requirements changes, and incorporation of more advanced technologies.
Palumbo [2000] is assumed to be the most authoritative source of Shadow 200 data.
Palumbo describes an evolutionary design history beginning in 1990 that has not ended.
The design attempts to leverage legacy components and tooling, and does not represent a
clean sheet design synthesis. For example, the wing configuration is driven by a
constraint to reuse Pioneer program wing tooling.
10.3.3.2 Geometry Characteristics
The Shadow 200 geometry characteristics are shown graphically in Figure 1027, and
numerically in Table 107.
Figure 1027 Shadow 200
Item Value Source
Wing span 12.75 ft Jane’s [1999]
Aspect ratio 7.07 Jane’s [1999]
Sweep (quarter chord) 0
o
Jane’s [1999]
Overall Length 11.17 ft Jane’s [1999]
Table 107 Shadow 200 Geometry Characteristics
167
10.3.3.3 Propulsion
The Shadow 200 uses a UEL AR 741 rotary engine. The engine drives a twoblade
propeller. The engine weight listed includes the alternator. Major characteristics of the
Shadow 200 engine are shown in Table 108.
Item Value Source
Maximum Power (S/L) 38 HP Jane’s [1999]
BSFC (Max power, S/L) 0.57 lbm/HPhr Jane’s [1999]
Weight 28 lbs Palumbo [2000]
Propeller Diameter 2.33 ft Jane’s [1999]
Table 108 Shadow 200 Propulsion Characteristics
The thrust and efficiency loss factor was found to be 0.59 through the calibration process.
The aerodynamics technologies are already very conservative, so the only remaining
performance factor for modification is the propulsion losses. Figure 1028 and Figure
1029 show the Shadow 200 predicted installed propulsion characteristics.
Shadow 200
Installed Thrust Performance
0
10
20
30
40
50
60
70
80
90
0 50 100 150
Velocity, kts
M
a
x
i
m
u
m
T
h
r
u
s
t
,
l
b
s
Alt = 0 kft
Alt = 3.75 kft
Alt = 7.5 kft
Alt = 11.25 kft
Alt = 15 kft
Figure 1028 Shadow 200 Installed Thrust Performance
168
Shadow 200
Installed Thrust Specific Fuel Consumption
0
0.1
0.2
0.3
0.4
0.5
0.6
0 50 100 150
Velocity, kts
T
S
F
C
,
l
b
m
/
l
b

h
r
Alt = 0 kft
Alt = 3.75 kft
Alt = 7.5 kft
Alt = 11.25 kft
Alt = 15 kft
Figure 1029 Shadow 200 Installed Specific Fuel Consumption Performance
10.3.3.4 Avionics
The Shadow 200 uses a relatively simple avionics suite. There is no indication that any
of the avionics components are redundant.
General Avionics
Palumbo does not provide a detailed breakout of avionics weights. The total weight
allocated to avionics is 57 pounds, which includes avionics, communications equipment,
and elements of the electrical system [Palumbo 2000]. The avionics, at minimum,
includes the autopilot and a Mode IV IFF transponder. No further information about the
avionics suite is provided by Palumbo.
For the purposes of this analysis, 30 pounds was allocated to avionics and 27 pounds was
allocated to communications equipment. The electrical system weight was set to 0
pounds, and the alternator weight was included in the propulsion weight.
The avionics suite apparently has evolved since 2000 to include more modern equipment.
The Shadow 200 is known to use the Tactical Automatic Landing System (TALS), which
has an airborne component weighing 3 pounds. The Shadow 200 currently uses the
Athena GS211 GuideStar
TM
autopilot, which weighs 2 pounds. This autopilot includes
an INS and air data measurement equipment. The allocated weight of 30 pounds is
applied, due to the year 2000 design year assumption.
10.3.3.5 Subsystems
An overview of the Shadow 200 subsystems weight results is in Figure 1030.
169
Shadow 200 Subsystems Weights
0
5
10
15
20
25
30
FCS ECS Fuel Communications
W
e
i
g
h
t
,
l
b
s
Figure 1030 Shadow 200 Subsystems Weights Summary
Flight Control Systems
The Shadow 200 Flight Control System (FCS) consists of one aileron per wing and Vtail
ruddervators. The custom FCS weight method (FCS Method 7) was used to estimate
FCS weight.
Electrical System
The electrical system weight was set to 0 to accommodate the weight accounting system
differences between this model and the weights described by Palumbo [2000]. The
Palumbo [2000] alternator weight is included in the engine weight. Other elements of the
electrical system weights are captured in the avionics weight, due to the weights
accounting provided by Palumbo [2000]. The electrical system weight is fully accounted
for by the engine and avionics weight categories.
Fuel System
The available literature provides little information on the Shadow 200 fuel system. The
Shadow 200 wing tanks are believed to be selfsealing. The Raymer [1992] Transport
fuel system weight equation (Fuel System Method 2) was used, despite being outside the
applicable range. The fuel system numbers appear to be reasonable.
Environmental Control System
The available literature provides no information on the Shadow 200 Environmental
Control System (ECS). A ramair ECS was assumed, and the associated performance
impacts are included in the flight performance model. The Roskam [1990]  General
170
Dynamics lowsubsonic aircraft ECS weight method (ECS Method 2) was used. This
weight method appears to scale well to small UAVs.
Communications
The Shadow 200 is equipped with the Tactical Common Data Link, which is a moderate
bandwidth lineofsight communications link. No actual weights information is known
for this system, but 27 pounds was assumed, based on the assumed allocation between
avionics and communications weights described earlier. Separate communications relay
payloads may be integrated, but they are not baseline avionics, and they were not
included here.
10.3.3.6 Structures
The Shadow 200 structure is 90% composites, which is primarily composed of graphite
and Kevlar epoxy [Jane’s 1999]. Jane’s [1999] lists the limit load at 3.6 G, and Palumbo
[2000] lists the limit load at 3.8 G. The limit load of 3.8 G was applied here. This limit
was applied directly at the design gross weight condition, since the fuel fraction is
relatively low on this UAV. The Shadow 200 structural weight estimations are
summarized in Figure 1031.
Shadow 200 Structural Weights
0
5
10
15
20
25
30
35
40
Wing Tails Fuselage Gear Installation
W
e
i
g
h
t
,
l
b
s
Figure 1031 Shadow 200 Structural Weight Summary
Wing
The piecewise linear beam model was used to determine the wing weight. The reference
material is the 2017 Aluminum alloy with a minimum gage of 0.02 inches. A composites
wing weight multiplication factor of 0.85 was used, as suggested by Raymer [1992].
Tails
171
The piecewise linear beam model was used to determine the tail weight. The reference
material is the 2017 Aluminum alloy with a minimum gage of 0.02 inches. A composites
tail weight multiplication factor of 0.83 was used, as suggested by Raymer [1992].
Fuselage
The shadow 200 fuselage consists of a main pod and two booms that connect the tail to
the wing. The primary contributor to the weight is the main pod. This houses the engine,
payload, landing gear, most subsystems, and the avionics. The Raymer [1992] General
Aviation fuselage weight equation was used to estimate the fuselage weight. Although
this equation is operating outside of its intended application, the resulting weight appears
to be reasonable when combined with the overall installation weight. The installation
weight was primarily added to the fuselage weight. A 0.9 fuselage weight multiplication
factor was applied to account for composites usage, as recommended by Raymer [1992].
Landing Gear
The Shadow has both fixed landing gear and a parachute recovery capability. The
standard landing gear weight equation is used to account for the fixed landing gear
weight. The gear drag contributions are accounted for in the aerodynamics analysis. The
parachute recovery gear was estimated using the custom parachute recovery system
weight method.
10.3.3.7 Payloads
The maximum payload weight is 52 pounds [Palumbo 2000]. This weight was assumed
to be split evenly between EO/IR and SAR functionality for the purposes of calibration
and technology modeling.
10.3.3.8 Overall Weights
Palumbo [2000] provides group weights for structures, avionics, propulsion, payload, and
fuel. The structural weight in Palumbo appears to contain many elements of the
subsystems weights. 27 pounds of avionics weights are moved to subsystems to account
for the communications subsystems assumptions. These parameters as modified for this
research, the takeoff gross weight (design gross weight here), and the corresponding
values from the analysis are provided in Table 109. The Shadow 200 weight and mass
fraction summaries are presented in Figure 1032 and Figure 1033, respectively.
172
Palumbo
Original/Modified
Analysis Difference
Structure
(struct + subs)
115/142 lbs 144.5 lbs
(Includes wt. growth)
1.77%
Avionics 57/30 lbs 30 lbs 0%
Engine/Alternator 28 lbs 28 lbs 0%
Payload/Ballast 52 lbs 52 lbs 0%
Fuel 64 lbs 64.63 lbs 1.0%
Design Gross Weight 316 lbs 319.1 lbs 0.99%
Table 109 Shadow 200 Calibration weight comparison
Shadow 200 Weight Summary
1
10
100
1000
A
v
i
o
n
ic
s
S
u
b
s
y
s
t
e
m
s
P
r
o
p
u
l
s
i
o
n
S
t
r
u
c
t
u
r
e
s
P
a
y
l
o
a
d
F
u
e
l
G
r
o
w
t
h
E
m
p
t
y
D
e
s
i
g
n
G
r
o
s
s
W
e
i
g
h
t
,
l
b
s
Figure 1032 Shadow 200 Weights Summary
Shadow 200 Mass Fraction Summary
0%
10%
20%
30%
40%
50%
60%
70%
A
v
i
o
n
ic
s
S
u
b
s
y
s
t
e
m
s
P
r
o
p
u
l
s
i
o
n
S
t
r
u
c
t
u
r
e
s
P
a
y
l
o
a
d
F
u
e
l
G
r
o
w
t
h
E
m
p
t
y
M
a
s
s
F
r
a
c
t
i
o
n
Figure 1033 Shadow 200 Mass Fraction Summary
173
10.3.3.9 Aerodynamics
Figure 1034 and Figure 1035 show the estimated aerodynamic performance of the
Shadow 200.
Shadow 200 Lift to Drag Ratio
0
2
4
6
8
10
12
14
16
0 0.2 0.4 0.6 0.8 1 1.2
Lift Coefficient
L
i
f
t
t
o
D
r
a
g
R
a
t
i
o
Re/L = 186K
Re/L = 259K
Re/L = 461K
Re/L = 737K
Figure 1034 Shadow 200 Trimmed Aerodynamic Performance
Note that gear drag was included in the drag contributor summary. The Shadow 200 flies
without retracting the landing gear.
Shadow 200 Drag Contributors
0.00
0.01
0.02
0.03
0.04
0.05
0 0.2 0.4 0.6 0.8 1 1.2
Lift Coefficent
D
r
a
g
C
o
e
f
f
i
c
i
e
n
t
CDprof
CDind
CDfuse
CDint
CDgear
Figure 1035 Shadow 200 Trimmed Drag Contributors
10.3.3.10 Performance
The Shadow 200 flight performance is described differently by various sources. Jane’s
[1999] indicates that the Shadow has an operational radius of 43 nautical miles and 6
hours of endurance, though it is not explicitly states that this occurs on the same flight or
that the maximum payload weight is included. An AAI Shadow 200 brochure presented
174
at the Association of Unmanned Vehicle Systems International (AUVSI) conference on
or before 2003 states that the performance is 56 hours without further details. A second
AAI brochure presented at AUVSI at or before 2003 states that the Shadow 200 can fly
50 ‘klicks’ (kilometers) and remain on station up to 4 hours. The second brochure also
states (on the opposite side) that the endurance at 100 kilometers is 4 hours.
Palumbo [2000] describes the design mission profile as 4 hours time on station at a 200
kilometer radius. Here, an additional 30 minutes of endurance was added at the end of
the mission to account for recovery operations and mission planning uncertainties.
Palumbo [2000] and Jane’s [1999] state the Shadow 200 ceiling is 15,000 feet, which was
applied here. The ingress segment climb was assumed to be from sea level to 12,500
feet. The aircraft then performed a cruise climb until reaching the ceiling. The aircraft
descended from the final loiter altitude to sea level in the egress segment. Figure 1036,
Figure 1037, and Figure 1038 show the Shadow 200 altitude and velocity performance.
Note that Figure 1038 is assembled from the data points of the optimized mission
profile, and does not represent the boundaries of the flight envelope. Rapid changes in
the velocityaltitude profile occur during changes of mission segment, such as
transitioning from loiter to egress.
Shadow 200 Altitude Profile
0
2,000
4,000
6,000
8,000
10,000
12,000
14,000
16,000
0 2 4 6 8
Time, hrs
A
l
t
i
t
u
d
e
,
f
t
Figure 1036 Shadow 200 Altitude Profile
175
Shadow 200 Velocity Profile
0
20
40
60
80
100
120
0 2 4 6 8
Time, hrs
V
e
l
o
c
i
t
y
,
k
t
s
Figure 1037 Shadow 200 Velocity Profile
Shadow 200 VelocityAltitude Envelope Used
0
2,000
4,000
6,000
8,000
10,000
12,000
14,000
16,000
50 60 70 80 90 100
Velocity, kts
A
l
t
i
t
u
d
e
,
f
t
Figure 1038 Shadow 200 VelocityAltitude Envelope
10.3.3.11 Design Technology
The weights and performance calibration process resulted in the design technology levels
shown in Table 1010.
176
Design Item Tech Level (01)
Volume Efficiency 0.5
Induced Drag 0.2
Interference Drag 1.
Wave Drag 1. (No compressible effects)
Laminar Flow 0.2
Factor Of Safety 0.5
Weight Growth 0.35
Installation Weight 0.35
Table 1010 Shadow 200 Design Technology Levels
177
Chapter 11 Reference Year Calibration
Optimization Comparisons
11.1 Overview
Optimized designs using the same requirements and same technology levels as the
calibration cases were developed. The reference year calibration cases were described in
the previous chapter. The reference year calibration and optimization cases are compared
in this chapter.
The calibration cases are inappropriate for optimization, so a specialized parametric
geometry formulation is necessary. The calibrated fuselage geometry will not reshape to
account for wing junctures, radome changes, or other considerations. The calibrated
wings and tails are located and sized based on manual inputs, not parametric
relationships. The parametric geometry has similarities among all cases, so it is not
customized for any one case. Details of the parametric geometry are described in
Chapter 5. The comparison between the reference year optimized parametric and
calibrated designs is necessary to better understand the progression of design into the
future.
There are more similarities than differences between the calibrated and optimized
reference year cases. The avionics, subsystems, and payload inputs are identical. The
technology levels and design levels are the same. There are no differences in analysis
methods, such as aerodynamics, propulsion, mass properties, and performance analysis.
The mission profile and input flight envelope constraints are identical.
The results of the Global Hawk, Predator, and Shadow 200 reference year optimized
solutions are presented. Emphasis is placed on the differences between the calibrated and
optimized solutions.
11.2 Reference Case Comparisons
11.2.1 Global Hawk 1994 Optimum Design
Table 111 shows the Global Hawk active design variables and the associated limiting
bounds. These values were used for all optimized Global Hawk cases without
modification.
178
Design Variable Lower Bound Upper Bound
MF
Fuel
0.4 0.65
b
W
20 140
AR
W
10 30
T/W 0.18 0.4
L
Fuse
15 50
t/c
W
0.09 0.17
l
W
0.2 1.
Λ
W
0
o
15
o
Table 111 Global Hawk Design Variables
The optimized Global Hawk geometry has significant differences relative to the
calibration case. The comparison of the calibrated and optimized 1994 Global Hawk
cases are shown graphically in Figure 111, Figure 112, and Figure 113. The calibrated
Global Hawk case is on the left, and the optimum Global Hawk is on the right for these
figures. The numeric differences in the geometry between these cases are presented in
Figure 114.
The parametric fuselage applied to all optimized Global Hawk cases uses the lower
fairing to mount the wing, whereas the 1994 calibration case uses the lower fairing to
house the SAR payload. The basic parametric fuselage prior to the addition of the
radome and wing fairing has a lower lengthtodiameter ratio than the calibration
fuselage. There is no EO/IR housing in the optimized case. The optimized fuselage has a
slightly shorter fuselage length.
The tail configurations for the two cases are essentially identical. The optimized case
uses the parametric tail volume coefficient method. The tail moment arm is less on the
optimized case, so the tail area is greater.
The optimized wing has a shorter span, lower aspect ratio, and much reduced sweep than
the baseline wing. The optimized Global Hawk has a wing loading 3.5% higher than the
baseline, permitting an efficient higher speed loiter. The overall wing area went down
8.6%. Compressibility impacts do not appear to limit either case.
179
Figure 111 1994 Global Hawk Geometry Comparison – Top View
Figure 112 1994 Global Hawk Geometry Comparison – Side View
180
Figure 113 1994 Global Hawk Geometry Comparison – Isometric View
Global Hawk Geometry Comparison
0
20
40
60
80
100
120
Fuse Length, ft Wing Span, ft Aspect Ratio
Baseline
94 Opt
Figure 114 Global Hawk Geometry Comparison
The primary weights differences between the optimal and baseline cases reside within the
structures and propulsion weights. The optimized Global Hawk has a reduced aspect
ratio and shorter wingspan, resulting in reduced aerodynamic efficiency but also a
reduced structural weight. The avionics and payload weights are identical. The minor
subsystems weight differences are caused primarily driven by the flight control system.
The differences in the weight and mass fraction summaries are shown in Figure 115 and
Figure 116, respectively.
181
Global Hawk Weight Summary
1
10
100
1,000
10,000
100,000
A
v
i
o
n
ic
s
S
u
b
s
y
s
t
e
m
s
P
r
o
p
u
l
s
i
o
n
S
t
r
u
c
t
u
r
e
s
P
a
y
l
o
a
d
F
u
e
l
G
r
o
w
t
h
E
m
p
t
y
D
e
s
i
g
n
G
r
o
s
s
W
e
i
g
h
t
,
l
b
s
Baseline
94 Opt
Figure 115 Global Hawk Summary Weights Comparison
Global Hawk Mass Fraction Summary
0%
10%
20%
30%
40%
50%
60%
70%
A
v
i
o
n
ic
s
S
u
b
s
y
s
t
e
m
s
P
r
o
p
u
l
s
i
o
n
S
t
r
u
c
t
u
r
e
s
P
a
y
l
o
a
d
F
u
e
l
G
r
o
w
t
h
E
m
p
t
y
M
a
s
s
F
r
a
c
t
i
o
n
Baseline
94 Opt
Figure 116 Global Hawk Summary Mass Fraction Comparison
There are no significant differences in the mission profiles. The calibration case achieves
the maximum altitude of 65,000 feet shortly before the optimized case. The lift to drag
ratio of the optimized case is lower than the calibration case, which is compensated for
with the higher fuel fraction. The minor differences in specific fuel consumption are
caused by the cruise Mach number and altitude experienced at a given point in the
mission. The mission performance differences between the calibrated and optimized
1994 Global Hawk cases are presented in Figure 117, Figure 118, Figure 119, Figure
1110, and Figure 1111.
182
Global Hawk Altitude Profile Comparison
0
10,000
20,000
30,000
40,000
50,000
60,000
70,000
0 5 10 15 20 25 30 35
Time, hrs
A
l
t
i
t
u
d
e
,
f
t
Baseline
94 Opt
Figure 117 Global Hawk Altitude Profile Comparison
Global Hawk Mach Profile Comparison
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0 5 10 15 20 25 30 35
Time, hrs
M
a
c
h
Baseline
94 Opt
Figure 118 Global Hawk Mach Profile Comparison
183
Global Hawk MachAltitude Envelope Used
0
10,000
20,000
30,000
40,000
50,000
60,000
70,000
0 0.2 0.4 0.6 0.8
Mach
A
l
t
i
t
u
d
e
,
f
t
Baseline
94 Opt
Figure 119 Global Hawk MachAltitude Envelope Used
Global Hawk
Specific Fuel Consumption Comparison
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0 5 10 15 20 25 30 35
Time, hrs
S
F
C
,
l
b
m
/
l
b

h
r
Baseline
94 Opt
Figure 1110 Global Hawk Specific Fuel Consumption Comparison
184
Global Hawk Lift to Drag Ratio Comparison
0
5
10
15
20
25
30
0 5 10 15 20 25 30 35
Time, hrs
L
i
f
t
t
o
D
r
a
g
R
a
t
i
o
Baseline
94 Opt
Figure 1111 Global Hawk Lift to Drag Comparison
185
11.2.2 Predator 1994 Optimum Design
Table 112 shows the Predator active design variables and the associated limiting bounds.
These values were used for all optimized Predator cases without adjustment.
Design Variable Lower Bound Upper Bound
MF
Fuel
0.25 0.45
b
W
6 55
AR
W
10 30
P/W 0.04 0.1
L
Fuse
5 30
t/c
W
0.09 0.17
l
W
0.2 0.9
Table 112 Predator Design Variables
The Predator 1994 optimal solution exhibits geometry variations relative to the
calibration case. The geometry differences between these two cases are presented
graphically in Figure 1112, Figure 1113, and Figure 1114. The calibrated Predator is
on the left and the optimum Predator is on the right for these figures. Key geometry
differences are presented numerically in Figure 1115.
The Predator parametric fuselage has distinct differences from the calibration case. The
parametric Predator and parametric Global Hawk have the same radome sizing
relationship, which scales the radome dimensions linearly with the parabolic dish
diameter. The parametric Predator fuselage contains a wing fairing on the underside,
whereas the calibrated Predator does not.
The optimization process resulted in a significantly lower lengthtodiameter ratio and
shorter fuselage length for the optimized Predator than for the calibration case. Although
less visually pleasing than the calibration case, this is a valid solution.
The parametric Predator tail geometry is different for the two cases. Although still a V
tail configuration, the parametric Predator has a traditional Vtail, and the calibrated
Predator has an inverted Vtail. The traditional Vtail was selected to minimize
erroneous interference impacts in the Trefftz plane analysis that might be exploited by the
optimizer and artificially drive the optimized solutions. The inverted Vtail was found to
be free of interference problems on the baseline case, despite the Vtail plane intersecting
the wing plane. The dihedral angle is higher on the optimized case because the wing
aspect ratio is higher, which drives the vertical component of the tail. The tail
proportions of the optimized case appear exaggerated because the tail moment arm is
short.
The optimized solution has a slightly shorter span and nearly identical aspect ratio to the
calibration case.
186
Figure 1112 1994 Predator Geometry Comparison – Top View
Figure 1113 1994 Predator Geometry Comparison – Side View
187
Figure 1114 1994 Predator Geometry Comparison – Isometric View
Predator Geometry Comparison
0
10
20
30
40
50
Fuse Length, ft Wing Span, ft Aspect Ratio
Baseline
94 Opt
Figure 1115 Predator Geometry Comparison
A comparison of the calibrated and optimized 1994 Predator weight and mass fraction
summary is presented in Figure 1116 and Figure 1117, respectively. The optimized
Predator has a lower design gross weight than the calibration case. The primary
differences reside in structures and propulsion weights. The avionics and payload
weights are identical. The differences in the subsystems are driven primarily by reduced
flight control system and fuel system weights.
The uninstalled power to weight ratio for the optimized 1994 Predator is 15.5% less than
the calibrated case. This directly drives the propulsion weight down, which has
significant synergistic effects on the design. This appears to be a driver for the weight
difference between the two cases. If a takeoff analysis were performed, then a higher
power to weight ratio might result.
188
In addition to the reduced propulsion weight influence, the structures weight reduction
also appears to be a function of reduced wing, fuselage, and tail size.
Predator Weight Summary Summary
1
10
100
1,000
10,000
A
v
io
n
ic
s
S
u
b
s
y
s
te
m
s
P
r
o
p
u
l
s
io
n
S
tr
u
c
t
u
r
e
s
P
a
y
lo
a
d
F
u
e
l
G
r
o
w
t
h
E
m
p
t
y
D
e
s
ig
n
G
r
o
s
s
W
e
i
g
h
t
,
l
b
s
Baseline
94 Opt
Figure 1116 Predator Weights Summary Comparison
Predator Mass Fraction Summary
0%
5%
10%
15%
20%
25%
30%
35%
40%
45%
50%
A
v
i
o
n
ic
s
S
u
b
s
y
s
te
m
s
P
r
o
p
u
ls
io
n
S
tr
u
c
t
u
r
e
s
P
a
y
lo
a
d
F
u
e
l
G
r
o
w
t
h
M
a
s
s
F
r
a
c
t
i
o
n
Baseline
94 Opt
Figure 1117 Predator Summary Mass Fraction Comparison
The internally optimized mission profiles for the two reference year Predator cases are
similar. The mission performance characteristics are compared in Figure 1118, Figure
1119, Figure 1120, Figure 1121, and Figure 1122. The optimized Predator has a wing
loading 10.9% higher than the baseline case, resulting in higher speeds in some portions
of the mission.
The difference in specific fuel consumption appears to be driven by the propeller, which
is scaled as a function of engine power. The propeller efficiency is a function of the
power required, the altitude, and airspeed.
189
Predator Altitude Profile Comparison
0
5,000
10,000
15,000
20,000
25,000
30,000
0 10 20 30 40
Time, hrs
A
l
t
i
t
u
d
e
,
f
t
Baseline
94 Opt
Figure 1118 Predator Altitude Profile Comparison
Predator Velocity Profile Comparison
0
20
40
60
80
100
120
0 10 20 30 40
Time, hrs
V
e
l
o
c
i
t
y
,
k
t
s
Baseline
94 Opt
Figure 1119 Predator Velocity Profile Comparison
190
Predator VelocityAltitude Profile Used
0
5,000
10,000
15,000
20,000
25,000
30,000
40 50 60 70 80 90 100 110
Velocity, kts
A
l
t
i
t
u
d
e
,
f
t
Baseline
94 Opt
Figure 1120 Predator VelocityAltitude Envelope Used
Predator Specific Fuel Consumption Comparison
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0 5 10 15 20 25 30 35 40
Time, hrs
S
F
C
,
l
b
m
/
l
b

h
r
Baseline
94 Opt
Figure 1121 Predator Specific Fuel Consumption Comparison
191
Predator Lift to Drag Ratio Comparison
0
5
10
15
20
25
0 5 10 15 20 25 30 35 40
Time, hrs
L
i
f
t
t
o
D
r
a
g
R
a
t
i
o
Baseline
94 Opt
Figure 1122 Predator Lift to Drag Comparison
192
11.2.3 Shadow 200 2000 Optimum Design
Table 112 shows the Shadow 200 active design variables and the associated limiting
bounds. These values were used for all optimized Shadow 200 cases.
Design Variable Lower Bound Upper Bound
MF
Fuel
0.08 0.25
b
W
3 12.75
AR
W
5 20
P/W 0.05 0.2
L
Fuse
3 12
t/c
W
0.09 0.17
l
W
0.2 1
Table 113 Shadow 200 Design Variables
The Shadow 200 reference year optimal solution varies significantly from the calibration
case. The geometry comparison of these cases is presented in Figure 1123, Figure
1124, and Figure 1125. The calibrated Shadow 200 is on the left and the optimum
Shadow 200 is on the right in these figures. The numerical geometry comparison is
shown in Figure 1125.
The fuselage geometry is noticeably different. The calibration Shadow 200 case fuselage
has a pod and twinboom configuration. This configuration is more difficult to integrate
parametrically than the parametric fuselage formulation used for the other cases.
Additionally, the parametric wing is attached to the lower fuselage with a wing fairing,
whereas the calibration case has a midwing attachment to the pod fuselage. Future
versions of the software could easily integrate a parametric twinboom configuration.
The tail geometry is also different between the two cases. The calibration case has an
inverted Vtail attached at the tips to the booms. The parametric tails are a traditional V
tail configuration attached to the fuselage.
The optimization process resulted in a higher aspect ratio wing than the calibration case.
A maximum wingspan constraint of 12.75 feet, the span for the calibration case, was
imposed to account for transportation considerations. The optimized case does not reach
the wingspan constraint. The Shadow 200 was constrained by AAI to maintain the chord
of the available tooling, which largely drove the wing design. The optimal case does not
have this restriction, so the aspect ratio and wingspan are free to vary independently. The
resulting aspect ratio was much higher than the calibration case. The optimal Shadow
200 flies faster than the calibrated Shadow 200, due to an increased wing loading.
193
Figure 1123 2000 Shadow 200 Geometry Comparison – Top View
Figure 1124 2000 Shadow 200 Geometry Comparison – Side View
Figure 1125 2000 Shadow 200 Geometry Comparison – Isometric View
194
Shadow 200 Geometry Comparison
0
2
4
6
8
10
12
14
Fuse Length, ft Wing Span, ft Aspect Ratio
Baseline
00 Opt
Figure 1126 Shadow 200 Geometry Comparison
The Shadow 200 2000 weight and mass fraction summary comparisons are presented in
Figure 1127 and Figure 1128, respectively. The avionics and payload weights between
the two cases are identical. The differences in subsystems weights are largely due to
flight controls and fuel system weight variations. The payload mass fraction is higher for
the optimized case, showing an increase in relative utility. The higher payload mass
fraction is made possible by a reduction in structural mass fraction. The fuel mass
fractions are similar, which is a result of offsetting effects of the improved lift to drag
ratio and degraded specific fuel consumption of the optimized case. The degraded
specific fuel consumption is caused by propeller efficiency reduction at higher speeds.
Shadow 200 Weight Summary
1
10
100
1000
A
v
io
n
ic
s
S
u
b
s
y
s
t
e
m
s
P
r
o
p
u
l
s
i
o
n
S
t
r
u
c
t
u
r
e
s
P
a
y
lo
a
d
F
u
e
l
G
r
o
w
t
h
E
m
p
t
y
D
e
s
ig
n
G
r
o
s
s
W
e
i
g
h
t
,
l
b
s
Baseline
00 Opt
Figure 1127 Shadow 200 Summary Weights Comparison
195
Shadow 200 Mass Fraction Summary
0%
10%
20%
30%
40%
50%
60%
70%
A
v
i
o
n
ic
s
S
u
b
s
y
s
t
e
m
s
P
r
o
p
u
l
s
i
o
n
S
t
r
u
c
t
u
r
e
s
P
a
y
l
o
a
d
F
u
e
l
G
r
o
w
t
h
E
m
p
t
y
M
a
s
s
F
r
a
c
t
i
o
n
Baseline
00 Opt
Figure 1128 Shadow 200 Summary Mass Fraction Comparison
Comparison of the sameyear Shadow 200 flight performance characteristics are shown
in Figure 1129, Figure 1130, Figure 1131, Figure 1132, and Figure 1133.
Shadow 200 Altitude Profile Comparison
0
2,000
4,000
6,000
8,000
10,000
12,000
14,000
16,000
0 1 2 3 4 5 6 7
Time, hrs
A
l
t
i
t
u
d
e
,
f
t
Baseline
00 Opt
Figure 1129 Shadow 200 Altitude Profile Comparison
196
Shadow 200 Velocity Profile Comparison
0
20
40
60
80
100
120
0 1 2 3 4 5 6 7
Time, hrs
V
e
l
o
c
i
t
y
,
k
t
s
Baseline
00 Opt
Figure 1130 Shadow 200 Velocity Profile Comparison
Shadow 200 VelocityAltitude Envelope Used
0
2,000
4,000
6,000
8,000
10,000
12,000
14,000
16,000
40 50 60 70 80 90 100 110 120
Velocity, kts
A
l
t
i
t
u
d
e
,
f
t
Baseline
00 Opt
Figure 1131 Shadow 200 VelocityAltitude Envelope Used
197
Shadow 20 Specific Fuel Consumption Comparison
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
0.40
0.45
0.50
0 1 2 3 4 5 6 7
Time, hrs
S
F
C
,
l
b
m
/
l
b

h
r
Baseline
00 Opt
Figure 1132 Shadow 200 Specific Fuel Consumption Comparison
Shadow 200 Lift to Drag Ratio Comparison
0
2
4
6
8
10
12
14
0 1 2 3 4 5 6 7
Time, hrs
L
i
f
t
t
o
D
r
a
g
R
a
t
i
o
Baseline
00 Opt
Figure 1133 Shadow 200 Lift to Drag Ratio Comparison
198
Chapter 12 Technology Case
Optimization Applications
12.1 Overview
The technology impacts investigation evaluated the design synthesis of air vehicles with
constant requirements and advancing technologies. Optimization aided the design
synthesis substantially. The Global Hawk, Predator, and Shadow 200 technology
solutions are presented in this chapter.
12.2 Global Hawk Technology Impacts
12.2.1 Sizing Overview
The Global Hawk optimized cases experienced steady weight reduction through the
advancing years. The relative improvement per period steadily diminished through 2010.
The improvement degradation began to stabilize near 2015 through 2020. The 2025
optimum design was an unusually good design, taking a significantly different approach
than the two preceding cases. The initial advances were mostly driven by the SAR
payload weight reduction and propulsion improvements, followed by avionics/
subsystems, structures and aerodynamics. Beyond 2010, the relatively unvarying EO/IR
payload weight limited the sizing improvement potential, despite advances in other
technologies.
12.2.2 Geometry
The graphical summary of the Global Hawk technology cases is presented in Figure 121,
Figure 122, and Figure 123. The earliesttolatest date progresses from left to right in
these figures. Wing geometry trends are shown in Figure 124, Figure 125, Figure 126,
Figure 127, and Figure 128. The fuselage length trend is shown in Figure 129.
Global Hawk is the only design class that contends with compressibility effects. The
Global Hawk cases can ingress and egress at a low transonic speed, so wing sweep,
reduced wing thickness to chord ratio, or both were typically selected by the optimizer to
delay the onset of compressibility effects. Two design families, or niches, emerged from
the optimization, as shown in Figure 128. The 1994, 2010, and 2015 cases took the
approach using low aspect ratio wings with little sweep and reduced thickness to chord
ratios. The 2000, 2005, and 2025 cases had high aspect ratio wings with some sweep and
high thickness to chord ratio wings. For a fixed technology year, the unswept, low aspect
199
ratio niche has a relatively low structural mass fraction due to the smaller wing size and
more efficient structure, but a higher fuel fraction due to the relatively poor aerodynamic
performance. For a fixed technology year, the swept, highaspect ratio niche has a higher
aerodynamic efficiency and a lower fuel mass fraction that the competing niche, but this
comes at the expense of structural weight.
The wing span and aspect ratio tended to be more significant drivers than taper ratio and
wing sweep. The optimizer tended to select wing thickness to chord ratios of 15.717%,
which was well above the minimum value of 9%. The wing span for sameniche
vehicles consistently decreased over time.
The fuselage length decreased from 1994 through 2010, but then remained relatively
constant from 2015 through 2020. The radome geometry influence on the fuselage outer
loft became insignificant by 2020.
Figure 121 Global Hawk Geometry Trends – Top View, Years (Left to Right):
1994, 2000, 2005, 2010, 2015, 2020, 2025
Figure 122 Global Hawk Geometry Trends – Side View, Years (Left to Right):
1994, 2000, 2005, 2010, 2015, 2020, 2025
200
Figure 123 Global Hawk Geometry Trends – Isometric View, Years (Left to
Right): 1994, 2000, 2005, 2010, 2015, 2020, 2025
Global Hawk Wingspan
0
20
40
60
80
100
120
94 Opt 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt
W
i
n
g
s
p
a
n
,
f
t
Figure 124 Global Hawk Wing Span Trends
201
Global Hawk Aspect Ratio
0
5
10
15
20
25
30
94 Opt 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt
A
s
p
e
c
t
R
a
t
i
o
Figure 125 Global Hawk Aspect Ratio Trends
Global Hawk Wing Sweep
0
2
4
6
8
10
94 Opt 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt
W
i
n
g
S
w
e
e
p
,
D
e
g
r
e
e
s
Figure 126 Global Hawk Wing Sweep Trends
Global Hawk Thickness to Chord Ratio
15.0%
15.5%
16.0%
16.5%
17.0%
17.5%
94 Opt 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt
T
h
i
c
k
n
e
s
s
t
o
C
h
o
r
d
R
a
t
i
o
Figure 127 Global Hawk Thickness to Chord Ratio Trends
202
Global Hawk
Design Variable Values Relative to Limits
0%
20%
40%
60%
80%
100%
1990 1995 2000 2005 2010 2015 2020 2025
Design Year
R
e
l
a
t
i
v
e
V
a
l
u
e
ARw
t/c
Sweep
Figure 128 Global Hawk Wing Parameter Comparisons
Global Hawk Fuselage Length
0
10
20
30
40
50
94 Opt 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt
F
u
s
e
l
a
g
e
L
e
n
g
t
h
,
f
t
Figure 129 Global Hawk Fuselage Length Trends
12.2.3 Propulsion
The global hawk uninstalled thrust and uninstalled thrusttoweight trends are shown in
Figure 1210 and Figure 1211, respectively. The Global Hawk uninstalled thrust level
rapidly shrank from 1994 to 2000, but then was reduced more gradually and
inconsistently from 2000 to 2025. There was no consistent trend for uninstalled thrust to
weight ratio referenced to the design gross weight as a function of design year. The
vehicle thrust to weight ratio became less costly to the design as time progressed due to
engine thrust to weight improvements, so the thrust to weight ratio increased from 2010
to 2020. The decrease in thrust to weight ratio of the 2025 design relative to its
predecessor was a result of more aggressive aerodynamic design, such as higher aspect
ratio and increased sweep.
203
Global Hawk Uninstalled Thrust
0
1,000
2,000
3,000
4,000
5,000
6,000
94 Opt 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt
U
n
i
n
s
t
a
l
l
e
d
T
h
r
u
s
t
,
l
b
s
Figure 1210 Global Hawk Uninstalled Thrust Trends
Global Hawk Thrust to Weight Ratio
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
94 Opt 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt
T
h
r
u
s
t
t
o
W
e
i
g
h
t
R
a
t
i
o
Figure 1211 Global Hawk Thrust to Weight Ratio Trends
12.2.4 Weights
Weight Summary
The Global Hawk weight and mass fraction trends are shown in Figure 1212 and Figure
1213, respectively. The weight improvement trends are shown in Figure 1214. As
expected, the synthesized vehicle weight improved as technology improves. The mass
fraction of the payload increased as the mass fractions consumed by fuel, propulsion, and
avionics were reduced. Note that the weights in Figure 1212 are on a log scale, so
weight categories such as the growth weight appear exaggerated despite having a
relatively constant mass fraction.
204
Global Hawk Weight Summary
1
10
100
1000
10000
100000
A
v
i
o
n
ic
s
S
u
b
s
y
s
t
e
m
s
P
r
o
p
u
l
s
i
o
n
S
t
r
u
c
t
u
r
e
s
P
a
y
l
o
a
d
F
u
e
l
G
r
o
w
t
h
E
m
p
t
y
W
e
i
g
h
t
,
l
b
s
94 Opt
00 Opt
05 Opt
10 Opt
15 Opt
20 Opt
25 Opt
Figure 1212 Global Hawk Weight Trends
Global Hawk Mass Fraction Summary
0%
10%
20%
30%
40%
50%
60%
70%
A
v
i
o
n
ic
s
S
u
b
s
y
s
t
e
m
s
P
r
o
p
u
l
s
i
o
n
S
t
r
u
c
t
u
r
e
s
P
a
y
l
o
a
d
F
u
e
l
G
r
o
w
t
h
E
m
p
t
y
M
a
s
s
F
r
a
c
t
i
o
n
94 Opt
00 Opt
05 Opt
10 Opt
15 Opt
20 Opt
25 Opt
Figure 1213 Global Hawk Mass Fraction Trends
Global Hawk
Improvement Over Previous Period
0%
10%
20%
30%
40%
50%
60%
70%
2000 2005 2010 2015 2020 2025
Design Year
I
m
p
r
o
v
e
m
e
n
t
,
%
W_DG
W_Obligated
Figure 1214 Design Gross  Obligated Weights Trends
205
Avionics Weights
The avionics weights improved dramatically through future years. The results can be
seen in Figure 1215. The relative influence of the overall avionics weight contribution
to vehicle sizing compared to the payload weight became less significant as time
advanced. The avionics weight reductions follow the technology trends provided in
Chapter 4.
Global Hawk Avionics Weight
0.00001
0.0001
0.001
0.01
0.1
1
10
100
1000
94 Opt 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt
W
e
i
g
h
t
,
l
b
s
General Avionics
Autopilot
INS
GPS
ADS
ATC
Recorder
Figure 1215 Global Hawk Avionics Weight Trends
Subsystems Weights
The subsystems weight trends are shown in Figure 1216. The relative contribution of
the communications system was greater in the future cases when using the assumed
technology curves. This was caused by the parabolic dish weight that was a function of
dish diameter. The dish diameter and associated dish weight followed a technology trend
separate from the subsystems curve.
206
Global Hawk Subsystems Weight
1
10
100
1000
94 Opt 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt
W
e
i
g
h
t
,
l
b
s
FCS
ECS
Electrical System
Fuel System
Communications
Figure 1216 Global Hawk Subsystems Weight Trends
Payload Weights
The Global Hawk payload weight trends are shown in Figure 1217. The SAR payload
weight saw aggressive improvement in weight relative to the EO/IR payload. Relative to
the 1994 SAR value, the SAR payload weight was reduced by 87.2 % in 2005 and 98.9%
by 2025. Relative to the 1994 EO/IR value, the EO/IR payload weight was reduced by
only 7.4% in 2005 and 14.6% in 2025. The largeEO/IR payload technology prediction
was applied here. The EO/IR payload became the primary vehicle sizing driver in 2025.
Global Hawk Payload Weight
1
10
100
1000
94 Opt 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt
W
e
i
g
h
t
,
l
b
s
EO/IR
SAR
Figure 1217 Global Hawk Payload Weight Trends
Structures Weights
The Global Hawk structural weight trends are shown in Figure 1218. The installation
weight became the most dominant structural weight contributor in 2005. The physical
207
size of the structure decreased due to volume reductions of many components, fuel
reduction, and general sizing phenomena. The combined relative weight of the obligated
weights did not change as rapidly as the major structural components. The installation
weight associated with these items was proportional to the component weights, and
therefore the installation weight did not decrease rapidly. The wing structural weight
trends would likely vary if an aerolastic and stiffness analyses were performed. Future
research should fully account for these relevant phenomena.
Global Hawk Structures Weight
1
10
100
1000
10000
94 Opt 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt
W
e
i
g
h
t
,
l
b
s
Wing
Tail
Fuselage
Nacelle
Gear
Installation
Figure 1218 Global Hawk Structures Weight Trends
12.2.5 Performance
The optimized Global Hawk mission profiles and related performance data are shown in
Figure 1219, Figure 1220, Figure 1221, and Figure 1222. The internal mission
optimization routine selected slightly different flight profiles for the various solutions.
The 1994 optimum had the highest average Mach number and reached the maximum
altitude quickest.
208
Global Hawk Altitude Profile
0
10,000
20,000
30,000
40,000
50,000
60,000
70,000
0 5 10 15 20 25 30 35
Time, hrs
A
l
t
i
t
u
d
e
,
f
t
G 94
G 00
G 05
G 10
G 15
G 20
G 25
Figure 1219 Global Hawk Altitude Profile Trends
Global Hawk Mach Profile
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0 5 10 15 20 25 30 35
Time, hrs
M
a
c
h
G 94
G 00
G 05
G 10
G 15
G 20
G 25
Figure 1220 Global Hawk Mach Profile Trends
Global Hawk Lift to Drag Ratio
0
5
10
15
20
25
30
0 5 10 15 20 25 30 35
Time, hrs
L
i
f
t
t
o
D
r
a
g
R
a
t
i
o
G 94
G 00
G 05
G 10
G 15
G 20
G 25
Figure 1221 Global Hawk Mission Lift to Drag Ratio Trends
209
Global Hawk Specific Fuel Consumption
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0 5 10 15 20 25 30 35
Time, hrs
S
F
C
l
b
m
/
l
b

h
r
G 94
G 00
G 05
G 10
G 15
G 20
G 25
Figure 1222 Global Hawk Mission Fuel Consumption Trends
12.2.6 Technology Sensitivities
The Global Hawk technology sensitivities are shown in Figure 1223. The three bars
represent different ways of viewing the technology impacts. The total bar is the design
gross weight difference between the 1994 optimized Global Hawk and the 2005
optimized Global Hawk with all technologies applied simultaneously in the design. The
components bar shows the weight improvement for each individual technology group
when applied individually. Finally, the combined bar is the multiplication of all of the
improvements from each of the technology groups. The combined bar is found from:
( )
∏
− − =
Components
Component
t Improvemen t Improvemen 1 1
Global Hawk was most sensitive to payload and then propulsion technology impacts
between 1994 and 2005. The avionics/subsystems technologies also had significant
influence. There was overall synergy among the combined technologies. The combined
improvement of all individual technology groups is 67.9%, and the total difference in
design gross weight from 1994 to 2005 is 74.5%. The overall synergy, defined here as
the difference of the total and combined technologies, is 6.68%.
210
Global Hawk Technology Sensitivities
19942005
0%
20%
40%
60%
80%
100%
Total Components Combined
W
e
i
g
h
t
R
e
d
u
c
t
i
o
n
,
%
Combined
Total
Structures
Propulsion
Payload
Design
Avion/Subs
Aerodynamics
Figure 1223 Global Hawk Design Gross Weight Sensitivity to Technology
211
12.3 Predator Technology Impacts
12.3.1 Sizing Overview
Like Global Hawk, Predator sizing was strongly influenced by the payloads and other
obligated weights. Predator used the small EO/IR technology trend, which was far more
aggressive than the Global Hawk large EO/IR trend. The notable reduction in EO/IR
weight throughout the dates considered permitted continuing weight reduction beyond
2005.
12.3.2 Geometry
Graphical representation of the Predator technology case geometry behavior is shown in
Figure 1224, Figure 1225, and Figure 1226. Numerical wing geometry trends are
presented in Figure 1227, Figure 1228, Figure 1229, and Figure 1230. The fuselage
length trend is provided in Figure 1231.
The most dominant visual feature of the optimized predator solutions is the radome. The
Predator parametric radome strongly affected the aerodynamic performance of the
vehicles under consideration. As the assumed satellite communications technology
improved, the dish diameter and radome size decreased, permitting the fuselage to
assume more visually pleasing and aerodynamically efficient contours. The fuselage
length generally decreased with time due to decreasing avionics, subsystems, propulsion,
and fuel volume.
The strongest wing geometry contributors to design gross weight were the wing span and
aspect ratio, which define the wing area. The wing area tended to decrease with
decreasing vehicle weight. The taper and thickness to chord ratios had no correlation
with design year, and were not major design drivers.
The optimization process resulted in a relatively low wing taper ratio of 0.81 for the 2015
optima. One possible explanation for this result is the impact of the wing on the fuselage
geometry. The parametric wing fairing length is proportional to the wing root chord.
The wing root chord may have been reduced to manage the fuselage length to diameter
ratio for fuselage drag reduction, given the dominance of the radome. Wing taper
provides wing structural weight and induced drag reductions. Apparently, the negative
impacts of low taper ratios on the fuselage drag overcame the wing structural and
aerodynamic benefits for the 2015 case. The optimizer generally selected more
aggressive wing taper for this family of UAV.
212
Figure 1224 Predator Geometry Trends – Top View, Years (Left to Right): 1994,
2000, 2005, 2010, 2015, 2020, 2025
Figure 1225 Predator Geometry Trends – Side View, Years (Left to Right): 1994,
2000, 2005, 2010, 2015, 2020, 2025
213
Figure 1226 Predator Geometry Trends – Isometric View, Years (Left to
Right): 1994, 2000, 2005, 2010, 2015, 2020, 2025
Predator Wing Span
0
10
20
30
40
50
94 Opt 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt
W
i
n
g
S
p
a
n
,
f
t
Figure 1227 Predator Wing Span Trends
214
Predator Aspect Ratio
0
5
10
15
20
25
30
94 Opt 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt
A
s
p
e
c
t
R
a
t
i
o
Figure 1228 Predator Aspect Ratio Trends
Predator Thickness to Chord Ratio
0
0.05
0.1
0.15
0.2
94 Opt 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt
T
h
i
c
k
n
e
s
s
t
o
C
h
o
r
d
R
a
t
i
o
Figure 1229 Predator Thickness to Chord Ratio
Predator Wing Area
0
20
40
60
80
100
120
94 Opt 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt
W
i
n
g
A
r
e
a
,
s
q
.
f
t
.
Figure 1230 Predator Wing Area
215
Predator Fuselage Length
0
5
10
15
20
94 Opt 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt
F
u
s
e
l
a
g
e
L
e
n
g
t
h
,
f
t
Figure 1231 Predator Fuselage Length Trends
12.3.3 Propulsion
The Predator uninstalled power and uninstalled powertoweight trends are shown in
Figure 1232 and Figure 1233, respectively. As expected, the uninstalled power
decreased with vehicle weight as time progresses. The powertoweight ratio tended to
increase as time progressed, since the weight impacts attributable to the propulsion
system were less stressing on future designs.
Predator Uninstalled Power
0
20
40
60
80
100
94 Opt 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt
U
n
i
n
s
t
a
l
l
e
d
P
o
w
e
r
,
H
P
Figure 1232 Predator Uninstalled Power Trends
216
Predator Power to Weight Ratio
0.00
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
94 Opt 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt P
o
w
e
r
t
o
W
e
i
g
h
t
R
a
t
i
o
,
H
P
/
l
b
Figure 1233 Predator Power to Weight Ratio
12.3.4 Weights
Weight Summary
The Predator technology case weight and mass fraction trends are shown in Figure 1234
and Figure 1235, respectively. The Predator overall weight improvement trends are
shown in Figure 1236. The Predator weights uniformly decreased with improving
technology. The behavior of the mass fractions had variation, but the general trend was
for the fuel mass fraction to decrease while the payload mass fraction increased. The
overall design gross weight improved faster than the reduction in obligated weights,
which revealed that the other technology improvements were providing benefit.
Predator Weight Summary
0.1
1
10
100
1000
10000
A
v
i
o
n
ic
s
S
u
b
s
y
s
t
e
m
s
P
r
o
p
u
l
s
i
o
n
S
t
r
u
c
t
u
r
e
s
P
a
y
l
o
a
d
F
u
e
l
G
r
o
w
t
h
E
m
p
t
y
D
e
s
i
g
n
G
r
o
s
s
W
e
i
g
h
t
,
l
b
s
94 Opt
00 Opt
05 Opt
10 Opt
15 Opt
20 Opt
25 Opt
Figure 1234 Predator Weight Summary Trends
217
Predator Mass Fraction Summary
0%
10%
20%
30%
40%
50%
60%
A
v
i
o
n
ic
s
S
u
b
s
y
s
t
e
m
s
P
r
o
p
u
l
s
i
o
n
S
t
r
u
c
t
u
r
e
s
P
a
y
l
o
a
d
F
u
e
l
G
r
o
w
t
h
M
a
s
s
F
r
a
c
t
i
o
n
s
94 Opt
00 Opt
05 Opt
10 Opt
15 Opt
20 Opt
25 Opt
Figure 1235 Predator Summary Mass Fraction Trends
Predator
Improvement Over Previous Period
0%
10%
20%
30%
40%
50%
60%
70%
2000 2005 2010 2015 2020 2025
Design Year
I
m
p
r
o
v
e
m
e
n
t
,
%
W_DG
W_Obligated
Figure 1236 Predator Design Gross – Obligated Weight Trends
Avionics Weights
The Predator technology case avionics weights trends are presented in Figure 1237. As
seen on Global Hawk, the avionics weights became negligible as technology advanced.
218
Predator Avionics Weights
0.001
0.01
0.1
1
10
100
94 Opt 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt
W
e
i
g
h
t
,
l
b
s
General
Autopilot
INS
GPS
ADS
ATC
Recorder
Figure 1237 Predator Avionics Weight Trends
Subsystems Weights
The Predator technology case subsystems weight trends are shown in Figure 1238. The
subsystems weights all decreased with time. The slowest to decrease was the
communications system weight, because this weight was a function of the satellite
communications dish diameter that followed a technology trend distinct from the other
subsystems.
Predator Subsystems Weight
0.01
0.1
1
10
100
94 Opt 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt
W
e
i
g
h
t
,
l
b
s
FCS
ECS
Electrical
Fuel Sys
Comms
Figure 1238 Predator Subsystems Weight Trends
Payload Weights
The Predator payload weight trends are presented in Figure 1239. The payload weights
exhibited similar behavior to those seen on Global Hawk. The relative weight reduction
of the SAR payload from 1994 was identical to that on Global Hawk. The EO/IR saw
219
more weights improvement than on Global Hawk, because the more aggressive small
EO/IR technology trend was applied.
Predator Payload Weight
1
10
100
1000
94 Opt 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt
W
e
i
g
h
t
,
l
b
s
EO/IR
SAR
Figure 1239 Predator Payload Weight Trends
Structural Weights
The Predator technology case structural weight trends are shown in Figure 1240. The
structural weights all decreased with time. The installation weight became increasingly
dominant. This weight accounts for the structural mounting of avionics, subsystems,
payload, and propulsion. The slowly decreasing payload weight was the primary
contributor to the relatively intractable installation weight.
Predator Structural Weight
1
10
100
1000
94 Opt 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt
W
e
i
g
h
t
,
l
b
s
Wing
Tail
Fuselage
Gear
Installation
Figure 1240 Predator Structures Weight Trends
220
12.3.5 Performance
The Predator technology case mission profiles and associated performance information
are presented in Figure 1241, Figure 1242, Figure 1243, and Figure 1244. The
Predator flight performance showed some variation.
The 1994 optimum reached the maximum altitude during loiter much more rapidly than
later aircraft in the family. The higher flight velocity brought the 1994 Predator design to
the loiter point rapidly. During loiter, the high lift to drag ratio permitted this aircraft to
climb faster than its less aerodynamically efficient descendents. The 1994 Global Hawk
also had a faster ingress than its descendants.
In general, the lift to drag ratio degraded with time. The culprit was that the size of the
EO/IR ball relative to the wing made this very high drag sphere a larger contributor in the
future cases. Improved specific fuel consumption in later years made aerodynamic
performance less important to achieve the same performance results. Aerodynamic
performance improvements can be attained by increasing the aspect ratio and wing span,
but both of these options add structural weight that offsets the benefits of reduced fuel
consumption.
Predator Altitude Profile
0
5,000
10,000
15,000
20,000
25,000
30,000
0 10 20 30 40
Time, hrs
A
l
t
i
t
u
d
e
,
f
t
94 Opt
00 Opt
05 Opt
10 Opt
15 Opt
20 Opt
25 Opt
Figure 1241 Predator Altitude Profile Trends
221
Predator Velocity Profile
0
20
40
60
80
100
120
0 10 20 30 40
Time, hrs
V
e
l
o
c
i
t
y
,
k
t
s
94 Opt
00 Opt
05 Opt
10 Opt
15 Opt
20 Opt
25 Opt
Figure 1242 Predator Velocity Profile Trends
Predator Lift to Drag Ratio
0
2
4
6
8
10
12
14
16
18
20
0 10 20 30 40
Time, hrs
L
i
f
t
t
o
D
r
a
g
R
a
t
i
o
94 Opt
00 Opt
05 Opt
10 Opt
15 Opt
20 Opt
25 Opt
Figure 1243 Predator Mission Lift to Drag Ratio Trends
222
Predator Specific Fuel Consumption
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0 10 20 30 40
Time, hrs
S
F
C
,
l
b
m
/
l
b

h
r
94 Opt
00 Opt
05 Opt
10 Opt
15 Opt
20 Opt
25 Opt
Figure 1244 Predator Mission Fuel Consumption Trends
12.3.6 Technology Sensitivities
The Predator design gross weight sensitivities to various technologies are shown in
Figure 1245. The Predator showed 7.1% overall synergy, which is the highest of the
three vehicle classes evaluated here. Predator is most sensitive to the payload technology
category, followed by avionics/subsystems and propulsion technology categories.
Predator Technology Sensitivities
19942005
0%
20%
40%
60%
80%
100%
120%
Total Components Combined
W
e
i
g
h
t
R
e
d
u
c
t
i
o
n
,
%
Combined
Total
Structures
Propulsion
Payload
Design
Avion/Subs
Aerodynamics
Figure 1245 Predator Design Gross Weight Sensitivity to Technology
223
12.4 Shadow 200 Technology Impacts
12.4.1 Sizing Overview
The Shadow 200 family has shown the least improvement over time. This was caused by
the relatively low fuel mass fraction lowering the weight escalation factor.
12.4.2 Geometry
The Shadow 200 technology case trends are shown graphically in Figure 1246, Figure
1247, and Figure 1248. The wing geometry behavior is provided in Figure 1249 and
Figure 1250. The fuselage length is shown presented in Figure 1251. Two families of
solutions, or niches, emerged through optimization. The first family sought to minimize
physical size to reduce structural weight. The second family had a combination of high
aspect ratio and low wing loading to improve aerodynamic performance and propulsion
efficiency, which improved the fuel fraction. Figure 1252 shows the wing loading and
aspect ratio trends for all technology cases. The 2015 and 2025 cases fall into the latter
family. No clear aspect ratio and taper ratio relationships are evident.
Figure 1246 Shadow 200 Geometry Trends – Top View, Years (Left to Right):
2000, 2005, 2010, 2015, 2020, 2025
Figure 1247 Shadow 200 Geometry Trends – Side View, Years (Left to Right):
2000, 2005, 2010, 2015, 2020, 2025
224
Figure 1248 Shadow 200 Geometry Trends – Isometric View, Years (Left to
Right): 2000, 2005, 2010, 2015, 2020, 2025
Shadow 200 Wingspan
0
2
4
6
8
10
12
14
00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt
W
i
n
g
s
p
a
n
,
f
t
Figure 1249 Shadow 200 Wing Span Trends
225
Shadow 200 Aspect Ratio
0
5
10
15
20
00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt
A
s
p
e
c
t
R
a
t
i
o
Figure 1250 Shadow 200 Aspect Ratio Trends
Shadow 200 Fuselage Length
0
1
2
3
4
5
6
7
8
00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt
F
u
s
e
l
a
g
e
L
e
n
g
t
h
,
f
t
Figure 1251 Shadow 200 Fuselage Length Trends
226
Shadow 200 Wing Loading and Aspect Ratio
0
5
10
15
20
25
30
2000 2005 2010 2015 2020 2025
Design Year
W/S, lb/sq.f t.
AR
Figure 1252 Shadow 200 Wing Loading – Aspect Ratio Relationships
12.4.3 Propulsion
The Shadow 200 technology case uninstalled power and uninstalled power to weight
tends are shown in Figure 1253 and Figure 1254, respectively. The maximum power to
weight difference among all optimized cases was 17%. This variation was more a
function of the solution niche than the design year.
Shadow 200 Uninstalled Power
0
5
10
15
20
25
30
35
00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt
U
n
i
n
s
t
a
l
l
e
d
P
o
w
e
r
,
H
P
Figure 1253 Shadow 200 Uninstalled Power Trends
227
Shadow 200 Power to Weight Ratio
0
0.02
0.04
0.06
0.08
0.1
0.12
00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt
P
o
w
e
r
t
o
W
e
i
g
h
t
R
a
t
i
o
,
H
P
/
l
b
Figure 1254 Shadow 200 Power to Weight Ratio Trends
12.4.4 Weights
Weight Summary
Figure 1255 and Figure 1256 show the Shadow 200 technology case weight and mass
fraction trends, respectively. The Shadow 200 overall weights decreased with improving
technology, as expected.
Figure 1257 shows the weight improvement trends. The overall design gross weight
improvement tracked the obligated weight improvement with an offset. The offset
represented improvement in structures technology and technologies that affect the fuel
fraction, including the aerodynamics and propulsion.
The decrease in the avionics and fuel mass fractions allowed an overall increase in
payload mass fraction. The higher payload mass fraction translated to a higher utility.
Shadow 200 Weight Summary
0.01
0.1
1
10
100
1000
A
v
i
o
n
ic
s
S
u
b
s
y
s
te
m
s
P
r
o
p
u
ls
io
n
S
tr
u
c
t
u
r
e
s
P
a
y
lo
a
d
F
u
e
l
G
r
o
w
t
h
E
m
p
t
y
D
e
s
ig
n
G
r
o
s
s
W
e
i
g
h
t
,
l
b
s
00 Opt
05 Opt
10 Opt
15 Opt
20 Opt
25 Opt
Figure 1255 Shadow 200 Weight Trends
228
Shadow 200 Mass Fraction Summary
0%
10%
20%
30%
40%
50%
60%
70%
A
v
io
n
ic
s
S
u
b
s
y
s
te
m
s
P
r
o
p
u
ls
io
n
S
t
r
u
c
t
u
r
e
s
P
a
y
lo
a
d
F
u
e
l
G
r
o
w
t
h
E
m
p
ty
M
a
s
s
F
r
a
c
t
i
o
n
00 Opt
05 Opt
10 Opt
15 Opt
20 Opt
25 Opt
Figure 1256 Shadow 200 Summary Mass Fraction Trends
Shadow 200
Improvement Over Previous Period
0%
10%
20%
30%
40%
50%
60%
2005 2010 2015 2020 2025
Design Year
I
m
p
r
o
v
e
m
e
n
t
,
%
W_DG
W_Obligated
Figure 1257 Shadow 200 Design Gross – Obligated Weights Trend
Avionics Weights
The Shadow 200 technology case avionics weights trends are shown in Figure 1258.
The avionics weights rapidly became negligible contributors by 2010 using the assumed
avionics technology curves.
229
Shadow 200 Avionics Weight
0.001
0.01
0.1
1
10
100
00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt
W
e
i
g
h
t
,
l
b
s
General Avionics
Figure 1258 Shadow 200 Avionics Weight Trends
Subsystems Weights
Figure 1259 shows the Shadow 200 subsystems weight trends. The subsystems weights
all decreased with time. All subsystems except for the communications system were
affected by vehicle sizing, so the communications subsystem weight did not decrease
rapidly. The flight control system was a function of control surface size, so the large area
of the 2015 and 2025 solutions resulted in a higher flight control weight than the adjacent
cases.
Shadow 200 Subsystems Weight
0.1
1
10
100
Baseline 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt
W
e
i
g
h
t
,
l
b
s
FCS
ECS
Fuel
Communications
Figure 1259 Shadow 200 Subsystems Weight Trends
Payload Weights
The Shadow 200 technology case payload weight trends are shown in Figure 1260. Like
the previous two cases, the SAR payload weight decreased much more rapidly than the
EO/IR. Despite applying the small EO/IR technology curve, the EO/IR weight remained
high relative to the SAR. The EO/IR became the primary air vehicle sizing driver in later
years.
230
Shadow 200 Payload Weight
0.1
1
10
100
00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt
W
e
i
g
h
t
,
l
b
s
EO/IR
SAR
Figure 1260 Shadow 200 Payload Weight Trends.
Structures Weights
The Shadow 200 structures weights technology trends are presented in Figure 1261. The
structures weights all decreased with time. The installation weight is the dominant
contributor in all years because of the small vehicle size and high obligated weight
fraction.
Shadow 200 Structures Weight
0.1
1
10
100
00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt
W
e
i
g
h
t
,
l
b
s
Wing
Tails
Fuselage
Gear
Installation
Figure 1261 Shadow 200 Structures Weight Trends
12.4.5 Performance
Figure 1262, Figure 1263, Figure 1264, and Figure 1265 show the Shadow 200 flight
profile and related performance trends. There were some variations in the optimized
flight profiles.
As seen on Predator, the low technology cases tended to have a higher lift to drag ratio
than the later cases. The relative contribution of the EO/IR ball drag was less on the
231
earlier optimal Shadow 200 cases because the ratio of ball area to wing area was smaller.
The high lift to drag ratio of the early cases facilitated a more rapid climb.
Shadow 200 Alitude Profile
0
2,000
4,000
6,000
8,000
10,000
12,000
14,000
16,000
0 1 2 3 4 5 6 7
Time, hrs
A
l
t
i
t
u
d
e
,
f
t
S 00
S 05
S 10
S 15
S 20
S 25
Figure 1262 Shadow 200 Altitude Profile Trends
Shadow 200 Velocity Profile
0
20
40
60
80
100
120
0 1 2 3 4 5 6 7
Time, hrs
V
e
l
o
c
i
t
y
,
k
t
s
S 00
S 05
S 10
S 15
S 20
S 25
Figure 1263 Shadow 200 Velocity Profile Trends
232
Shadow 200 Lift to Drag Ratio
0
2
4
6
8
10
12
14
0 1 2 3 4 5 6 7
Time, hrs
L
i
f
t
t
o
D
r
a
g
R
a
t
i
o
S 00
S 05
S 10
S 15
S 20
S 25
Figure 1264 Shadow 200 Mission Lift to Drag Ratio Trends
Shadow 200 Specific Fuel Consumption
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
0.40
0.45
0.50
0 1 2 3 4 5 6 7
Time, hrs
S
F
C
,
l
b
m
/
l
b

h
r
S 00
S 05
S 10
S 15
S 20
S 25
Figure 1265 Shadow 200 Mission Fuel Consumption Trends
12.4.6 Technology Sensitivities
The Shadow 200 design gross weight sensitivities to various technologies are shown in
Figure 1266. The Shadow 200 had a 4.22% overall synergy, which is the lowest of all
the cases evaluated here. Technology that affected the obligated weight directly had the
greatest impact, given the small fuel fraction and associated low weight escalation factor.
The strongest technology contributors were payloads and avionics/subsystems groups.
233
The higher fuel fraction of the previous cases created greater sensitivities to propulsion
technology. The more modest Shadow 200 total improvement was only slightly higher
than the combined payload and avionics/ subsystems groups.
Shadow 200 Technology Sensitivities
20002010
0%
20%
40%
60%
80%
100%
Total Components Combined
W
e
i
g
h
t
R
e
d
u
c
t
i
o
n
,
%
Combined
Total
Structures
Propulsion
Payload
Design
Avion/Subs
Aerodynamics
Figure 1266 Shadow 200 Design Gross Weight Sensitivity to Technology
234
Chapter 13 Conclusions
13.1 AV Design Methodology Conclusions
The goal of achieving an effective and robust aircraft design methodology for unmanned
aircraft has been realized. The methodology used here captures important phenomena
affecting UAV design for the AV classes considered. Several analysis methods are
available for many disciplines, and those ultimately selected for the final analysis
adequately blend effectiveness and computational speed. Many models, such as airfoil
profile performance, lift distribution prediction, propeller performance, and semianalytic
structures modeling provide high utility conceptual design capabilities beyond parametric
equations. Integration of this MDO code onto a personal computer offers great
convenience and severs reliance on more expensive computer resources. The
methodology has been applied to three different classes of UAVs.
The applied genetic algorithm is well suited to the unmanned aircraft MDO problem.
The noisy design space caused by the selfoptimizing mission profiles and tabular drag
polars necessitate a method as tolerant as a GA.
The aerodynamics methods are capable of modeling all cases considered. The parametric
airfoil formulation adequately captures driving LRN phenomena. Other aerodynamic
models employed proved effective for the problem set. The tabular drag polars are
convenient for mission analysis lookup requirements and for postdesign evaluations.
The propulsion models applied, while simple, adequately cover the propulsion
phenomena of interest. The propulsion analysis methods are wellsuited to the simple
propulsion technology trends used here. The unused complex propeller performance
model shows great promise despite the long relative computation time.
The AV size, weight and power model is flexible and covers the considerations of
interest. The specialized avionics methods were unused with the exception of the air data
system model, because every system has specific avionics suites. The subsystems models
rely largely on manned aircraft parametric equations that generally operate outside the
intended application. The piecewise linear beam model works well for all AV classes
covered, though the lack of aerolastic analysis and stiffnessbased design create
limitations. The empirical fuselage weights equations should be improved in future
research.
Integration of the selfoptimizing mission profiles enables additional freedom in AV
optimization. The alternative of simply specifying a fixed mission profile is more
constraining on the design problem.
235
The parametric geometry formulation used in the optimization cases is robust. The
vertical separation of the VTails and the wing ensure that no singularities occur in the
lift distribution methods. The minimal tail length parametric functions effectively
prevent the tail size to attain unrealistic proportions without adding additional constraints
for the optimizer.
13.2 Calibration Cases Conclusions
The overall calibration achieved for the three UAV classes was successful. However, the
calibration is based on often inconsistent and incomplete data. The calibration process
provides the user with many means of achieving the desired results, so the target results
are not difficult to obtain.
Unmanned aircraft, such as the Global Hawk and Shadow 200, experienced turbulent
design histories that are difficult to replicate. Requirements and design goals evolve over
time. The technologies change such that the reference technology year is difficult to
identify. This calibration process, or perhaps any other, could not replicate the design
history of the UAVs of interest.
Despite the inherent limitations of this process, all calibrations are within the bounds of
the formulation, which lends credibility to the AV analysis methodology. The mission
profiles, design gross weight, empty weight, and other known data was brought into close
agreement with the available information on the UAVs of interest.
13.3 Technology Cases Conclusions
The technology impact study demonstrated the utility of the AV design methodology.
The code optimized many AVs within a family over many technology years without a
need for tailoring the design variables, constraint weighting, or analysis methods.
This research identified the formats for technology trends information that can be applied
to AV trade studies. The process of tailoring existing technology trends to those applied
resulted in substantial format modification. Existing trends frequently did not suit the
needs of this research. Future AV technology modeling could benefit from following the
model data requirements developed here.
The design gross weight improves with advancing technology within each vehicle family.
The design gross weight is sensitive to weights that directly affect the obligated weights
for all cases. Additionally, the propulsion technology strongly affects the high
performance Global Hawk and Predator families, which have high fuel mass fractions
relative to the Shadow 200. All three families show overall technology synergies.
The relatively steady EO/IR sensor payload weight limits the achievable size reductions.
The EO/IR relative weight trends are assumed to match the EO/IR relative performance
236
trends here, which is probably conservative. The SAR performance per weight improved
approximately 12 orders of magnitude more than the SAR performance during the
timeframe under consideration. Unfortunately, there is no equivalent performance per
weight prediction available for the EO/IR payload.
Only a limited set of potential payloads are considered. Electronics Intelligence
(ELINT), Communications Intelligence (COMINT), Measurements and Signals
Intelligence (MASINT), and other relevant payloads are not covered. Any one of these
potential payloads could significantly impact the vehicle design.
Avionics technology impacts are significant for all families considered here. This is to be
expected, because the avionics category includes a significant portion of the obligated
weights. The avionics contribution to vehicle weight becomes negligible after
approximately 2010 with the technology modeling used.
Not all avionics are likely to see the full potential gains modeled in this research. For
example, the ATC equipment is largely driven by general aviation and commercial
aviation, and is very slow to adapt to change. The unique ATC equipment trades are not
considered in this research.
Miniaturization as a function of technology improvement has been practically
demonstrated through the DARPA Micro Aerial Vehicle (MAV) program. One of the
MAV program objectives is to demonstrate miniaturized technologies in vehicles with
wingspans less than 6 inches. Many MAVs with sizes either achieving or approaching
the goal were successfully developed. The MAV program achievements lend credibility
to the weight reduction predictions.
Flight performance and payload performance requirements were held constant for this
research. The resulting designs decreased in weight as technology advanced. This
behavior is well represented by historical trends. For example, the requirement for close
range battlefield surveillance has existed for decades. The vehicle sizes that support this
role have greatly decreased over time. The technically attainable UAV sizes for this role
have decreased from the 9foot wingspan Pointer in the 1980’s to the MAV class of
today. Technology advancement for fixed requirements promotes miniaturization.
Technologies enable UAV requirements to improve, which often results in larger vehicles
over time. Often, historical UAV designs within a manufacturer’s line tend to increase in
size as technology improves. This is due to new capabilities, previously not applicable to
a vehicle class, become available to potential UAV systems that are slightly larger than
the existing capability. Rather than wait for the technology to further improve, the UAV
manufacturer will develop a new system to accommodate the new mission capability.
The manufacturer, in this scenario, has an established capacity for UAV development and
gains approval to develop a larger version of the existing family. This scenario has been
demonstrated by Aerovironment with solar powered aircraft, General Atomics with the
GnatPredator family, and TeledyneRyan (now Northrop Grumman) with the long
lineage of reconnaissance UAVs that ultimately led to the Global Hawk. The General
237
Atomics GnatPredator family evolution is shown in Figure 1  note that time progresses
from right to left. The Teledyne Ryan UAV family evolution is shown in Figure 13 
again, time progresses from right to left. UAV design tools that capture technology
insertion can facilitate requirements evolution.
Figure 131 General Atomics GnatPredator family evolution (Left to Right:
Predator B, Predator A, Gnat)
13.4 Recommendations
Technology trends that are directly applicable to UAV design should be developed and
refined. As an example, the EO/IR payload weight technology trend became a limitation
to the ultimate size reductions of all cases considered here, despite its poor quality. The
greatest resources should be placed on technologies that have the greatest impact on AV
design, such as the EO/IR payload weight trend.
Detailed UAV subsystems definition and design methodology development should be
conducted. UAV subsystems often operate beyond typical limitations of manned aircraft.
Improved UAV subsystems weight methods are not possible without knowledge of the
applicable design considerations.
Performance requirements for future UAV systems that leverage technology
advancements should be further explored. These requirements could be applied to the
AV design methodology to assess the potential solution space more realistically.
Tomorrow’s UAV systems will probably not be designed against today’s requirements.
Alternatively, this AV design methodology could be used to help define future
requirements.
The performance models can be improved. The detailed propeller analysis method
should be evaluated for gains in computational efficiency. The methodology should
incorporate offdesign propulsion models for reciprocating engines and turbofans.
Electric and fuel cell propulsion system models should be added so that UAVs
incorporating these options can be evaluated.
Future UAV design code enhancements should incorporate launch and recovery
performance and system models. The constraints associated with this analysis will likely
drive the configuration. Takeoff and landing speeds, climb angles, field length, and
other considerations may be drivers for UAVs equipped with conventional landing gear.
238
Pneumatic and rocket launch system integration will have structural impacts. Parachute
recovery systems take up volume that scales as a function of the vehicle weight and
descent requirements.
The wing structural model should be improved to capture more physical phenomena.
Improvements should include higher quality secondary structure and control surface
analysis. Key structural phenomena of interest include flutter analysis and stiffness
based design capabilities. The model should be more highly tailored to composites
materials.
Future model should capture local poststall behavior on the configuration. The current
limitation that all surfaces operate in the approximately linear region prior to stall
imposes limitations on the maximum permissible lift coefficient of the vehicle. In reality,
local stall conditions on some portion of the vehicle typically occur prior to the vehicle
achieving the maximum lift coefficient. The nonlinear behavior of the airfoils prior to
and post stall creates limitations on the fidelity of the matrix methods for lift distribution
calculations. The numerical lifting line theory presented in Anderson [1991] addresses
these considerations for a wing, but not a lifting system. A similar numeric method could
be applied to the vortex lattice method, incorporating the poststall airfoil models
developed in this research.
An entire graduatelevel research effort could be dedicated exclusively to defining
technology trends. Potential emphasis can be placed on generating physics and history
based technology trends. This research helps define useful formats for these technology
predictions. As discussed previously, existing technology trends do not always lend
themselves to integration into aircraft design and optimization codes.
As the vehicle size is reduced, the survivability will likely increase. Decreased visual,
acoustic, and infrared signature levels, and potential decrease in radar cross section will
reduce the probability of detection. Additionally, the overall vulnerability of the AV will
decrease due to smaller presented area of flight critical components to threat systems.
Future research efforts could focus on survivability benefits gained though integration of
advanced technologies.
239
Appendix A Geometry Methods
Wing Geometry
This section provides the trapezoidal geometry formulation for the wing segments. The
index for the number of wings is i, and the index for the number of segments for a given
wing is j. For clarity, the index i is dropped from this discussion.
Some general geometry parameters of interest for the trapezoidal wing elements are
shown in Figure A1.
Figure A1 Wing Geometry Parameters
First, a nondimensional wing shape factor, SF, parameter must be defined. SF is
analogous to a complex taper parameter. Note that if a single panel wing with a taper
ratio of 1 is applied, then SF is equal to 2.
( ) ( ) ( )  
∑
=
− − ⋅ − − + + ⋅ =
npan
j
w r t w r t tip tip r,w t tip
j c c j c c j j c c SF
2
, ,
) 1 ( ) ( ) 1 ( ) ( ) 1 ( 1 ) 1 ( η η η
The wing area equations are:
SF c b S
w r w w
⋅ ⋅ ⋅ =
,
2
1
w
w
w
AR
b
S
2
=
2 2
,
4
1
SF c AR S
w r w w
⋅ ⋅ ⋅ =
The wing root chord equations are:
240
SF AR
b
c
w
w
w r
⋅ ⋅
=
2
1
,
SF b
S
c
w
w
w r
⋅ ⋅
=
2
1
,
SF
AR
S
c
w
w
w r
⋅
=
2
,
The wing span equations are:
SF c
S
b
w r
w
w
⋅
⋅
=
,
2
SF c AR b
w r w w
⋅ ⋅ ⋅ =
,
2
1
w w w
S AR b ⋅ =
The wing aspect ratio equations are:
SF c
b
AR
w r
w
w
⋅
⋅
=
,
2
w
w
w
S
b
AR
2
=
2 2
,
4
SF c
S
AR
w r
w
w
⋅
⋅
=
To convert from the complex wing geometry to an equivalent single taper, the following
equivalent taper ratio equation is used:
1
2
,
−
⋅
⋅
=
w r w
w
eq
c b
S
λ
The segment paneling can now begin. It is necessary to know the leading edge Cartesian
coordinates of the root and tip, leading edge sweep, quarter chord sweep, root and tip
chords, and the root and tip incidence angles of all wing panels.
The segment dihedralindependent semispan, s
seg
, for a wing or horizontal tail is found
by:
2
) 1 ( ) 1 (
w
tip seg
b
s ⋅ =η if j=1
241
( ) ) 1 ( ) (
2
) ( − − ⋅ = j j
b
j s
tip tip
w
seg
η η if j>1
and for a vertical tail:
w tip seg
b s ⋅ = ) 1 ( ) 1 ( η if j=1
( ) ) 1 ( ) ( ) ( − − ⋅ = j j b j s
tip tip w seg
η η if j>1
Similarly, the dihedralindependent segment tip location relative to the wing root for a
wing or horizontal tail is:
2
) ( ) (
w
tip tip
b
j j s ⋅ =η
and for a vertical tail:
w tip tip
b j j s ⋅ = ) ( ) ( η
The segment root and tip chords are found by:
) ( ) ( ) (
, , , ,
j c j c c j c
w r w r r seg w r
⋅ =
) ( ) ( ) (
, , , ,
j c j c c j c
w r w r t seg w t
⋅ =
The segment root incidence angle is found by:
) 1 ( ) 1 (
, w w r
inc inc = if j=1
) 1 ( ) (
,
− − = j inc j inc
w w w r
ε if j>1
The segment tip incidence angle is found by:
) ( ) (
,
j inc j inc
w w w t
ε − =
The segment quarter chord sweep for a wing or horizontal tail is:
( )
( )


.

\

− ⋅
−
+ Λ = Λ
Λ −
) ( ) (
2 /
25 . 0 ) ( /
tan ) ( ) (
, , , ,
, 1
. , 4 /
j c j c
b
j c x
j j
seg w t seg w r
w
w
w x seg c
and for a vertical tail:
242
( )
( )


.

\

− ⋅
−
+ Λ = Λ
Λ −
) ( ) (
25 . 0 ) ( /
tan ) ( ) (
, , , ,
, 1
. , 4 /
j c j c
b
j c x
j j
seg w t seg w r
w
w
w x seg c
The segment leading edge sweep for a wing or horizontal tail is:
( ) ( )


.

\

− ⋅ + Λ = Λ
Λ −
) ( ) (
2 /
) ( /
) ( tan tan ) (
, , , ,
,
,
1
,
j c j c
b
j c x
j j
seg w t seg w r
w
w
w x seg LE
and for a vertical tail:
( ) ( )


.

\

− ⋅ + Λ = Λ
Λ −
) ( ) (
) ( /
) ( tan tan ) (
, , , ,
,
,
1
,
j c j c
b
j c x
j j
seg w t seg w r
w
w
w x seg LE
Now the process of finding the coordinates for the segment root and tip leading edges can
begin. Starting with the most inboard segment (j=1), the wing root coordinates are:
w w r
P P = ) 1 (
,
otherwise (j>1),
) 1 ( ) (
, ,
− = j P j P
w t w r
Next, the difference in leading edge coordinates between the segment root and tip if
dihedral is neglected is calculated:
( ) ( )
( ) ( )
( ) ( ) ) ( cos 1 ) ( / ) (
) ( cos 1 ) ( / ) (
) ( cos ) ( tan ) (
, , ,
, , ,
, ,
j inc j c x j c
j inc j c x j c
j inc j j s x
w r seg w r
w t seg w t
w r seg LE seg
− ⋅ ⋅
− − ⋅ ⋅
+ ⋅ Λ ⋅ =
ε
ε
δ
) ( j s y
seg
= δ
( ) ( )
( )
( ) ) ( sin ) ( / ) (
) ( sin ) ( / ) (
) ( sin ) ( tan ) (
, , ,
, , ,
, ,
j inc j c x j c
j inc j c x j c
j inc j j s z
w r seg w r
w t seg w t
w r seg LE seg
⋅ ⋅
− ⋅ ⋅
+ − ⋅ Λ ⋅ =
ε
ε
δ
Now the effects of dihedral can be taken into consideration:
( ) ( ) ( ) ) ( cos ) ( sin ) ( sin
, ,
j inc j inc j y x x
w r w r w
⋅ ⋅ Γ ⋅ + = ∆ δ δ
( ) ) ( cos j y y
w
Γ ⋅ = ∆ δ
( ) ( ) ) ( 2 cos ) ( sin
,
j inc j y z z
w r w
⋅ Γ ⋅ + = ∆ δ δ
243
The segment wing tip locations are then found by:
x j X j X
w r w t
∆ + = ) ( ) (
, ,
y j Y j Y
w r w t
∆ + = ) ( ) (
, ,
z j Z j Z
w r w t
∆ + = ) ( ) (
, ,
The overall projected planform reference area for a wing is:
( ) ( )  
∑
=
Γ ⋅ ⋅ − =
npan
j
w seg seg w t seg w r ref
j j s j c j c S
1
, , , ,
) ( cos ) ( ) ( ) (
The overall projected reference span for a wing and horizontal tail is:
( )  
∑
=
Γ ⋅ ⋅ =
npan
j
w seg ref
j j s b
1
) ( cos ) ( 2
and for a vertical tail:
( )  
∑
=
Γ ⋅ =
npan
j
w seg ref
j j s b
1
) ( cos ) (
Lift Distribution Paneling
There are two primary methods of paneling the wing. The first method panels each
segment individually. The second method panels the entire wing by blending segments.
For the first method, each wing segment is broken into n
pan
panels. For the index k
values from 1 to n
pan
, the following procedure is followed. The spanwise position of the
segment root and tip before dihedral considerations is determined. Either even or
sinusoidal spacing may be used. The sinusoidal spacing places more panels towards the
tip to more fully capture the rapidly changing behavior at the wing tip. The even spacing
root and tip panel spanwise locations, respectively, are:
pan
seg
n
k
j s ys
1
) ( 1
−
⋅ = root
pan
seg
n
k
j s ys ⋅ = ) ( 2 tip
The sinusoidal spacing equivalents are:
244


.

\

⋅
−
⋅ =
2
1
sin ) ( 1
π
pan
seg
n
k
j s ys root


.

\

⋅ ⋅ =
2
sin ) ( 2
π
pan
seg
n
k
j s ys tip
Notice that no index is associated with ys1 or ys2. This is because each of these
parameters is calculated for each panel, and the information does not need to be retained.
The difference between ys1 and ys2, ∆ys(k), will be used for other calculations.
The nondimensional panel spanwise station is:
w
seg tip
b
ys ys
j s j s
k
2
2 1
) ( ) (
2 ) (
+
+ −
⋅ = η
The root and tip chords of this panel are:
( )
) (
1
) ( ) ( ) ( 1
, , , , , ,
j s
ys
j c j c j c c
seg
seg w r seg w t seg w r
⋅ − + = root chord
( )
) (
2
) ( ) ( ) ( 2
, , , , , ,
j s
ys
j c j c j c c
seg
seg w r seg w t seg w r
⋅ − + = tip chord
The incidence angles of the root and tip of this panel are:
( )
) (
1
) ( ) ( ) ( ) ( 1
, , ,
j s
ys
j inc j inc j inc k inc
seg
w r w t w r
⋅ − + = root incidence
( )
) (
2
) ( ) ( ) ( ) ( 2
, , ,
j s
ys
j inc j inc j inc k inc
seg
w r w t w r
⋅ − + = tip incidence
The cartesian coordinates for the quarter chord of the panel root are:
( ) ) ( 1
4
1
) (
1
) ( ) ( ) ( ) ( 1
, , ,
k c
j s
ys
j X j X j X k X
seg
w r w t w r
⋅ + ⋅ − + =
( )
) (
1
) ( ) ( ) ( ) ( 1
, , ,
j s
ys
j Y j Y j Y k Y
seg
w r w t w r
⋅ − + =
( )
) (
1
) ( ) ( ) ( ) ( 1
, , ,
j s
ys
j Z j Z j Z k Z
seg
w r w t w r
⋅ − + =
The cartesian coordinates for the quarter chord of the panel tip are:
245
( ) ) ( 2
4
1
) (
2
) ( ) ( ) ( ) ( 2
, , ,
k c
j s
ys
j X j X j X k X
seg
w r w t w r
⋅ + ⋅ − + =
( )
) (
2
) ( ) ( ) ( ) ( 2
, , ,
j s
ys
j Y j Y j Y k Y
seg
w r w t w r
⋅ − + =
( )
) (
2
) ( ) ( ) ( ) ( 2
, , ,
j s
ys
j Z j Z j Z k Z
seg
w r w t w r
⋅ − + =
The second wing paneling method blends the wing segments such that the panels are
distributed from wing root to wing tip rather than from segment root to tip. This method
is primarily used for the lifting line theory, though the vortex lattice method and Trefftz
plane analysis can use it as well.
The even spacing root and tip panel spanwise locations are:
pan
w
n
k
b ys
1
1
−
⋅ = root
pan
w
n
k
b ys ⋅ = 2 tip
The sinusoidal spacing equivalents are:


.

\

⋅
−
⋅ =
2
1
sin 1
π
pan
w
n
k
b ys root


.

\

⋅ ⋅ =
2
sin 2
π
pan
w
n
k
b ys tip
The nondimensional panel spanwise station is:
w
b
ys ys
k
2 1
) (
+
= η
The parameter θ(k) is used for the lifting line method. It can be thought of as an angular
equivalent of the spanwise station. It is defined as:


.

\
 +
=
−
w
b
ys ys
k
2 1
cos ) (
1
θ
Calculation of the other parameters uses the same equations as shown above. The one
major difference is the selection of segment index j, which is now dependent upon the
value of ys1 and ys2.
246
Regardless of which paneling method is used, other parameters can now be determined.
The average panel chord is:
2
2 1
) (
c c
k c
ave
+
=
The 3/4chord panel control point used in the vortex lattice method and, except for the x
coordinate, in the Trefftz plane analysis is:
) (
2
1
2
) ( 2 ) ( 1
) ( k c
k X k X
k Xm
ave
⋅ +
+
=
2
) ( 2 ) ( 1
) (
k Y k Y
k Ym
+
=
2
) ( 2 ) ( 1
) (
k Z k Z
k Zm
+
=
The point used to simulate the Trefftz plane (infinitely downstream) for induced drag
calculations in the vortex lattice method is:
w
b k Xi ⋅ =100 ) (
) ( ) (
) ( ) (
k Zm k Zi
k Ym k Yi
=
=
where b
w
is the span of the main wing.
The equivalent panel dihedral angle, φ(k), is found by:


.

\

−
−
=
−
) ( 1 ) ( 3
) ( 1 ) ( 2
tan ) (
1
k Y k Y
k Z k Z
k φ
The panel incidence angle is:
2
) ( 2 ) ( 1
) (
k inc k inc
k inc
+
=
247
Appendix B Blade Element Propeller
Methods
B.1 Overview
The propeller design and evaluation codes were developed from the work of Charles
Adkins and Robert Liebeck described in their paper entitled, Design of Optimum
Propellers [1994]. The methods used here differ from those of previous optimal
propeller design codes in three ways. First, small angle approximations are not
necessary. Second, there is no light disk loading requirement on the propeller. Third,
viscous terms are included in the induced velocity calculations. (The airfoil drag affects
the circulation distribution radially across the propeller blade). These methods bring
optimal propeller design into agreement with analysis, which is an improvement on
earlier blade element methods.
B.2 Blade Element Model Propeller Design
Design Results
A sample design and analysis case for a general aviation propeller was provided in
Liebeck [1994]. In general, the results of this code correspond to the sample case with
suitable repeatability.
There is little difference between the two airfoil interpolation methods. The differences
between the code and the sample case can be partially explained by the use of different
airfoil data methods. Adkins and Liebeck [1994] does not discuss the airfoil data lookup
method in detail. Other differences may be explained by subtleties of two different
codes. The chord distribution agreement is much better than that of the twist
distribution.
The code was also evaluated for a threebladed propeller under the same design
conditions. The results in Figure B (AB) indicate that the local chord size is sensitive to
the blade count, but the form of the chord distribution remains the same. The blade twist
is very insensitive to blade count.
248
Chord Comparison of 2 and 3 Blade
Props
0
0.1
0.2
0.3
0.4
0.5
0.6
0 1 2 3
Local Radius, FT
L
o
c
a
l
C
h
o
r
d
,
F
T
2Blade
3Blade
Beta Comparison of 2 and 3 Blade
Props
0
10
20
30
40
50
60
70
0 1 2 3
Local Radius, FT
B
e
t
a
,
D
E
G
(A) (B)
Figure B1 Comparison Between Two and Three Bladed Propeller Designs
B.3 Blade Element Model Propeller OffDesign Analysis
The offdesign analysis is necessary to determine how the propeller performs during the
majority of operation. Propellers never act as point designs, rather propellers experience
a range of operational conditions. The bladeelement propeller offdesign analysis is
much more realistic at predicting offdesign conditions than the disk actuator method.
An added advantage of this method is the ability to predict static thrust.
The lift and drag coefficients are found as a function of angle of attack and Reynolds
number through the airfoil interpolation subroutine. Note that the angle of attack and
drag coefficients were determined in a similar way in the design code, except these
parameters were functions of the lift coefficient and Reynolds number. By using the
angle of attack as a lookup variable, it is possible to find feasible solutions of the lift and
drag coefficients even beyond stall. A special model was created as part of the airfoil
interpolation subroutines to calculate poststall lift and drag values from 0 to 360 degrees
angle of attack. These poststall airfoil methods are described in the aerodynamics
chapter.
The agreement with Liebeck [1994] is good in all cases. Both airfoil interpolation
methods are presented here as well.
249
Appendix C Selected Avionics and
Subsystem Weights Methods
C.1 Autopilot
Autopilots permit UAV systems to fly without pilots. Autopilots are defined here as
electronic systems that control AV functions. There are four categories considered: radio
control receivers, RPV autopilots, autonomous autopilots, and intelligent autopilots.
Radio Control (RC) Receivers
RC receivers have a broad range of sizes and capabilities. RC systems are generally
driven by the model aircraft hobby industry.
The minimum weight of a RC receiver is assumed to be 0.1 ounces. The average RC
receiver weight is 3 ounces. These numbers are based on typical commercially available
RC systems. If a commercially available GPS system is included, the weight increases
by approximately 4.4 ounces. If a low grade INS system is included, the weight increases
by approximately 8 ounces.
RPV Autopilots
RPV autopilots contain, at minimum, an inner loop autopilot and a means of determining
vehicle angular rates. The pilot typically controls the vehicle by issuing control
commands in response to video and other data from the AV. RPV autopilots are
becoming displaced by small autonomous autopilots.
The minimum weight of a RPV autopilot is assumed to be 2 ounces. The average RPV
autopilot is assumed to weigh 1.5 pounds. If a GPS system is included, the weight
increases by 8 ounces. If an INS system is included, the weight increases by 8 ounces.
These weight increases are causes by the additional components plus the weight of the
processors.
Autonomous Autopilots
Autonomous autopilots have the ability to perform most flight operations without pilot
intervention. Pilots may or may not be in the flight control loop for launch and recovery
operations. Most autonomous autopilots permit pilots to take control of the AV, and
operate in a sophisticated RPV mode.
250
The minimum weight of an autonomous autopilot is assumed to be 4 ounces. The
average autonomous autopilot weight is assumed to weigh 2 pounds. The addition of an
INS system is assumed to add 1 pound. The addition of a GPS system is assumed to
weigh 8 ounces.
Intelligent Autopilots
Intelligent autopilots provide additional functionality over autonomous autopilots.
Intelligent autopilots can determine vehicle health, optimize collection, assist group UAV
operations, and have adaptable characteristics.
The weights are assumed to be the same as for the autonomous autopilots, because little
additional hardware is required. The difference between autonomous and intelligent
autopilots primarily lies in the processing, which is driven by software.
General Autopilot Considerations
The autopilot weight is corrected for ruggedness, performance, and miniaturization. The
corrections are valid for avionics boards and boxes. The factors are shown in Table C1.
Driver Minimum (0) Mid (0.5) Maximum (1)
Ruggedness 0.4 1.0 1.5
Performance 0.4 1.0 2.0
Miniaturization 0.4 1.0 1.5
Table C1 Autopilot Factors
A limited database of autopilot systems was established to provide guidance on the range
of autopilot solutions. This research influenced the above factors. While not exact, these
correction factors provide an approximation where none previously existed.
After corrections, if the uninstalled autopilot weight is less than the minimum autopilot
weight, then the uninstalled autopilot weight will be set to the minimum autopilot weight.
The volume for an autopilot is calculated. The reference density is assumed to be 45
pounds/cubic foot, which is based on Nicolai [1984]. The actual installed volume
occupied by the autopilot is assumed to be 300% of the autopilot alone.
The autopilot power is assumed to be 4 Watts/pound.
C.2 Inertial Navigation System (INS)
Inertial Navigation Systems (INS) are used for almost all UAVs. Some unsophisticated
systems that are little more than radiocontrolled aircraft are the exception. All UAVs
251
under consideration in this study use an INS. The two categories of INS covered here are
rate based and navigation grade INS systems.
Rate based INS systems provide angular rates and sometimes linear accelerations. Rate
based INS systems are typically used by the inner loop of the flight control system. The
minimum weight for a rate based INS system is assumed to be 2 ounces. A typical rate
based INS system is assumed to weigh 2 pounds.
Navigation INS systems integrate the angular rates to find the angles and integrate the
linear accelerations to find velocity. Often times the drift is contained by corrections
enabled by GPS or air data sources. The minimum weight for a navigation INS system is
assumed to be twice the weight of a C/A code GPS, because GPS is necessary if the INS
sensors are crude. A typical navigation INS system is assumed to weight 5 pounds.
The INS weight is corrected for ruggedness, performance, and miniaturization. The
factors are shown in Table C2.
Driver Minimum (0) Mid (0.5) Maximum (1)
Ruggedness 0.4 1.0 1.5
Performance 0.4 1.0 2.0
Miniaturization 0.1 1.0 1.5
Table C2 INS Factors
A limited database of INS systems was established to provide guidance on the range of
INS solutions. This research influenced the above factors. While not exact, these
correction factors provide an approximation where none previously existed.
After corrections, if the uninstalled INS weight is less than the minimum INS weight,
then the uninstalled INS weight will be set to the minimum INS weight.
The installed weight of the INS is assumed to be 150% of the uninstalled INS.
The volume for an INS is calculated. The reference density is assumed to be 70
pounds/cubic foot, which is based upon the popular LN100G INS. The actual installed
volume occupied by the INS is assumed to be 200% of the INS alone.
The INS power is assumed to be 1.73 Watts/pound, which is also based on the Litton
LN100G.
C.3 Global Positioning System (GPS)
GPS is an essential element of modern UAV systems. Prior to the advent of GPS, it was
necessary to fly unmanned aircraft as radio controlled models, as RPVs, or to accept
252
inaccurate navigation. GPS, when coupled to an autopilot or other navigation system,
allows a UAV to navigate long distances accurately.
The basic GPS weight for a commercially available (C/A) code GPS is assumed to be 4.4
ounces based upon the TracPac GPS receiver. A precision (P) code GPS is assumed to
be 50% heavier, because there are no commercial applications to drive the weight down.
The GPS power, cost, and volume are calculated based upon the weight using the
functions for the INS. The weight and volume installation factors are calculated using
the INS functions.
C.4 Processors
Processors are typically necessary to perform a variety of tasks. While autopilots
typically take care of flight controls and managing flight critical systems, processors are
often required. Some processor tasks may involve communications management,
payload management, or preprocessing sensor data prior to sending to the autopilot.
A typical processor is assumed to weigh 3 pounds. This weight is corrected for
ruggedness, performance, and miniaturization according to the boards and boxes
functions described earlier. The power, volume, and installation factors are also found
using boards and boxes functions.
C.5 Cameras
Cameras are often needed on UAVs. This can be to aid in takeoff in landing operations
or to provide general situational awareness. These cameras are considered distinct from
cameras that are part of the payload.
Three types of cameras are considered: Electrooptical (EO), Infrared (IR), and lowlight.
The weight of a typical EO camera is assumed to be 0.5 pounds. The weight of a typical
IR camera is assumed to be 2 pounds. The weight of a typical lowlight camera is
assumed to be 1 pound.
The camera weight is corrected for ruggedness, performance, and miniaturization. The
factors are shown in Table C3.
Driver Minimum (0) Mid (0.5) Maximum (1)
Ruggedness 0.5 1.0 2.0
Performance 0.5 1.0 4.0
Miniaturization 0.5 1.0 2.0
Table C3 Camera Factors
253
The power and volume are found using the boards and boxes functions. The weight and
volume installation factors are also found using the boards and boxes functions.
C.6 Recorders
Recorders are sometimes necessary for recording payload data, flight data, and any other
data of interest.
The recorder is assumed to weigh 0.5 pound per gigabyte.
The volume and power is found using the associated boards and boxes formulas. The
weight and installation factors are from the boards and boxes formulas.
C.7 Air Traffic Control (ATC)
Several ATC options are covered in the code. ATC equipment makes it possible for
UAVs to operate in the same airspace as manned aircraft. While the FAA and foreign
ATC services have evolving UAV airspace operability requirements, the equipment
covered here will be relevant.
ATC Radios
UAV systems may use ATC radios in one of three ways. First, the voice
communications between the ATC personnel and the UAV ground personnel may be
relayed through the AV. Second, the AV may generate a voice signal to ATC and
interpret the ATC voice via voice recognition software. Third, the UAV system ground
personnel may communicate directly with the local ATC with ground station ATC radios.
Only the first two options affect the AV weight, so the third option is dropped from
consideration.
The Filser ATR 720 A [Aircraft Spruce and Specialty 2002] ATC radio is used as a
reference. The radio characteristics are shown below.
Weight: 1.56 pounds
Volume: 51 in
3
Power: 4 Watts output (Assume 16 Watts input – 25% efficiency)
Cost: $1095 (2002)
The ATC radio requires an antenna. A typical VHF ATC antenna has the following
characteristics:
Weight: 0.5 pounds
Volume: 0. (Almost all external)
Power: 0.
Cost: $97.50 (2002)
254
Mode 3A
Mode 3A transponders provide an identification code to ATC when interrogated. The
ATC determines the aircraft position as the ATC radar makes the 360
o
sweeps. The
reference transponder is the Garmin GTX327 [Aircraft Spruce and Specialty Co. 2002].
This is used for modes 3A, C, and S. The characteristics are shown below.
Weight: 2.1 pounds
Volume: 83.5 in
3
Power: 15 Watts
Cost: $1895 (2002)
The weight of the Mode 3A ATC also includes the ATC antenna weight.
Mode C
Mode C transponders have all of the features of the Mode 3A, but also add an altitude
reporting capability. This requires an altitude encoder. The AmeriKing altitude encoder
[Aircraft Spruce and Specialty Co 2002] is used as a reference. The characteristics are
shown below.
Weight: 0.4 pounds
Volume: 33.25 in
3
Power: 1 Watt (Guess)
Cost: $500 (The cost is altitude dependent)
All other components and weights from the Mode 3A are included.
Mode S
Mode S transponders enable pilots to determine the position of other Mode S aircraft
relative to their aircraft. This is a relatively new mode that is not widely adopted. Mode
S is incorporated into TCAS, an aircraft collision avoidance system. Mode S uses all of
the features of Mode C, and the weights are considered identical for the purposes of this
code.
General ATC Considerations
No technology factors are applied to the ATC equipment for this formulation. Factors
are applied when entered directly as avionics. ATC equipment is relatively mature. The
hardware is largely driven by commercial and general aviation, which are slow to adopt
new technologies. It is conceivable that a UAV program could modify the equipment,
but this may introduce FAA acceptance issues.
255
Installation weight and installation volume factors are applied to the ATC equipment.
The functions for avionics boards and boxes are used.
256
Bibliography
Abbot, Ira H., and Von Doenhoff, Albert E., Theory of Wing Sections, Including a
Summary of Airfoil Data, Dover Publications, Inc., New York, 1959.
Adkins, Charles N., Liebeck, Robert H., Design of Optimum Propellers, AIAA830190,
AIAA 21
st
Aerospace Sciences Meeting, January 1983.
Adkins, Charles N., Liebeck, Robert H., Design of Optimum Propellers, AIAA Journal
(NFI), Volume 10, Number 5, SeptOctober 1994.
Aircraft Spruce and Specialty Co., 20022003 Catalog, Aircraft Spruce and Specialty
Co., 2002.
Air Force Scientific Advisory Board (SAB), UAV Study, Air University, 1996.
Althaus, Dieter, Profilpolaren fur den Modellflug, Band 2, NeckarVerlag GMBH,
Germany, 1985.
Altmann, Herman, A Study of a Reconnaissance Surveillance Vehicle, Course Materials
Presented at the UCLA UAV Aircraft Design Short Course, August 2002.
Anderson, John D., Fundamentals of Aerodynamics, Second Edition, McGrawHill, Inc.,
United States of America, 1991.
Bertin, John J., and Smith, Michael L., Aerodynamics for Engineers, Second Edition,
Prentice Hall, Inc., Englewood Cliffs, NJ, 1989.
Boucher, Robert J., Electric Motor Handbook, Robert J. Boucher, Los Angeles, 1994.
Brown, David R., Unmanned Combat Aerial Vehicles: Evolution or Potential
Revolution?, Air University, April 1998.
Brown, Robert B., Compass Cope Airframe Design History, AIAA Journal of Aircraft,
Volume 17, Number 12, December 1980.
Brown, Stuart F., "Batteries are Getting Charged Up," Fortune, October 27, 1997.
Carmichael, Bruce W., DeVine, Troy E., et. al., Strikestar 2025, Research Paper
Presented to Air Force 2025, August 1996.
Carroll, David L., GA170 Code Listing, University of Illinois at Urbana Champaign,
December 1998.
257
Davenport, William, Virginia Tech AOE 5104 Graduate Incompressible Aerodynamics
Course Notes, 1998.
Defense Airborne Reconnaissance Office (DARO), UAV Annual Report FY 1996, 6
November 1996.
Defense Airborne Reconnaissance Office (DARO), UAV Annual Report FY 1997, 1997.
Defense and Aerospace, Technology Foresight – Progress Through Partnerships, Office
of Science and Technology for HMSO, Vol 12, April 1995.
Drela, Mark, XFOIL 6.9 User Guide, MIT Aero and Astro, 2000.
Drezner, Jeffrey A., and Leonard, Robert S., Innovative Development, Global Hawk and
Darkstar, Executive Summary and Vol 13, RAND Project Air Force, RAND, 2002.
Etkin, Bernard, and Reid, Lloyd D., Dynamics of Flight, Stability and Control, Third
Edition, John Wiley and Sons, Inc, Canada, 1996.
Fahlstrom, Paul G., and Gleason, Thomas J., Introduction to UAV Systems, Second
Edition, UAV Systems Inc., Columbia, Maryland, 1998.
Foch, Richard J., Ailinger, Kevin G., Low Reynolds Number Long Endurance Aircraft
Design, AIAA 921263, AIAA, 1992.
Goldberg, David E., Genetic Algorithms in Search, Optimization, and Machine Learning,
Addison Wesley Longman, Inc., USA, 1989.
Grasmeyer, Joel, TrussBraced Wing Code Description and User's Manual, VPIAOE
255, Virginia Tech, 1998.
1. Gundlach, John F., Development of the Virginia Tech Forestry Unmanned Aerial
Vehicle, AIAA 990003, 1999.
2. Gundlach, John F., Multidisciplinary Design Optimization and Industry Review of a
2010 StrutBraced Wing Transonic Transport, Masters Thesis, Virginia Tech, 1999.
Gundlach, John F., Tetrault, P.A., Gern, F.H., NagshinehPour, A.H., Ko, A.K., Schetz,
J.A., Mason, W.H., Kapania, R.K., Grossman, B., Haftka, R.T., Conceptual Design
Studies of a StrutBraced Wing Transonic Transport, AIAA Journal of Aircraft, Volume
37, Number 6, December 2000.
Hall, David W., Fortenbach, Charles D., et. al., A preliminary Study of Solar Powered
Aircraft and Associated Power Trains, NASA CR 3699, 1983.
258
Hall, David W., Hall, Stan A., Structural Sizing of a Solar Powered Aircraft, NASA
Contractor Report 172313, 1984.
Hill, Philip G., Peterson, Carl R., Mechanics and Thermodynamics of Propulsion, Second
Edition, AddisonWesley Publishing Company, Inc., Reading, Massachusetts, 1992.
Hoerner, S.F., FluidDynamic Drag, Hoerner Fluid Dynamics, Vancouver, WA, 1992.
Holder, Bill, Unmanned Vehicles, An Illustrated History, Schiffer Publishing Inc, China,
2001.
Irwin, Col. David W., Jr., Reconnaissance is Black, Turner Publishing Co., Paducah,
Kentucky, 2000.
Johnstone, R., Arntz, N., CONDOR – High Altitude Long Endurance (HALE)
Automatically Piloted Vehicle (APV), AIAA 903279, AIAA, 1990.
McDaid, Hugh, and Oliver, David, Smart Weapons, Top Secret History of Remote
Controlled Airborne Weapons, Barnes and Noble, Inc., Italy, 1997.
Munson, Kenneth, Jane’s Unmanned Aerial Vehicles and Targets, Issue Thirteen, Jane’s
Information Group Limited, UK, 1999.
Munson, Kenneth, World Unmanned Aircraft, Jane’s Publishing Company Limited,
London, 1988.
L3 Communications, S/TAR Strategic/Tactical Airborne Recorder, Brochure presented
at AUVSI 2003, 2003.
Libeau, Michael, Design Visualization Using the DXF.F Subroutine, Unpublished
documentation supporting the Virginia Polytechnic and State University StrutBraced
Wing (SBW) transonic transport research, May 1999.
Macci, S. H., SemiAnalytical Method for Predicting Wing Structural Mass, 54
th
Annual
Conference of the Society of Allied Weights Engineers, Inc., Huntsville Alabama, 1995.
Mattingly, J. D., Et. Al., Aircraft Engine Design, AIAA Education Series, AIAA, New
York, 1987.
McCormick, Barnes W., Aerodynamics Aeronautics and Flight Mechanics, Second
Edition, John Wiley and Sons, Inc., New York, 1995.
Mises, Richard V., Theory of Flight, Dover Publications, Toronto, 1959.
259
NaghshinehPour, Amir H., Preliminary Structural Analysis of a StrutBraced Wing,
VPIAOE256, Department of Aerospace and Ocean Engineering, Virginia Polytechnic
Institute and State University, 1998.
1. The National Academy of Sciences, Review of ONR’s Uninhabited Combat Air
Vehicles Program, National Academy Press, Washington DC, 2000.
2. The National Academy of Sciences, Uninhabited Air Vehicles, Enabling Science for
Military Systems, National Academy Press, Washington DC, 2000.
Nelson, Robert C., Flight Stability and Automatic Control, McGraw Hill, Inc., Mexico,
1989.
Nichols, Mark J., and Martel, William C., UCAVs and Commercial Satellites: "The
Missing Link", Air War College, Air University, April 1998.
Nicolai, Leland M., Fundamentals of Aircraft Design, METS, inc., USA, 1984.
Noor, Ahmed K., Future Aeronautical and Space Systems, Volume 172, Progress in
Astronautics and Aeronautics, AIAA, Reston, VA, 1997.
Noor, Ahmed K., Structures Technology for Future Aerospace Systems, Volume 188,
Progress in Astronautics and Aeronautics, AIAA, Reston, VA, 2000.
Northrop Grumman, RQ4A Global Hawk, High Altitude Endurance Unmanned Aerial
Vehicle, Brochure presented at AUVSI 2003, Northrop Grumman 452AS4209_06.03,
2003.
Oates, G. C., The Aerothermodynamics of Gas Turbine and Rocket Propulsion, AIAA
Education Series, AIAA, New York, 1984.
Office of the Secretary of Defense (OSD), Unmanned Aerial Vehicles Roadmap 2000
2025, Washington DC, April 2001.
Office of the Secretary of Defense (OSD), Unmanned Aerial Vehicles Roadmap, 2002
2027, Draft Version 7, Washington DC, November 2002.
O'Reilly, Thomas G., and Beno, Andrew M., Uninhabited Air Vehicles  Critical
Leverage Systems for Our Nation's Defense in 2025, Air Command and Staff College,
Air University, April 1999.
Palumbo, Dominic J., Design Evolution of the Shadow 200 Tactical Unmanned Aerial
Vehicle, Unmanned Systems 2000, Association of Unmanned Vehicle Systems
International, 2000.
260
Power Computing Solutions, Inc. (PCS), Electric Power System for High Altitude UAV
Technology Survey, NASA/CR97206337, 1997.
Raymer, Daniel P., Aircraft Design: A Conceptual Approach, 2
nd
Edition, AIAA
Education Series, Washington, D.C., 1992.
Raymer, Daniel P., UAV Aircraft Design, UCLA Short Course Materials, August 2002.
Ritchie, D., Aerodynamics Course Notes, Academy of Aeronautics, Long Island, New
York, 1968.
Roskam, Jan, Airpane Design, Parts IVIII, Roskam Aviation and Engineering
Corporation, Ottawa, Kansas, 1990.
SAWE (Society of Allied Weights Engineers), Introduction to Aircraft Weight
Engineering, SAWE Inc, Los Angeles, 1996.
Selig, Michael S., Donovan, John F., and Fraser, David B., Airfoils at Low Speed,
Soartech 8, H.A. Stokely, Virginia Beach, VA, 1989.
Selig, Michael S., Et. Al., Summary of LowSpeed Airfoil Data, Volume 1, SoarTech
Publications, Virginia Beach, VA, 1995.
Sheldahl, R. E. and Klimas, P. C., Aerodynamic Characteristics of Seven Airfoil Sections
Through 180 Degrees Angle of Attack for Use in Aerodynamic Analysis of Vertical Axis
Wind Turbines, SAND802114, March 1981, Sandia National Laboratories,
Albuquerque, New Mexico.
Sloan, Joel, Design and Packaging of Electronics Equipment, Van Nostrand Reinhold
Company, NY, 1985.
Somers, Dan M., Maughmer, Mark D., The SM701 Airfoil, An Airfoil for World Class
Sailplanes, The Pennsylvania State University, Circa 1992.
Stinton, Darrol, The Design of the Aeroplane, Blackwell Sciences Ltd., Oxford, 1995.
Sullivan, Lt. Col. James F., Brouillette, Maj. Gregory, Joles, Maj. Jeffery, 1998 Army
After Next Unmanned Aerial Vehicle Studies, Operations Research Center, July 1998.
Wagner, William, Sloan, William P., Fireflies and other UAVs (Unmanned Aerial
Vehicles), Aerofax Inc, Arlington, Texas, 1992.
Wagner, William, Lightning Bugs and other Reconnaissance Drones, The cando story of
Ryan’s unmanned ‘spy planes’, Armed Forces Journal International, Fallbrook,
California, 1982.
261
Werrell, Kenneth P., The Evolution of the Cruise Missile, Air University Press, Maxwell
AFB, Alabama, September 1985.
Whitford, R., Fundamentals of Fighter Design, Airlife Publishing Ltd., UK, 2000.
262
List of Acronyms
AAI Company name
AC Aerodynamic Center
ACTD Advanced Concept Technology Demonstrator
AIAA American Institute of Aeronautics and Astronautics
ATC Air Traffic Control
AUVS Association of Unmanned Vehicle Systems
AUVSI Association of Unmanned Vehicle Systems International
AV Air Vehicle
BLOS Beyond Line of Sight
CFD Computational Fluid Dynamics
COMINT Communications Intelligence
DGPS Differential Global Positioning System
ECS Environmental Control System
ELINT Electronics Intelligence
EO ElectroOptical
EO/IR ElectroOptical/Infrared
FCS Flight Control System
FOM Figure of Merit
FOS Factor of Safety
GN&C Guidance, Navigation and Control
GPS Global Positioning System
HALE High Altitude Long Endurance
HAPP High Altitude Powered Platform
IFF Identify Friend of Foe
INS Inertial Navigation System
IR Infrared
LHS LeftHandSide
LLT Lifting Line Theory
LOS Line of Sight
LRN Low Reynolds Number
MAE Medium Altitude Endurance
MASINT Measurements and Signals Intelligence
MAV Micro Aerial Vehicle
MDO MultiDisciplinary Design Optimization
RHS RightHandSide
ROA Remotely Operated Aircraft
RPV Remotely Piloted Vehicle
SATCOM Satellite Communications
SAWE Society of Allied Weights Engineers
SBW StrutBraced Wing
SLS Sea Level Static Conditions
SWaP Size, Weight and Power
UAV Unmanned Aerial Vehicle
263
UCAS Uninhabited Combat Aircraft System
UCAV Uninhabited Combat Aerial Vechicle
UFP Unit Flyaway Price
UIUC University of Illinois at Urbana Champaign
VLM Vortex Lattice Method
264
Formatting Notes
The code architecture descriptions use capitalized and italicized letters to name
subroutines. The names presented may differ from the source code names when clarity is
gained from doing so. The functional architecture is accurate. When practical, minor
subroutines are not listed when there are significant differences in level.
The consolidated nomenclature definition list is defined initially within the document.
When a single variable name represents two or more variables, then the variable name is
used once and is followed by all definitions. Some variables with limited usage are
defined within the text of the document and not captured within the consolidated
nomenclature definition list. The units of variables are covered in the nomenclature list
to streamline the rest of the document.
265
Vita
John F. Gundlach IV has worked on several manned and unmanned aircraft research and
development projects. John has built over 50 radio controlled model aircraft starting at
the age of 13, with many original designs. At age 15, John helped build a wind tunnel
and static display model of a 5060 passenger turboprop design for a fledgling aircraft
company. At 17, John joined Aurora Flight Sciences, where he helped build the full
scale mockup of the Theseus highaltitude UAV. He also developed radiocontrolled
flying engineering models and conducted Multidisciplinary Design Optimization (MDO)
work for Aurora Flight Sciences on Jason, a Mars atmospheric flight vehicle. He
graduated Suma Cum Laude in undergraduate Aerospace Engineering at Virginia Tech in
1998. While an undergraduate, John received the 1998 Sigma Gamma Tau National
Aerospace Honor Society Ammon S. Andes National Award, won first prize in the AIAA
1998 MidAtlantic Regional Student Conference, was chairman of the Virginia Tech
AIAA, was the team leader of the 2
nd
place winner of the 1997 AIAA Design/Build/Fly
UAV Competition, published three AIAA papers, led the NASA Forestry UAV project,
and received several other honors and awards. During the undergraduate years, he
worked at Naval Research Laboratory on over a dozen UAV programs. John went on to
complete his Masters degree in Aerospace Engineering from Virginia Tech in 1999,
working on the StrutBraced Wing (SBW) MDO project. His SBW research resulted in
four published papers, including a Journal of Aircraft article. John works for the United
States Government today.
MultiDisciplinary Design Optimization of Subsonic FixedWing Unmanned Aerial Vehicles Projected Through 2025
John F. Gundlach
Abstract
Through this research, a robust aircraft design methodology is developed for analysis and optimization of the Air Vehicle (AV) segment of Unmanned Aerial Vehicle (UAV) systems. The analysis functionality of the AV design is integrated with a Genetic Algorithm (GA) to form an integrated Multidisciplinary Design Optimization (MDO) methodology for optimal AV design synthesis. This research fills the gap in integrated subsonic fixedwing UAV AV MDO methods. No known single methodology captures all of the phenomena of interest over the wide range of UAV families considered here. Key advancements include: 1) parametric Low Reynolds Number (LRN) airfoil aerodynamics formulation, 2) UAV systems mass properties definition, 3) wing structural weight methods, 4) selfoptimizing flight performance model, 5) automated geometry algorithms, and 6) optimizer integration. Multiple methods are provided for many disciplines to enable flexibility in functionality, level of detail, computational expediency, and accuracy. The AV design methods are calibrated against the HighAltitude LongEndurance (HALE) Global Hawk, MediumAltitude Endurance (MAE) Predator, and Tactical Shadow 200 classes, which exhibit significant variations in mission performance requirements and scale from one another. Technology impacts on the design of the three UAV classes are evaluated from a representative system technology year through 2025. Avionics, subsystems, aerodynamics, design, payloads, propulsion, and structures technology trends are assembled or derived from a variety of sources. The technology investigation serves the purposes of validating the effectiveness of the integrated AV design methods and to highlight design implications of technology insertion through future years. Flight performance, payload performance, and other attributes within a vehicle family are fixed such that the changes in the AV designs represent technology differences alone, and not requirements evolution. The optimizer seeks to minimize AV design gross weight for a given mission requirement and technology set. All three UAV families show significant design gross weight reductions as technology improves. The predicted design gross weight in 2025 for each class is: 1) 12.9% relative to the 1994 Global Hawk, 2) 6.26% relative to the 1994 Predator, and 3) 26.3% relative to the 2000 Shadow 200. The degree of technology improvement and ranking of contributing technologies differs among the vehicle families. The design gross weight is sensitive to technologies that directly affect the nonvarying weights for all cases, especially payload and avionics/subsystems technologies. Additionally, the propulsion
technology strongly affects the high performance Global Hawk and Predator families, which have high fuel mass fractions relative to the Tactical Shadow 200 family. The overall technology synergy experienced 1011 years after the initial technology year is 6.68% for Global Hawk, 7.09% for Predator, and 4.22% for the Shadow 200, which means that the technology trends interact favorably in all cases. The Global Hawk and Shadow 200 families exhibited niche behavior, where some vehicles attained higher aerodynamic performance while others attained lower structural mass fractions. The high aerodynamic performance Global Hawk vehicles had high aspect ratio wings with sweep, while the low structural mass fraction vehicles had straight, relatively low aspect ratios and smaller wing spans. The high aerodynamic performance Shadow 200 vehicles had relatively low wing loadings and large wing spans, while the lower structural mass fraction counterparts sought to minimize physical size.
iii
Acknowledgements
This research would not be possible without the support, advice, guidance, and other contributions from a number of individuals and organizations. My wife and best friend, Katie, provided invaluable encouragement, emotional support, practical advice, time, personal sacrifice, and patience. I will never forget her loving dedication. My advisory committee has shown great flexibility in accommodating an unconventional approach to PhD research. Dr. Schetz, my advisory committee chairman, showed great resourcefulness and creativity in making distance learning arrangements. Dr. Schetz’s frequent requests to limit research scope enabled this effort to be completed within a practical timeframe. Guidance and direction provided by the advisory committee has significantly shaped the research. I am grateful to Dr. Fred Lutze, who was a great mentor and guiding influence in my personal and educational development. I will cherish the memory of this great man. Rick Foch shared his extensive experience with me in this effort and during my college internships at NRL. My current employer generously provided tuition funding, which eliminated a financial burden. L.W. championed my quest for knowledge and understood the personal demands associated with pursuing an advanced degree while working full time. I would also like to thank R.L., S.O., M.C., P.A. and R.A. for their steadfast support throughout the duration.
iv
Table of Contents
Abstract ............................................................................................................................... ii Acknowledgements............................................................................................................ iv Table of Contents................................................................................................................ v List of Figures .................................................................................................................... ix List of Tables .................................................................................................................... xv Nomenclature.................................................................................................................. xvii Chapter 1 Introduction..................................................................................................... 1 1.1 Background ......................................................................................................... 1 1.2 Problem Statement .............................................................................................. 6 1.3 Approach............................................................................................................. 6 Chapter 2 Methodology ................................................................................................... 9 2.1 Methodology Overview ...................................................................................... 9 2.2 Methodology Development ................................................................................ 9 2.3 Applied Methodology Architecture .................................................................. 10 2.4 Methods............................................................................................................. 11 2.5 Calibration......................................................................................................... 12 Chapter 3 Optimization ................................................................................................. 13 3.1 Optimization Overview..................................................................................... 13 3.2 Optimization Code Architecture ....................................................................... 14 3.3 Design Variables............................................................................................... 16 3.4 Constraints ........................................................................................................ 17 Chapter 4 Technology Modeling................................................................................... 21 4.1 Technology Overview....................................................................................... 21 4.2 Technology Methods ........................................................................................ 23 4.3 Technology Modeling....................................................................................... 25 4.3.1 Aerodynamics ........................................................................................... 25 4.3.2 Avionics and Subystems ........................................................................... 30 4.3.3 Design ....................................................................................................... 35 4.3.4 Payloads .................................................................................................... 36 4.3.5 Propulsion ................................................................................................. 39 4.3.6 Structures .................................................................................................. 42 Chapter 5 Geometry....................................................................................................... 46 5.1 Geometry Overview.......................................................................................... 46 5.2 Geometry Definition ......................................................................................... 46 5.2.1 Wings and Tails ........................................................................................ 46 5.2.2 Fuselage .................................................................................................... 46 5.2.3 Propeller.................................................................................................... 48 5.3 Parametric Geometry ........................................................................................ 49 5.3.1 Parametric Fuselage .................................................................................. 49 5.3.2 Parametric Tails ........................................................................................ 52 5.4 Wing Lift Distribution Paneling ....................................................................... 53 5.5 Wing Structures Paneling ................................................................................. 54 5.6 Graphics File Generator.................................................................................... 57 v
.... and Power........................................................................................................7 Cooling Drag........................................................1 Subsystems.1 Propulsion Overview ............ 107 8.................1 PistonProp Engine Deck...................................................................................................3 Propeller Methods....... Weight........4................ 70 6...........................................3.............................1 Aerodynamics Overview ..............4.. 105 Chapter 8 Size... 97 7........................................................................................................................................3 Performance Methodology....... 62 6...................... Weight............................................ 116 8.........................................4................................ 111 8.... 62 6............ 100 7.................. 102 7....................3..................................................................................2 Profile Drag Overview.......................4....4 Drag Methodology ...............................................5 Generic Systems Thermal Management ................. 68 6............................3.................................................................3 Lift Methodology .............1 Overview..................3 Lift Distribution Methods ....... Weight..........................2 Avionics ...............................................2 Disk Actuator Model............ and Power Methods ...........4............. 104 7.............................. 128 9.................... 98 7........................... 131 9........................................................1 Turbojet/Turbofan Engine Decks .................................................................................. 93 6......................................... 97 7................................................................................................................................... 111 8......................................................3.............................................................................3 Structures ... 95 6...... 60 6............................................................................2 Performance Constraints...2............................................................................................................................................................................................................3..................................... 109 8. 104 7..1 Propeller Diameter Sizing............ 94 6........................3.............................2 Generic Systems.................................................................................................................. and Power Overview ..........4.....1 Drag Methodology Overview ..........................4 Generic Systems Power ...................3............3......1 Performance Equations ..3........ 110 8......4 Propulsion .............................4..................................1 Generic Systems Overview.............................................1 Lift Methodology Overview ................. 62 6.. 96 Chapter 7 Propulsion ......... 125 Chapter 9 Flight Performance............... 129 9.................2 Code Architecture .........6 Landing Gear Drag ................................ 108 8.2 Generic Systems Weights ......................................2............. 59 6......... 70 6.....................2.............. 107 8..............4 Pitching Moment........ 110 8................................5 Turbojets and Turbofans ........................................4 Fuselage Drag .................5 Wave Drag ....2.........4......................3 Specialized Size....................................................... 97 7.................................................................... 71 6..3....5.......1 Size................................................ 107 8..........................................................................2 CG Location Determination.......3 Interference Drag ........................ 91 6..........................................................................................................................3 Generic Systems Size............................. 89 6. 118 8.. 107 8............2...........Chapter 6 Aerodynamics ................................................................................... 99 7.........................................................................3..4.........2 Performance Code Architecture..............................................................................4 PistonProp Engines....... 63 6........3............................. 134 vi ............................... 131 9......2 Aerodynamics Code Architecture........................................................... 128 9...................................................................................................................... 59 6..........................................................................8 Miscellaneous Drag .....................
..2...................................................2 Calibration Cases Conclusions ..............................3.......................6 Technology Sensitivities.........2 Geometry.................................4 Weights ...3.......................................... 223 12....4 Weights .......1 Sizing Overview.1 Sizing Overview............................................................................................3 Propulsion .............. 198 12........................................... 226 12..............................4.......................................................5 Mission Evaluation ....................................................................... 198 12...............................................1 Overview............2...................................1 Calibration Overview......................... 198 12.....................................................................2..6 Technology Sensitivities.. 211 12.............................................................................. 227 12........... 177 11...................2...............4...............................................................3 Predator Technology Impacts ..........4.................3..........3.......3.....................................................................3 Propulsion ........... 211 12.3..............2 Predator .................. 198 12.................... 234 13..........................................................................................3 Flight Velocity Optimization ......2 Geometry................................................ 166 Chapter 11 Reference Year CalibrationOptimization Comparisons .....................................4 Climb Performance ...................................................................1 Overview............. 177 11...................3.9......... 177 11.................3 Propulsion .................... 239 vii ................................................3.......................3 Calibration Cases .................................................3 Shadow 200................................. 192 Chapter 12 Technology Case Optimization Applications ..................... 177 11............................................... 215 12........ 142 10..... 235 13... 202 12............................................................................. 216 12..........................................2......4............. 207 12......................... 185 11............................................................... 237 Appendix A Geometry Methods.................6 Technology Sensitivities.................6 Mission Completion Criterion ... 223 12...................................3............................................................... 232 Chapter 13 Conclusions........... 209 12...2 Reference Case Comparisons ...................................4 Recommendations...........................4............................................................................................................................................................ 203 12................................3........................ 230 12....................3................................................... 136 9. 139 9.3 Technology Cases Conclusions ........................ 220 12...........................................................................2...........................1 Global Hawk .................... 235 13...............................................3.............................5 Performance ...................................................................................................................................................................................................................................3 Shadow 200 2000 Optimum Design .........................4 Weights ...........5 Performance .......2 Predator 1994 Optimum Design ....................................................5 Performance .....................................2 Global Hawk Technology Impacts ....... 139 Chapter 10 Calibration.......................................................... 155 10....................................................................................................................................................................... 143 10...................2........ 135 9................................................ 198 12...........................................................2.................. 222 12.....................................................2 Calibration Methodology ..........................................................................................4....................................................................... 234 13.. 142 10........................1 Sizing Overview......................................................2 Geometry.........................1 AV Design Methodology Conclusions .....4 Shadow 200 Technology Impacts.................. 142 10................2..........3........................ 143 10.............. 211 12.........1 Global Hawk 1994 Optimum Design ..................................................... 223 12..............................
.................................... 264 Vita........................................... 265 viii ............................... 256 List of Acronyms ......................................................Appendix B Blade Element Propeller Methods.................................................................................................................................................................................................. 249 Bibliography ....................................................................................... 262 Formatting Notes ..................... 247 Appendix C Selected Avionics and Subsystem Weights Methods........................................................................................
................ AQM154 Compass Arrow................................................................... 23 Figure 42 Direct Technology Parameter Lookup Model..................... 34 Figure 414 Satellite Communications Dish Antenna Diameter Trends ................................................................................................................................................... Aquila) ................................................................................................. 7 Figure 21 ACMAIN Code Architectre .............. 31 Figure 49 GNC System Technology Trends.. 37 Figure 418 EO/IR Payloads.................................... 41 Figure 424 UAV Internal Combustion Engine........................................................................................................................................................................................................... 41 Figure 425 Internal Combustion Engine Technology Phase .................................................. Outrider.......................................................... ........ 39 Figure 421 UAV Jet Engine .............................................................................................................................. .................................................... Hunter.......... 43 Figure 428 Structural Technology Phase............................ 27 Figure 45 Wave Drag Technology Factor Trends .......List of Figures Figure 11 Model of German V1 Cruise Missile........................ 27 Figure 46 Profile Drag Multiplication Factor Trends........................ Compass Cope.................................................................. 37 Figure 417 SAR Payload Technology Trends.............. AQM34 Firefly)............... AQM147H Firebee......................... 42 Figure 426 Internal Combustion Engine Technology Trends ........... 14 Figure 41 Technology Parameter Ratio Model (Method 1) ........................................................ 1 Figure 12 AQM34 Firefly Currently in Operational Service........................... 31 Figure 410 INS Technology Phase............................................................. ........................................................ 40 Figure 423 Jet Engine Technology Trends........................ 28 Figure 47 Final Profile Drag Multiplication Factor Trends................ 44 Figure 429 Structural Weight Multiplication Factor Technology Trend ....................................................... and Global Hawk (Left to Right) ............................................................ 26 Figure 44 Interference Drag Trends ...... Predator...................................... 2 Figure 14 Models of Army Battlefield UAVs (Left to Right: Shadow 200.............. 45 ix ................................. 32 Figure 411 INS Technology Trends ............................................... 25 Figure 43 Induced Drag Multiplication Factor Technology Trends... 35 Figure 415 Weight Growth Technology..... 2 Figure 13 Models of the Teledyne Ryan Drones (Left to Right: Global Hawk..... 29 Figure 48 GNC System Technology Phase.................................... 42 Figure 427 Factor of Safety Technology Trends................................. 11 Figure 31 Genetic Algorithm Code Architecture ....................... 38 Figure 419 Large EO/IR Payload Performance Technology Trend .... 38 Figure 420 Small EO/IR Payload Performance Technology Trend.......................................................... 40 Figure 422 Jet Engine Technology Phase....................................................... 33 Figure 413 Subsystem Weight Technology Trend .......................... 36 Figure 416 Payload Technology Phase ............................................................ ............. 3 Figure 16 Predator .................................................................... 4 Figure 17 Shadow 200.................................... 4 Figure 18 Scale Comparison of Shadow 200............................................................................ 10 Figure 22 EVAL Code Architecture.............................................................................................................................................................................................................. 3 Figure 15 Global Hawk ......................................................... 32 Figure 412 Subsystems Technology Phase ........................................................................................................................................................................................................................
............................... 99 Figure 73 Comparison Between Predicted and Ideal Propeller Efficiencies................................................................................. 54 Figure 511 Vehicle Structural Paneling Example ........................................... 86 Figure 611 The Pressure Drag Comparison Between XFOIL and the Empirical Equation Matching for the NACA 2412 Airfoil ............................................................................................................................................. 135 Figure 94 Standard Mission Profile........................ 47 Figure 52 X45A Geometry.. 87 Figure 614 WingFuselage Interference Geometry ................Figure 51 Predator Fuselage Nose Region ........ 103 Figure 81 Macci Wing Factors ............................................... 98 Figure 72 Raymer Propeller Sizing Equation Results .................................................. 86 Figure 612 Comparison of Drag Polars Between the Direct Data and Parametric Airfoil Interpolation Methods............................ 53 Figure 510 Wing Aerodynamic Paneling for Virginia Tech Forestry UAV .................... 50 Figure 57 Radome Impacts on Fuselage Geometry.................. 77 Figure 68 Curve Fit for 360Degree Angle of Attack Sweep......... 92 Figure 616 Fuselage Drag Maximum Diameter Location Penalty......................................................................... 52 Figure 59 Virginia Tech Forestry UAV ................................................................................................................................................................... 134 Figure 93 Velocity Search Initialization....................................................................................................................................................... 145 x ............................... 72 Figure 64..... 48 Figure 53 Dragon Eye Fuselage Geometry ........................................................................ Global Hawk Engine Thrust Performance ........................................................................................................ 62 Figure 63 NACA 0009 at Several Reynolds Numbers.................................................... 102 Figure 74 Density Ratio Impacts on Brake Specific Fuel Consumption...................................................... 50 Figure 56 Wing Fairing Geometry ....................... 72 Figure 65 Airfoil 3D Interpolation Space ..................... 75 Figure 67 Angle of Attack Symmetry Diagram ............................ 92 Figure 71 Propulsion Code Architecture ...................................................................................................................................................................................................................................................................... 58 Figure 514 TiltWing Graphics Option ........................ 57 Figure 513 Example Unconventional Configuration.......................................................... Lift Coefficient Comparison Between XFOIL and Empirical Equation Matching for the NACA 2412 Airfoil.. ........................ 73 Figure 66 Drag Polar Interpolation Diagram for Two Airfoils . 58 Figure 61 AEROC code Architecture.............. 79 Figure 610 Upper and Lower Transition Location vs.............................................................................................. 87 Figure 613 Comparison Between XFOIL and Wind Tunnel Data............................................................. 144 Figure 102.............................................................. 90 Figure 615 Fuselage Drag Cross Section Penalty .................................................................................................................................................... 139 Figure 101 Global Hawk ........................................................................................................ 51 Figure 58 EO/IR Ball Geometry...................... 61 Figure 62 Pitching Moment Subroutine Architecture ........................... 57 Figure 512 Wing Tip Paneling Detail......................................... 77 Figure 69 Example Drag polar and Lift Curve of an Airfoil Through a Large Angle of Attack Sweep .......... 121 Figure 91 Performance Code Architecture....................... 49 Figure 55 Jet Integration on Generic Fuselage...................... Selig Donnovan SD7032 at Low Reynolds Numbers....................................................... 145 Figure 103 Global Hawk Engine Specific Fuel Consumption Performance......................... 129 Figure 92 Flight Envelope Boundaries.................................................................................. 48 Figure 54 Propeller Integration on Generic Fuselage........................................................
160 Figure 1020 Predator Weights Summary ...................... 147 Figure 106 Global Hawk Structural Weight Summary ....... 179 Figure 113 1994 Global Hawk Geometry Comparison – Isometric View ...................... 173 Figure 1035 Shadow 200 Trimmed Drag Contributors...................................................................................................... 182 Figure 119 Global Hawk MachAltitude Envelope Used .................................................................................................... 169 Figure 1031 Shadow 200 Structural Weight Summary...................................................................................................... 162 Figure 1022 Predator Trimmed Aerodynamic Performance ... 183 Figure 1110 Global Hawk Specific Fuel Consumption Comparison ............................................................................................................. 180 Figure 115 Global Hawk Summary Weights Comparison....................................................................................................... 151 Figure 108 Global Hawk Mass Fraction Summary .......................................................................... 168 Figure 1030 Shadow 200 Subsystems Weights Summary ......... 166 Figure 1028 Shadow 200 Installed Thrust Performance .. 157 Figure 1017 Predator Avionics Weights Summary.................................................. 174 Figure 1037 Shadow 200 Velocity Profile ....... 170 Figure 1032 Shadow 200 Weights Summary .......................................................................................... 175 Figure 111 1994 Global Hawk Geometry Comparison – Top View............... 163 Figure 1024 Predator Altitude Profile ........................................................................................................ 179 Figure 112 1994 Global Hawk Geometry Comparison – Side View .............................................. 151 Figure 109 Global Hawk Trimmed Incompressible Aerodynamic Performance....................................................... 156 Figure 1016 Predator Installed Specific Fuel Consumption Performance ......................Figure 104..................... 175 Figure 1038 Shadow 200 VelocityAltitude Envelope........................................................................................................................................................................................... 172 Figure 1033 Shadow 200 Mass Fraction Summary........... 167 Figure 1029 Shadow 200 Installed Specific Fuel Consumption Performance..................................................................... Global Hawk Avionics Weights Summary .................................................................................................... 184 xi ................................................................ 182 Figure 118 Global Hawk Mach Profile Comparison................................................ 162 Figure 1021 Predator Mass Fraction Summary......................................... 152 Figure 1011 Global Hawk Altitude Profile ..................................... 157 Figure 1018 Predator Subsystems Weights Summary ................................................................................................................................. 181 Figure 116 Global Hawk Summary Mass Fraction Comparison ........ 163 Figure 1023 Predator Trimmed Drag Contributors ...................................... 181 Figure 117 Global Hawk Altitude Profile Comparison...................... 180 Figure 114 Global Hawk Geometry Comparison......... 153 Figure 1012 Global Hawk Mach Profile .................................................. 153 Figure 1013 Global Hawk MachAltitude Envelope.................. 164 Figure 1026 Predator VelocityAltitude Envelope.......... 165 Figure 1027 Shadow 200. 172 Figure 1034 Shadow 200 Trimmed Aerodynamic Performance... 146 Figure 105 Global Hawk Subsystems Weights Summary...................................................................... 159 Figure 1019 Predator Structural Weight Summary ...... 149 Figure 107 Global Hawk Weights Summary ..... 164 Figure 1025 Predator Velocity Profile................................................................................................... 173 Figure 1036 Shadow 200 Altitude Profile.......... 152 Figure 1010 Global Hawk Trimmed Drag Contributors .................................................................................................................................... 154 Figure 1014 Predator ...................................... 183 Figure 1111 Global Hawk Lift to Drag Comparison...... 155 Figure 1015 Predator Installed Thrust Performance.............................................
.................. 2010......................................................................................................... 206 Figure 1218 Global Hawk Structures Weight Trends ................ 208 Figure 1220 Global Hawk Mach Profile Trends .................................................................................... 201 Figure 126 Global Hawk Wing Sweep Trends.................. 201 Figure 127 Global Hawk Thickness to Chord Ratio Trends ....................... 193 Figure 1126 Shadow 200 Geometry Comparison ..... 203 Figure 1212 Global Hawk Weight Trends .................................................................................................................................. 193 Figure 1124 2000 Shadow 200 Geometry Comparison – Side View..... 2000...... 186 Figure 1114 1994 Predator Geometry Comparison – Isometric View...... 196 Figure 1132 Shadow 200 Specific Fuel Consumption Comparison....................Obligated Weights Trends.............. 186 Figure 1113 1994 Predator Geometry Comparison – Side View...................... 2015..... 202 Figure 129 Global Hawk Fuselage Length Trends...................................... Years (Left to Right): 1994............................................................................ 2000................................................................................... 2015...................... 187 Figure 1115 Predator Geometry Comparison.......... 199 Figure 123 Global Hawk Geometry Trends – Isometric View....................................... 207 Figure 1219 Global Hawk Altitude Profile Trends ............................... 200 Figure 125 Global Hawk Aspect Ratio Trends ............................................................................................ 208 Figure 1221 Global Hawk Mission Lift to Drag Ratio Trends.................................. 197 Figure 121 Global Hawk Geometry Trends – Top View..... 208 xii .................................................................................................... 2010........................................................................ 188 Figure 1118 Predator Altitude Profile Comparison........................ 2005.......................... 187 Figure 1116 Predator Weights Summary Comparison.............. 201 Figure 128 Global Hawk Wing Parameter Comparisons ...... 190 Figure 1121 Predator Specific Fuel Consumption Comparison..................................... 195 Figure 1130 Shadow 200 Velocity Profile Comparison............................ 206 Figure 1217 Global Hawk Payload Weight Trends...... 189 Figure 1120 Predator VelocityAltitude Envelope Used.......... 205 Figure 1216 Global Hawk Subsystems Weight Trends..................................... 204 Figure 1213 Global Hawk Mass Fraction Trends ................................. 2025... 2010................... 191 Figure 1123 2000 Shadow 200 Geometry Comparison – Top View ..........Figure 1112 1994 Predator Geometry Comparison – Top View.......... 200 Figure 124 Global Hawk Wing Span Trends .. 2020......................................................................... 2025........................................................... 196 Figure 1131 Shadow 200 VelocityAltitude Envelope Used .... 2005........ Years (Left to Right): 1994......................... 203 Figure 1211 Global Hawk Thrust to Weight Ratio Trends ............................... 188 Figure 1117 Predator Summary Mass Fraction Comparison ...................................................... 199 Figure 122 Global Hawk Geometry Trends – Side View............... 204 Figure 1215 Global Hawk Avionics Weight Trends ................................................................................................................... 2005..................................................... 2000................................ 2015......... 2020........ 197 Figure 1133 Shadow 200 Lift to Drag Ratio Comparison............ 190 Figure 1122 Predator Lift to Drag Comparison......... Years (Left to Right): 1994............................................................ 194 Figure 1128 Shadow 200 Summary Mass Fraction Comparison ................... 195 Figure 1129 Shadow 200 Altitude Profile Comparison ............... 202 Figure 1210 Global Hawk Uninstalled Thrust Trends ........ 2020........... 204 Figure 1214 Design Gross ......... 194 Figure 1127 Shadow 200 Summary Weights Comparison ...................................... 2025..................................................... 189 Figure 1119 Predator Velocity Profile Comparison ... 193 Figure 1125 2000 Shadow 200 Geometry Comparison – Isometric View......
.............. 2010.......Figure 1222 Global Hawk Mission Fuel Consumption Trends......................... 219 Figure 1240 Predator Structures Weight Trends ........... 2025..................... Years (Left to Right): 2000.. 222 Figure 1246 Shadow 200 Geometry Trends – Top View................................................. 2025............. 217 Figure 1236 Predator Design Gross – Obligated Weight Trends........................ 214 Figure 1231 Predator Fuselage Length Trends.............................................................. 215 Figure 1232 Predator Uninstalled Power Trends ............................................................... 223 Figure 1248 Shadow 200 Geometry Trends – Isometric View................................................................... 2000.......... 2005....................................................... 2020...................................... 2005... 2020... Years (Left to Right): 2000........................................................................ 214 Figure 1230 Predator Wing Area.. 2010. 220 Figure 1242 Predator Velocity Profile Trends .......................................... 229 Figure 1260 Shadow 200 Payload Weight Trends....................................... 2025................ 225 Figure 1252 Shadow 200 Wing Loading – Aspect Ratio Relationships . 2015....................... Years (Left to Right): 1994................. 2025.. Years (Left to Right): 1994........ 230 Figure 1261 Shadow 200 Structures Weight Trends................................................................ 2020.. 225 Figure 1251 Shadow 200 Fuselage Length Trends ...... 2020.............. 216 Figure 1234 Predator Weight Summary Trends............. 218 Figure 1238 Predator Subsystems Weight Trends ............................................................ 2025............................................ 229 Figure 1259 Shadow 200 Subsystems Weight Trends .... 213 Figure 1228 Predator Aspect Ratio Trends ........ 2015......................................................................... 218 Figure 1239 Predator Payload Weight Trends ...................................... 227 Figure 1256 Shadow 200 Summary Mass Fraction Trends ...................... 2015........................................ 227 Figure 1255 Shadow 200 Weight Trends ....... ..... 2010..... 215 Figure 1233 Predator Power to Weight Ratio .......................................................................................... 219 Figure 1241 Predator Altitude Profile Trends .................................................................................... 226 Figure 1253 Shadow 200 Uninstalled Power Trends ............................................................................ 2005.......... 224 Figure 1250 Shadow 200 Aspect Ratio Trends ........................................... 228 Figure 1258 Shadow 200 Avionics Weight Trends................................... 226 Figure 1254 Shadow 200 Power to Weight Ratio Trends ....................... 212 Figure 1225 Predator Geometry Trends – Side View.................. 217 Figure 1237 Predator Avionics Weight Trends........ 2015............ 2000................... 221 Figure 1243 Predator Mission Lift to Drag Ratio Trends ....... 209 Figure 1223 Global Hawk Design Gross Weight Sensitivity to Technology............................... 2010............................................................................................ 230 xiii ......... 213 Figure 1227 Predator Wing Span Trends ................................................................................ 216 Figure 1235 Predator Summary Mass Fraction Trends........................ Years (Left to Right): 2000................................................. 222 Figure 1245 Predator Design Gross Weight Sensitivity to Technology ....... 2005................................... 223 Figure 1247 Shadow 200 Geometry Trends – Side View........................................ 2010................. 2010. 214 Figure 1229 Predator Thickness to Chord Ratio ........... 221 Figure 1244 Predator Mission Fuel Consumption Trends..... 2015................ 2020....................................... 2005........... 2015..................... 212 Figure 1226 Predator Geometry Trends – Isometric View............................. Years (Left to Right): 1994. 228 Figure 1257 Shadow 200 Design Gross – Obligated Weights Trend........................................................................................ 210 Figure 1224 Predator Geometry Trends – Top View........................................................................ 2000........ 2020................................ 2025....... 2005................. 224 Figure 1249 Shadow 200 Wing Span Trends..........
...... 233 Figure 131 General Atomics GnatPredator Family Evolution ……………………….......................... 231 Figure 1264 Shadow 200 Mission Lift to Drag Ratio Trends .237 Figure A1 Wing Geometry Parameters ...............................................................Figure 1262 Shadow 200 Altitude Profile Trends.........................................248 xiv ..................... 231 Figure 1263 Shadow 200 Velocity Profile Trends .................................. 232 Figure 1266 Shadow 200 Design Gross Weight Sensitivity to Technology .................................................................. 232 Figure 1265 Shadow 200 Mission Fuel Consumption Trends ................ 239 Figure B1 Comparison Between Two and Three Bladed Propeller Designs…………....
....... 115 Table 89 Fuel System Weight Method Functional Dependencies ......................................................................... ...........................................................................List of Tables Table 31 Universal Design Variables...................................................... 107 Table 82 Flight Control Systems Weight Methods ................................................ 165 Table 107 Shadow 200 Geometry Characteristics ................................................................................................................................................................................... 112 Table 84 Environmental Control System Weight Methods....................................................... 39 Table 49 Structures Technology Attributes........... 25 Table 44 Airfoil Characteristics.................................................... 113 Table 86 Electrical System Weight Methods ................................................................................................................. 144 Table 102 Global Hawk Propulsion ................................................................................ 117 Table 811 Wing Weight Methods ................ 118 Table 813 Tail Weight Methods............................. 63 Table 63 Profile Drag Methods ............... 123 Table 817 Fuselage Weight Method Functional Dependencies ...................................................................... 118 Table 812 Wing Weight Method Functional Dependencies ....................................................................................................................................................................... 18 Table 41 Technology Attributes ................................................... 17 Table 34 Constraints....................................................................... 124 Table 101 Global Hawk Geometry............................................... 115 Table 810 Air Data System Factors.. 43 Table 61 Primary aerodynamics product list..... 81 Table 66 Empirical equation airfoil database........ 123 Table 815 Vertical Tail Weight Method Functional Dependencies ...................................................... 22 Table 42 Attribute Ranking Description .......................................... 114 Table 88 Fuel System Weights Methods.................................................. 82 Table 81 Component Categories and Types................... 156 Table 106 Predator Design Technology Levels .......................... ..................................................................... 16 Table 32 Jet Propulsion Design Variables..................................................................................................................... 144 Table 103 Global Hawk Design Technology Levels................................... 123 Table 816 Fuselage Weight Methods .......... ....................................................................................................... 28 Table 45 Avionics and Subsystems Technology Attributes................................................................... 114 Table 87 Electrical System Weight Method Functional Dependencies ................................................................................................ 18 Table 35 Mission constraint types ........................................................................... 73 Table 64 Low Reynolds number airfoil database.................. 123 Table 814 Horizontal Tail Weight Method Functional Dependencies................ 154 Table 104 Predator Geometry Characteristics........... 80 Table 65 High Reynolds number airfoil database......................................................................................... 36 Table 48 Propulsion Technology Attributes................................ 17 Table 33 Reciprocating Engine Propulsion Design Variables ................................................................................................................................................................ 113 Table 85 ECS System Weight Method Functional Dependencies ...... 166 xv ...................................... 30 Table 46 Design Technology Attributes........................................................ 60 Table 62 Lift Distribution Methods.................... 22 Table 43 Aerodynamics Technology Attributes....................................................................... 111 Table 83 Flight Control System Weight Method Functional Dependencies...................................................................................................... 35 Table 47 Payload Technology Attributes ........................... 155 Table 105 Predator Propulsion Characteristics .............................................................
...................................... 178 Table 112 Predator Design Variables............................... 250 Table C2 INS Factors ................................................................................. 172 Table 1010 Shadow 200 Design Technology Levels ................................................. 167 Table 109 Shadow 200 Calibration weight comparison....................................................... 252 xvi ...... 185 Table 113 Shadow 200 Design Variables........................................................................................................................ 192 Table C1 Autopilot Factors .........Table 108 Shadow 200 Propulsion Characteristics .................................................................................................................................................................................................................................................... 251 Table C3 Camera Factors .. 176 Table 111 Global Hawk Design Variables .............
press CD.cavity CD. lb/HP*hr Brake specific fuel consumption at altitude. lb/HP*hr Brake specific fuel consumption at sea level. ft Maximum prescribed altitude.cool CD.frict CD. lb/HP*hr Wing span. HP Brake Horsepower at sea level. ft Local chord of interest (interference drag).wall CDint.strut CD. 1/deg or 1/radian Influence coefficient in lift distribution methods Brake Horsepower.wheel Axial interference factor Thrust ratio Rotational interference factor Influence coefficient in lift distribution methods Normalized distance from wing AC to trailing edge Altitude.sweep CDint. 1/deg or 1/radian Wing lift curve slope.gear Cdi(i) CDi CDint CDint.door Cd. Watts Section drag coefficient 3D drag coefficient 2D drag coefficient Landing gear bay cavity drag coefficient Cooling drag coefficient Landing gear door drag coefficient 2D friction coefficient Landing gear drag coefficient Local induced drag coefficient Induced drag coefficient Interference drag coefficient Interference drag coefficient due to perpendicular wall Interference drag coefficient increment due to lift Interference drag coefficient increment due to sweep Interference drag coefficient increment due to inclination 2D pressure drag coefficient Landing gear strut drag coefficient Landing gear wheel drag coefficient xvii .Nomenclature a a' A(i. ft Mean aerodynamic chord length. ft Cooling power. 1/deg or 1/radian Tail lift curve slope.lift CDint. ft Local chord (paneling).j) ACFunc alt altmax AR ARw asurface atail awing b(i) BHP BHP0 BHPh bref BSFC BSFC0 BSFCh bw c c(i) c (bar) C Cd CD Cd CD. HP Brake Horsepower at altitude. ft Blade section chord.incline Cd. ft Aspect ratio Wing aspect ratio Surface lift curve slope. ft Brake specific fuel consumption. HP Reference wing span.
2=elevator. ft Drag. hrs Prandtl momentum loss factor Interpolation factor Penalty factor Brake horsepower altitude correction factor Brake specific fuel consumption altitude correction factor Design level factor Airfoil form factor Fuselage form factor Flap function designator.w(j) CTV cwing Cx Cy D DCool DDish DJet DPow DRam DTire e Eseg Esubseg F fact(i) factpen(i) FBHP FBSFC FDesign FF FFFuse flap(j) Local flap chord.Max CLstall Clstall Cl0 CM CM. ft Power Extraction Drag. ft Jet engine diameter. 3=aileron xviii .0L(i) CP CPV crcr.fuselage CM. lbs Cooling Drag. 1=flap. lbs Communications dish diameter. hrs Subsegment endurance.0L Cm. lbs Tire width.cflap(i) Cl Cl(i) CL CL.Flap CM.Nac CM. ft Span efficiency factor Segment endurance.CL CM. ft Section lift coefficient Local 2D lift coefficient (paneling) Total 3D lift coefficient per surface or total vehicle 3D maximum lift coefficient 3D stall lift coefficient 2D stall lift coefficient 2D minimum drag lift coefficient Total pitching moment coefficient Pitching moment coefficient due to lift Pitching moment coefficient due to flap deflections Fuselage pitching moment coefficient Pitching moment coefficient due to fuselages and nacelles Pitching moment coefficient due to airfoil sections at zero lift Pitching moment coefficient of local airfoil Coefficient of power (conventional.w CT ctcr.normalized by rotation terms) Coefficient of power (normalized by V terms) Segment root chord to wing root chord ratio Propeller radiusnormalized chord length Wing root chord. ft Coefficient of thrust (conventional.w(j) c/Rp(j) cr.Fuse. lbs Ram drag.normalized by rotation terms) Segment tip chord to wing root chord ratio Coefficient of thrust (normalized by V terms) Wing chord Torque force coefficient Thrust force coefficient Propeller diameter.
ft Engine length. 1=split.Install FWt.Install G HEng hn incw Ixx Iyy. lbm*ft2 Advance ratio Goldstein momentum loss factor Lift dependent profile drag constant Aspect ratio correction factor for downwash derivative Taper ratio correction factor for downwash derivative Tailplane height correction factor for downwash derivative Length from the wing AC to the tail root trailing edge. ft Critical Mach number Drag divergent mach number Fuel mass fraction Maximum Mach number Index number in Prandtl LLT Curve fit parameter in parametric airfoil method Propeller revolutions per second. ft Vertical tail moment arm. ft4 YY mass moment of inertia. ft Electrical wiring length. ft Length of jet engine. ft Horizontal tail moment arm. ft Aircraft lift to drag ratio Fuselage length to maximum diameter ratio Length of landing gear door. 1/sec Maximum value of index n in Prandtl LLT Number of crew members in aircraft Blade count Number of flight control systems Number of generators Limit load factor Number of struts xix . ft Average tail moment arm.mass J K K ka ktr kh LACVT. ft Lifting load of a wing.FTech ftype(j) FVol. degrees XX area moment of inertia. lbs Overall vehicle length.TE L/D L/Dfuse LDoor LElec LEng Lfuse LH LJet Loadw LOverall LTail LV m(dot)cool M macW Mcrit MDD MFFuel Mmax n N NCrew Nblade NFCS NGen NLLF Nstrut Final technology factor Flap type. 4=double slotted Installation volume factor Weight installation factor Circulation function Engine height. ft Fuselage length. 3=slotted flap. ft Cooling air mass flow rate Flight Mach number Wing mean aerodynamic chord. ft Neutral point normalized by the MAC Wing incidence. 2=plain.
w Pw P/W q Qcool qmax Qseg qShear r R Re/L ROC ROCmin Rp r/Rp(j) Rseg Rsubseg RYear s Sc or SCont SF sm Sref sseg stip SVTail Sw Swet. ft2 Thrust. nmi Average technology ratio per year between two reference dates Semispan. Watts Absorbed power. nmi Reynolds number per unit length. Watts Maximum dynamic pressure. lb/ft2 Segment interference drag factor Shear flow Radial distance. Watts Wing rotation 3D point (ft. ft.NSurf NTanks Ntire/strut NUlt P Pabs PDish PEng PMax Prot Pr.w PShaft Pt. ft. nmi Subsegment range. ft2 Fuselage wetted area. ft2 Wing area.Fuse T Tair Number of control surfaces Number of fuel tanks Number of tires per strut Ultimate load factor Power absorbed by propeller. HP Maximum power draw of electrical system. HP Wing segment tip point (ft. HP Power used by dish antenna. ft Propeller radiusnormalized radial distance Segment range. ft) Uninstalled sea level horsepower to design gross weight. ft) Wing root 3D point (ft. Watts Engine power. ft Range. ft Vtail area. HP/lb Dynamic pressure. ft. ft/min Minimum rate of climb. HP Component power. ft. ft) Wing segment root point (ft. ft2 Wing shape factor Scale factor Static margin Reference area (defined as planform area of wing 1). L Control surface area. ft Wing tip semispan distance. ft) Shaft horsepower. L Distance in 3D parameter space for airfoil interpolation Propeller tip radius. ft2 Wing panel span increment. ft/min Propeller radius. lb/ft2 Cooled power. ft1 Rate of climb. lbs Air temperature xx .
lbs Panel thickness. kts Installation volume. ft3 Local total velocity.tBend t/c Tcomp Tech Tech1 Tech2 Temp TempSL throt TJet Tmax Tmin Tmax. kts or ft/sec Total flight velocity at infinity. kts or ft/sec Freestream velocity. lbs Total avionics weight. ft3 Local velocity in the spanwise (Y) direction. lbs Maximum installed thrust in flight. kts or ft/sec Flight velocity.Max VInstall vm VolAvail VolDish VolEng VolInternal VolJet VolMargin VolTank VolTot VUninstalled W WADS WAntiIce WAttach WATC WAvion Bending material thickness. sea level static thrust specific fuel consumption. ft Tail volume coefficient Uninstalled SLS thrust to design gross weight ratio Local velocity in the streamwise (X) direction. lbs Air data system weight. lbm/lbhr Installed thrust specific fuel consumption ratio Ref. ft3 Uninstalled volume. lbs Air traffic control system weight. kts or ft/sec Weight. lbs xxi . kts or ft/sec Maximum equivalent velocity. lbs Wing attachment weight. lbm/lbhr Skin thickness of shear material. kts Vortex displacement velocity. lbm/lbhr Installed sea level static thrust specific fuel consumption. ft3 Volume occupied by jet engine. ft3 Engine volume. lbs Maximum installed thrust at sea level static conditions. ft3 Total fuel volume required. ft Required thrust. ft3 Total fuel tank volume. lbs Antiice system weight. ft3 Internal fuel tank volume. lbs Minimum installed thrust in flight. ft Thickness to chord ratio Component temperature Technology parameter Technology parameter at data point 1 Technology parameter at data point 2 Absolute static temperature at altitude Absolute static temperature at sea level Throttle setting Jet static thrust rating. lbs Installed inflight thrust specific fuel consumption. kts or ft/sec Volume available.SLS TPan Treq TSFC TSFCrat TSFCref TSFCSLS tShear TVC T/W um U∞ V v' Veq. ft3 Volume margin. ft3 Radome volume occupied by dish antenna.
lbs Landing gear weight. lbs Installation weight. lbs EO/IR payload weight. lbs Engine width.w(j) x/cε. ft Segment flap chord ratio Segment reference x/c for wing sweep Segment reference x/c for washout X coordinate of the center of gravity. lbs Final weight. lbs Weight of wing box. lbs Width of tire. ft Inertial Navigation System (INS) weight. lbs Weight of wing ribs. lbs Weight of fuel in the wing. ft xxii . lbs Global Positioning System (GPS) weight. ft Uninstalled weight. lbs Wing weight. ft Local X coordinate of the panel quarter chord.wing wfuse(i) WFuse WGear WGPS Wi WidEng WINS WJet wm WNac WOverall WProcessor WRecip WRecorder WRibs WSAR Wseg WTails WtInstall Wtire WtUninstalled WWing x x(i) XAC. lbs Overall vehicle width (usually span). lbs Fuel burned. lbs Overall tail weight.WBend WBox WCamera WComms WCS. lbs Design gross weight. lbs Jet engine weight. ft Fuselage weight. lbs Fuel system weight. lbs Electrical system weight. lbs Local velocity in the vertical (Z) direction. kts or ft/sec Nacelle weight. lbs Camera weight. lbs SAR payload weight. lbs Data recorder weight. lbs Weight of control surfaces and secondary structure.Secondary WDG WDish WDoor WECS WElec WEO/IR Wf Wfuel Wfuelb WFuels WFuel. lbs Communications system weight.wing x/cflap(j) x/cΛ. lbs Initial weight. lbs Mean aircraft weight during segment. lbs Width of landing gear door. lbs Local fuselage width. ft Processor weight. ft Environmental Control System (ECS) weight.w(j) xcg xc/4(i) Bending material thickness. lbs Total fuel weight. lbs Nondimensional distance Ω*r/V Intermediate coefficient in Prandtl LLT Wing aerodynamic center location Xcoordinate. lbs Reciprocating engine weight. lbs Parabolic dish weight.
f. v'/V Flap effectiveness factor based on flap type Gear efficiency Propeller efficiency Ideal propeller efficiency xxiii . Zrot.flap ∆X ∆y(i) ε ε(i) εw(j) ζ ηf ηgear ηP ηP. degrees or radians Local downwash angle. degrees or radians Local washout.ideal Propeller x/c reference for sweep Propeller x/c reference for twist Fuselage leading edge x. δ(i) δf(i) δflap. ft Wing root leading edge x. ft Propeller rotation point x. ft Drag to lift ratio Downwash angle. y. y.f.p(j) x/cβ.w(j) ∆Cl. ft Zcoordinate of tail aerodynamic center. Yrot. ft Vehicle angle of attack. Yf. Zf X/LAC. z coordinates. y. ft Local spanwise panel difference. degrees Change in local 2D lift coefficient due to flap deflections Difference in X direction. degrees Displacement velocity ratio.wing x/Lf(j) XNP Xp. ft Y coordinate. Zp Xrot. z coordinates. z coordinates location. degrees or radians Segment washout relative to wing root.f(j) z/Lb. ft Segment flap span ratio Year Year at data point 1 Year at data point 2 Reference Year Fuselage lengthnormalized side ydistance from reference Fuselage lengthnormalized mid zdistance from reference Fuselage lengthnormalized top zdistance from reference Fuselage lengthnormalized bottom zdistance from reference Zcoordinate of wing aerodynamic center. ft Wing rotation point x.p. z coordinates. ft Fuselage x/L location of wing aerodynamic center Fuselage lengthnormalized xdistance from reference Neutral point Xcoordinate. y.p(j) Xf. z coordinates. Yrot.f(j) Zwing Ztail α αtwist(i) α0L(i) β ΓVTail Γw(j) δ. degrees VTail dihedral angle. degrees or radians Local zero lift angle of attack. Yw.f Xrot. degrees or radians Flap maximum deflection. Zw y y/bflap(j) Year Year1 Year2 YearRef y/Lf(j) z/Lm. Zrot Xrot. y. z coordinates.p. Yrot. degrees Segment dihedral.x/cΛ. degrees Angle of the VLM CP at zero angle of attack. degrees or radians Blade twist angle. y. Zrot. ft Fuselage rotation point x. ft Propeller center x.f(j) z/Lt.p Xw. degrees or radians Flap deflection. Yp.
degrees or radians Flow angle at the propeller tip Propeller angular velocity. sl/ft3 or kg/m3 Top conic parameter Local solidity Air density ratio Local flap effectiveness factor based on flap type Ultimate shear stress Flow angle Local dihedral angle.ytip/(bw/2) Normalized semispan angle for Prandtl LLT Inclination angle between wing and fuselage side at intersection. sl/ft3 Bottom conic parameter Density at altitude. degrees Viscosity Nondimensional propeller radius Nondimensional propeller hub radius 3. radians/sec Common Subscripts 0 avg eq fact Zero lift condition Sea level Average Equivalent Factor . degrees Segment sweep at x chordwise location.w(j) µ ξ ξ0 π ρ ρb(j) ρh ρ0 ρ t(j) σ τ(i) τult φ φ(i) φt Ω Nonideal propeller efficiency Radiation Efficiency Efficiency of radiating power Spanwise location of segment tip .f θrot.w(j) θ(i) θincline θrot. sl/ft3 or kg/m3 Density at sea level.p θrot.w κa λ λeq Λc/2 Λc/4 Λp(j) Λx. degrees or radians Fuselage rotation angle. degrees Transonic airfoil technology factor Speed ratio Taper ratio Equivalent taper ratio Sweep at 50% chord.nonideal ηRad ηradiation ηtip.ηP.14159… Air density. degrees or radians Propeller sweep angle.multiplication xxiv . ft Wing rotation angle. degrees Propeller rotation angle. degrees or radians Sweep at quarter chord.
or vehicle stall Table Target case Type xxv .neg mindat mindrag press prof ref seg stall table target type Friction drag Index Maximum negative angle of attack (most negative) Point of minimum available data Minimum drag condition Pressure drag Profile drag Reference case Segment Section.frict i max. wing.
The combined AirLaunched Cruise Missile (ALCM) and Surface Launched Cruise Missile (SLCM) programs of the 1970’s produced the first practical cruise missile programs [Werrell 1985]. and Micro Aerial Vehicle (MAV). Over 5. McDaid [1997]. . unmanned aircraft must be capable of controlled and sustained flight. Germany developed and employed numerous V1 cruise missiles with unprecedented results [Werrell 1985]. Unmanned aircraft have been dubbed drone. unmanned aircraft systems have an airborne component that performs at least one mission role without a pilot onboard. Wagner [1982 and 1992].Chapter 1 Introduction 1. Additionally.S. Cruise missiles and UAVs share a highly common lineage [Werrell 1985]. Although representing the state of the art for unmanned aircraft. to name only a few sources. Whatever called. the acronym UAV is adopted throughout this document. A model of the V1 is shown in Figure 11. The first operational experiments with cruise missiles began in World War I [Werrell 1985. Remotely Piloted Vehicle (RPV).000 V1s were launched by the Germans [Werrell 1985]. UAV systems was made possible by the Teledyne Ryan (now Northrop Grumman) Firebee/Firefly drones [Wagner 1982 and 1992]. UAV. McDaid 1997]. For consistency. Goebel [2003].1 Background Unmanned Aerial Vehicle (UAV) systems have taken many names and forms over their long history. and the distinctions between these types of systems are minor. the V1 is generally considered a cruise missile. Irwin [2000]. The history of unmanned aircraft is described by Werrell [1985]. Uninhabited Combat Aircraft System (UCAS). UAV systems began through cruise missile and target drone developments. to list a few. Organic Aerial Vehicle (OAV). shown in Figure 12. Holder [2001]. otherwise the first cast stone is also the first unmanned aircraft. During World War II. but these attempts met with little success. Uninhabited Combat Aerial Vehicle (UCAV). and Munson [1988 and 1999]. The United States developed many cruise missiles from the mid1940’s through the late 1960’s with generally disappointing results [Werrell 1985]. was initially developed in 1950. as appropriate. Figure 11 Model of German V1 Cruise Missile The first golden era of U. The first version of what became the Firebee drone.
Teledyne Ryan successfully developed two high altitude drones during the 1970’s. which cost over $1 billion to develop. Figure 12 AQM34 Firefly Currently in Operational Service The U. A model of Aquila can be seen on the far right of Figure 14.and variants are still in service today [Wagner 1982 and 1992]. Munson 1988. the AQM147 Compass Arrow and the Compass Cope. AQM154 Compass Arrow. which conducted over 3. Figure 13 Models of the Teledyne Ryan Drones (Left to Right: Global Hawk. UAV capabilities. Compass Cope.S. The most significant U. that were not fully utilized or pursued due to the changing political climate of the early postVietnam era [Wagner 1982 and 1992]. Goebel 1993].400 operational reconnaissance missions over Vietnam and other regions [Wagner 1982 and 1992]. The Firefly system evolved into the Firebee. The cancellation of the Aquila program represented the nadir of U. McDaid 1997]. Over 20 distinct variants of this system have been employed [Wagner 1982 and 1992]. saw significant schedule slips. UAV program of the 1980’s was the Lockheed Aquila battlefield surveillance RPV. AQM34 Firefly) 2 .S. and was ultimately never fielded due to poor performance [Fahlstrom 1998. experienced a notable decline in UAV funding and development success during the 1970’s and 1980’s [Wagner 1982 and 1992. Scale models of the Teledyne Ryan developments are shown in Figure 13.S. AQM147H Firebee.
these two systems are the backbone of the U. Today.Figure 14 Models of Army Battlefield UAVs (Left to Right: Shadow 200. A rendering of Global Hawk is shown in Figure 15. Both systems are equipped with satellite communications systems to provide Beyond Line Of Sight (BLOS) connectivity with remote ground stations [OSD 2002].S. is a HighAltitude LongEndurance (HALE) UAV that performs reconnaissance missions primarily with ElectroOptical/Infrared (EO/IR) and Synthetic Aperture Radar (SAR) payloads [Drezner 2002. UAV arsenal. The Defense Airborne Reconnaissance Office (DARO) and the Defense Advanced Research Projects Agency (DARPA) started. developed by Teledyne Ryan. Predator and Global Hawk went into production and have seen extensive operational service [OSD 2002]. OSD 2002]. but is more affordable [OSD 2002]. The MediumAltitude Endurance (MAE) Predator UAV. developed by General Atomics. flies at a lower altitude and has reduced payload capacity than Global Hawk. The Tier II+ Global Hawk. among other UAV efforts. A rendering of Predator is shown in Figure 16. Outrider. UAV systems are now in a second golden era. two Advanced Concept Technology Demonstrator (ACTD) programs in 1994. Figure 15 Global Hawk 3 . Hunter.S. Aquila) U.
Pilots have limitations such as the need for food. The second golden era is enabled by lightweight digital electronics. rest. Figure 17 shows a rendering of the Shadow 200 system. Unmanned aircraft have filled roles for “dull. The first golden era was made possible by analog autopilot advances. others have survived. and early small jet engine technology. and satellite communications. Dirty constitutes missions that would endanger human pilots due to exposure to radiation or biological/chemical agents. 4 . The obvious distinction from a manned aircraft is the lack of an onboard pilot. improved sensor technology. lightweight inertial navigation system accuracy enhancements. Technology has played a key role in UAV success. The success of the Global Hawk and Predator demonstrated the maturity of highly effective UAV systems. Dull typically means long reconnaissance and surveillance missions that stress human endurance. which helped lead the way for new UAV system developments. dirty. and the intrinsic value placed on human life. This effort seeks to explore the implications of potential future technology advances. water. Figure 17 Shadow 200 UAVs fill many missions and roles. Dangerous encompasses missions that would place human life at risk. One example of a surviving UAV system still under development is the tactical AAI Shadow 200 [Palumbo 2000]. though cost savings are not always realized. and dangerous” [OSD 2002] missions. While some recent UAV developments experienced failures.Figure 16 Predator Numerous other UAV programs are either in development or operational today [OSD 2002]. UAVs also offer the potential for significant cost savings over their manned counterparts.
Hall [1984] describes a method for sizing the structures of this platform. provide comprehensive guides to UAV needs and technologies projected into the future. the design methods employed are rarely published with sufficient detail to provide a complete methodology that can be transferred to other designs. so the ultimate validity of the methods is unknown. manufacturing. Boeing provided two noteworthy papers on the design of the Boeing Compass Cope [Brown 1980] and Condor [Johnstone 1990]. However. Unfortunately. Other reports [Carmichael 1996. Nearly every UAV system developed underwent an aircraft design effort. technology. though with some limitations due to proprietary considerations. Sullivan 1998. The Solar HAPP was never built. These reports integrate existing initiatives and platforms into the plan and projects what new technologies may mature during the same timeframe. Often the design methodologies are closely guarded as proprietary information in the highly competitive UAV industry. OSD 2002]. Detailed methodologies are occasionally provided on conceptual design studies that do not result in flight hardware. but specific design methodologies are not highlighted. The descriptive plan focuses on existing systems or currently planned initiatives. and flight test of the Boeing Compass Cope. Air Force Scientific Advisory Board 1996] describe future requirements. Brown [1980] describes the conceptual and preliminary design activities. and implementation possibilities. The level of engineering may not be worthy of publication in some cases. Johnstone [1990] describes technical challenges and solutions obtained for the Condor HALE UAV program. published by the Office of the Secretary of Defense. 5 . these sources do not attempt to directly determine the design impacts that may result from insertion of the provided projections through aircraft design and optimization. useful design information and technology trends supporting the technology modeling development are usually not available. Despite the wealth of technology information brought forward in the UAV roadmaps [OSD 2001. The Solar HAPP system was intended to fly extended missions at high altitudes using solar power and secondary power sources. These roadmaps constitute the most definitive sources of UAV technology projections available today. Many methods presented in the reports have sufficient detail to be repeatable. (12) The National Academy of Sciences 2000. Lockheed Missiles and Space provided a series of highly detailed papers on the Solar High Altitude Powered Platform (HAPP) design studies. Hall [1983] gives extensive conceptual and preliminary design methodologies for the aerodynamics and propulsion system.The DOD UAV Roadmaps [OSD 2001. The roadmap does not integrate the technology projections into UAV system designs. though details on specific design methodologies are limited. OSD 2002]. though many methods are tailored only to the Solar HAPP application.
All geometry. particularly LRN aerodynamics. Naval Research Laboratories (NRL) has advanced the state of the art in Low Reynolds Number (LRN) UAV design. GA optimizers offer a robust optimization capability that is insensitive to local optima and existence of smooth objective functions. and performance algorithms. Raymer [2002] presents a method for optimizing UCAV AVs with genetic algorithms. Technology trends are applied to the requirements of the three calibration cases to evaluate technology impacts on AV design. Wherever practical. Raymer’s research appears to be limited to UCAV cases. 1. Sources describing UAV concepts and systems abound. The methods are calibrated against HALE. There does not appear to be any previous research covering an integrated design and optimization methodology for relatively arbitrary Air Vehicles (AVs).3 Approach The AV MDO methodology is applied through a selfcontained software application. which are not covered here. The Forestry UAV flew with demonstrated performance similar to the predictions. The profile drag stripbuild up method and the lifting line theory application described in this source are applied here. fixedwing UAV AV design optimization problems. Few describe the design methodologies in sufficient depth to permit reuse. The AV analysis methodology is largely based upon legacy methods from Raymer [1992] and the associated RDS software. to name a few. mass properties. Existing research is limited to specific design problems that are narrow in scope.2 Problem Statement An integrated MDO methodology is developed and applied to subsonic. NRL publications [Foch 1992] highlight key technical issues affecting LRN AV design. relative to calculusbased 6 . developed by David Carrol of UIUC [Carrol 1998]. structures. 1. though no consolidated design synthesis approach is provided.Gundlach [11999] presents the design approach used for the Virginia Tech Forestry UAV. The selected GA is the GA170 code. Raymer [2002] emphasized exploration of GA utility for aircraft optimization. aerodynamics. multiple methods are provided to trade level of detail. are included in the provided methodology. and accuracy. Many sources describe individual design disciplines in isolation. A Genetic Algorithm (GA) is employed to optimize the vehicles for minimum design gross weight. propulsion. MAE. and tactical UAV classes. computational speed. Gundlach [11999] does not describe the design methodology in sufficient detail to be repeatable.
This permits the figure of merit to be minimum design gross weight. Predator. and technological revolutions. Medium Altitude and Endurance (MAE). funding. to cover a broad range of vehicles and missions. The relative size of these systems is shown in Figure 18. The presumption is that the evaluated vehicle design gross weight will decrease as technology advances. since there is a strong correlation between vehicle weight and cost. and High Altitude Long Endurance (HALE). The MAE reference system is the General Atomics Predator (A). Design gross weight is frequently used as a cost analogy by aircraft design codes. In other words. but the average weight of AV within a category may actually increase with time due to enhanced capabilities. History shows that the minimum possible AV size for a given mission does decrease with time. The 7 . Figure 18 Scale Comparison of Shadow 200. Three classes of UAV systems were selected for evaluation: the small tactical. However. The flight and payload performance requirements are held constant for all cases. The small tactical UAV requirement is filled by the AAI Shadow 200 system. and Global Hawk (Left to Right) Technology modeling is applied to the AV MDO code to demonstrate the capabilities of the methods employed and to provide insight into technology impacts of AV design. Determining future requirements is beyond the scope of this effort. this robustness comes at some expense of refinement efficiency. since this is largely a function of evolving military doctrine. The HALE reference system is the Northrop Grumman (previously Teledyne Ryan) Global Hawk. UAV requirements typically increase with advancing technology. politics.optimization methods.
The geometry. propulsion.direct. Technology impacts are evaluated by introducing advancing technologies to a reference technology year. OSD 2002] and other sources. and optimized UAVs utilizing advanced technologies. and structures. systems (size. The technology period under consideration spans 31 years. All technologies are ranked for substantiation. are applied to AV designs in this research. Then. weight and power). quality and potential design impact. the genetic algorithm optimizer and its integration are described. The conclusions chapter summarizes key findings and provides recommendations. optimized UAVs with technologies similar to the existing UAVs. an overview of the methodology and code architecture is presented. This document provides details of the AV MDO methodology and describes its application to existing and future UAV systems. and assumed technology trends. derived. and the Shadow 200 system represents year 2000 technology. The Global Hawk and Predator systems began development in 1994. First. Next. The technology areas explored include aerodynamics. The next three chapters describe the application of the methodology to existing UAVs. the technology modeling is detailed. general design. based on the roadmaps [OSD 2001. AV designs for 1994. and performance analysis methods are presented in the next chapters. and every five years from 2000 to 2025 are evaluated. from 19942025. 8 . propulsion. avionics and subsystems. aerodynamics. and individually at 1011 years after the reference technology year. payloads. The AV MDO code exploits any advantages gained by the applied technologies. The technologies are considered in combination for all years.
The strengths and limitations of several methods were evaluated by the author. 2. The final methods are integrated into the software. As the code 9 . full or partial functionality code was run to verify modification effectiveness. structures. systems analysis and other elements. a typical software development productivity metric is 2. The AV MDO methodology is applied in the JDES software. it was gradually replaced by improved versions.1 Methodology Overview The AV design synthesis is performed with the aid of an integrated Multidisciplinary Design Optimization (MDO) code. JDES is written in the Fortran 90 programming language. including algorithm reuse. including spaces and internal documentation. Other methods for the other analysis disciplines gradually replaced the prototype place holder subroutines.000 physical lines. in its current form. This software contains approximately 30.000 SLOCs per equivalent head per year. took approximately 4 years to develop. A spiral development approach was taken to generate the JDES software. The roughly estimated Source Line of Code (SLOC) count is approximately 8. Important aspects of the code were initially implemented with temporary prototype subroutines. Throughout development. The MDO code consists of an AV design code and a GA optimizer. since the aerodynamics methods are highly complex and innovative approaches are attempted. Once the prototype code proved functional.Chapter 2 Methodology 2. JDES. since the outputs are widely used by other methods. the most promising applicable methodologies were integrated and evaluated. For reference. weights.000. Original methods were developed where none existed or existing methods had unacceptable limitations for this application. The refined aerodynamics followed. running the thencurrent baseline code helped minimize gross errors through developer inspection of results. performance. aerodynamics. In several cases. propulsion.2 Methodology Development The methodology selection process involved a search of available methods. Throughout development. The optimizer proved to be a highly useful tool for identification and isolation of development errors. The first refined method was the geometry definition. The AV design code includes geometry.
The functional flow is shown in Figure 21. Next. subtle interactions among subroutines became increasingly difficult to identify. WEIGHTS is called to evaluate the vehicle mass properties and other systems evaluations. Finally. Then AEROC calculates all of the vehicle aerodynamic tables. GET_OPTFILE is called to read the optimizer settings.became increasingly complex and mature.3 Applied Methodology Architecture The main subroutine in JDES is called ACMAIN. Otherwise. A loop is entered for all the cases considered. which calls all input file reader subroutines. Through the development and result generation process. then the OPTIMIZER master subroutine is called to execute the optimization. TEMP represents a number of subroutines that assigns. The methods contained within these subroutines are described in later sections. and resets temporary values to data that may change during code execution. graphics are generated for the final vehicle by the DXF subroutine. First. a new mode of error management became possible. The first subroutine called from ACMAIN is FILELIST. Propulsion generates a tabular engine deck. the EVAL design analysis subroutine is called directly with the input design variable settings. The optimizer tends to exploit weaknesses in the code. Next. The first step within the loop is to call FILEMANAGER. Figure 21 ACMAIN Code Architectre The toplevel functional flow of EVAL. PERFORM is invoked to perform the AV mission performance evaluations and selfoptimizations. After the aircraft is evaluated or optimized. which contains a list of all the evaluation cases. much of the vehicle geometry is calculated in GEOMETRY. If optimization is required. stores.5x106 times. was called by the optimizer approximately 1. 10 . the primary aircraft evaluation subroutine EVAL. is shown in Figure 22. Once the optimizer was integrated. 2. the primary AV design evaluation subroutine.
Unfortunately.4 Methods Multiple approaches are available for several methods. The graphics file generator. The aircraft design code represents a large body of original development. For several disciplines. Latent code capabilities are covered. Many options have origins from prototype methods. Special emphasis is placed on the methods that are applied to generate the results for this research effort. and the ideal lift distribution method algorithms are strongly inspired by the VPI&SU StrutBraced Wing (SBW) transonic transport research publications. unused procedures are given coverage to highlight limitations to these approaches. piecewise linear beam wing bending material weight estimator. but these remain to enhance analysis flexibility. Other influences and applied methods are cited throughout the methodology chapter. though not extensively. multiple methods are available with similar levels of detail. the detailed methods are typically more computationally demanding. final method selection for this research is based upon inspection of results and assessed appropriateness of functional dependencies. The detailed options were generated when the simple methods did not satisfactorily represent the phenomena of interest.Figure 22 EVAL Code Architecture 2. By keeping both simple and detailed methods. The simplified prototype methods were largely supplanted by the more detailed options. In this case. 11 . the user can trade between detail and computational efficiency. Some algorithms were developed by the Author as early as 1993. The methodology chapter describes the methods utilized. Occasionally.
and mission performance are matched to the target values available from various sources. Individual methodologies are validated whenever practical. since experimental data or method test cases are unavailable. and structural material information. such as component lists. Predator. 12 . payload types.5 Calibration The integrated code is calibrated against the Global Hawk. subsystem descriptions. and penalty factors are applied as appropriate to facilitate the calibration process. takeoff gross weight. technology factors. the empty weight. Applicable experimental data and test case references are assembled to evaluate the method validity.2. Other data is obtained for calibration. For these new or modified cases. Various design levels. and Shadow 200. Some methodologies are new or incorporate modifications that make calibration difficult. flight envelope limitations. In all cases. loading. to name a few. calibration often occurs through comparison with alternative competing methods that are included in the code.
Chapter 3 Optimization 3. and the end result is typically highly dependent on the starting conditions. crossover. where calculusbased methods may have limitations. A GA is used to flexibly handle a wide range of design problems.1 Optimization Overview The aircraft design code. many calculus methods may handle constraints directly. GAs are adept at finding the global optima. operates under both analysis and optimization modes. Within the vicinity of local optima. The parameters are coded in binary strings of zeros and ones. GAs do not handle constraints directly. By comparison. Calculus methods require the existence of derivatives and smooth design space. Each string represents a member of the population. The design variables are selected to be powerful and intuitive in driving the optimizer to the global optima. Calculus methods start from a point in the design space. and combined in larger strings. GAs are highly capable of finding global optima within challenging design spaces. custom refinement algorithms are optionally available. GAs do not operate on design variables. not a single point. A Genetic Algorithm (GA) is employed to optimize the AV designs. and mutation. not necessarily the global optima. 13 . A nonsmooth design space with discontinuities is acceptable. known as gradients. The GA operates on the strings. JDES. To help contend with this limitation. The string operators are reproduction. refinement of the global optima is inefficient relative to calculus methods within the vicinity of the optima. However. Genetic algorithms handle the design problem in a significantly different manner. GAs do not require the existence of gradients. The utility and limitations of GAs can be elucidated best by comparison with the competing calculus methods. There are multiple individuals in the population for each generation. The analysis mode uses inputs provided in the input files with no optimized variable changes except for automatic internal convergence loops. and then decodes the strings into the parameters prior to evaluating the objective function. since the search is based on a population of design points. called the fitness function. The optimal solution is the local optima. calculus methods can be very efficient at finding the refined local optima. called parameters. However. Violated constraints must be converted to penalty functions that are applied to the objective function. directly. Calculusbased methods rely on derivatives of the objective function with respect to the design variables. The optimization mode uses the data from the input files and userdefined design variables. this comes at some expense of efficiency in finding the refined optima.
and optimizer search refinement.2 Optimization Code Architecture David Carrol’s GA170 code required significant modification and additional code to be useful for this application. but the language was modified to Fortran 90 by the Author for this application. New code was written for conditioning the design variables. The code was originally written in Fortran 77. This code. 3. using the methods described above.David Carrol’s GA170 [Carroll 1998] optimizer was selected for this research. The standard GA170 code acts as a main subroutine. constraints. is freeware and written in Fortran 77. but the UAV code has a main subroutine that controls the high level execution of the code. EVAL Set DV Code DV GA Inputs DV. developed at University of Illinois at Urbana Champaign (UIUC). GA170 was modified to act as a subservient subroutine called by ACMAIN via the OPTMIMIZER subroutine. 14 . Constr Inputs OPTIMIZER GA170 GA String Operators EVAL Decode DV Objective FUNC Penalty Reset DV Bounds Record Fitness ACMAIN EVAL Figure 31 Genetic Algorithm Code Architecture Figure 31 shows the GA code architecture. both key attributes contributing to its selection. fitness function. The shaded blocks represent new code developed by the Author. The basic principles of GAs are wellestablished. The GA170 code is a true GA. The functions represented by the blocks are not necessarily contained within a single subroutine or as a distinct subroutine. and the white blocks represent existing or slightly modified code contained within David Carrol’s GA170. A more extensive treatment of GA theory similar to that applied here can be found in Goldberg [1989].
OPTIMIZER has the option of refining the permissible solution set within the solution space based on the knowledge of the fit individual array. decodes the strings into design variables. 15 . This procedure may be performed many times until either the maximum number of iterations is reached or the difference between the minimum and maximum fitness values reaches a specified tolerance. The GA170 code is a highly useful GA. The random number generator selects the same random numbers starting at the initiation of software execution. This procedure occurs for the specified number of generations. Subroutine FUNC combines the objective function and the constraint penalty functions to produce the fitness value for GA170 to utilize. OPTIMIZER controls the execution of GA170. such as design variable limits and constraint information. ACMAIN will call the OPTIMIZER subroutine. The final product fully satisfies the needs of the research problem set. This new feature is not applied to the final cases. number of generations. The ACMAIN code will simply call EVAL. Once the GA data is input. the results are repeatable. These individuals are represented by genotypes of bit strings. The optimized solution and the fit individual array are sent back to Optimizer after the final generation is run. OPTIMIZER calls EVAL to evaluate the best design and write detailed design information to output files. and evaluates the fitness of the individuals. Optimizer can call GA170 multiple times with refined design variable limits such that the global optima may be refined. Although randomization is used by the GA. OPTIMIZER is the primary optimization control subroutine. reads the generic GA setting data and controls the execution of the GA. It reads the specialized GA inputs. the primary aircraft evaluation subroutine. since the baseline GA algorithms are sufficient. FUNC calls a subroutine to record the most fit members of populations over all generations that do not have violated constraints. EVAL calculates the objective function. Once the best solution is found. The constraint penalties are summed by FUNC and combined with the objective function to produce the fitness function. the main subroutine. and evaluates the individual constraint values and penalties. The fitness value for each individual is found in subroutine FUNC. GA170 performs the GA operations on these strings. when in the analysis mode. and probability of mutation and crossover settings. When in the optimization mode. FUNC calls EVAL with the design variable settings. The fittest individual in the final generation is output. The GA creates an initial population of individuals based on the design variable limitations. GA170 contains the GA algorithm. The optimized solutions are identical when evaluated in the same order. The generic GA setting data includes population size. the minimum and maximum design variable limits are brought closer to the minimum and maximum design variable values for the fittest individuals with no violated constraints. In this scenario. Substantial new code was required to integrate GA170 into a flexible and userfriendly UAV optimizer.The ACMAIN code.
The increased fairing size impact on drag can have positive or negative drag contributions. High aspect ratio wings have lower induced drag and higher available total lift coefficient for a given wing area than low aspect ratio wings.3 Design Variables The universally applicable design variables are shown in Table 31. because the absolute airfoil thickness is higher where the bending moment is highest. Wing sweep can delay the onset of compressibility and reduce the impacts of wave drag. which can drive up the profile drag. Taper improves structural weight. a reduced taper ratio increases the root chord length. Wing taper permits a wing to approach the ideal elliptical lift distribution more closely than without taper. The geometry design variables applied here were selected to have the greatest impact. Sweep is one way to counteract compressibility impacts on drag. but at the cost of a reduction in laminar flow and less structural efficiency. because higher Machnumber Global Hawk contends with compressibility effects. With a constant wing area and wingspan. The optimizer seeks to minimize the vehicle design gross weight while not violating constraints.3. The aspect ratio drives aerodynamic efficiency at the expense of structural weight. High aspect ratio wings have a higher structural weight than low aspect ratio wings of the same area due to higher bending moments and shear loads in the wing spar. Exclusive of high lift devices and for a fixed airfoil. the wing tip Reynolds number is lower than the average chord Reynolds number. which reduces the induced drag. 16 . The geometry is a strong contributor to vehicle weight. The taper ratio affects the induced drag and structural efficiency. Some of the trades performed by the optimizer with the design variable selections are described here. The increased area moment of inertia permits the wing skin thickness to be reduced for a given loading. though these effects are not evaluated here. higher aspect ratio wings tend to have lower wing areas due to the increase in maximum lift coefficient. Variable bW ARW λ t/c MFFuel Description Wing span Wing aspect ratio Wing taper ratio Wing thickness to chord ratio Fuel mass fraction Table 31 Universal Design Variables The Global Hawk case additionally provides wing quarter chord sweep as a design variable. At low Reynolds numbers. Wing sweep trades are documented in Gundlach [21999]. Sweep is a unique design variable to this case. Aerolastic phenomena are more dominant in high aspect ratio wings. which increases the wingfuselage fairing size in the parametric geometry.
the parametric geometry used for all optimization cases limits the minimum fuselage length via constraints that are functions of span and average chord. T/W. but tends to adversely affect the wing profile drag and hastens the onset of compressibility effects. The constraint types are listed below in Table 34.SLS .SLS Description Uninstalled SLS thrusttoweight ratio at TOGW Uninstalled SLS thrust Table 32 Jet Propulsion Design Variables The propulsion design variables for reciprocating engines are shown in Table 33. Transonic wings tend to manipulate wing sweep and airfoil thickness to chord ratio simultaneously. tail. as described in the geometry section. The weight in P/W is the takeoff gross weight of the aircraft.Increasing the airfoil thickness to chord ratio increases structural efficiency. Tmax. Variable T/W Tmax. Additionally. Uninstalled refers to the advertised engine characteristics that do not take into account for engine installation impacts due to inlets or nozzles. The optimizer tends to minimize fuselage length and wing span such that the volume constraints are not violated.4 Constraints Several constraints are used by the optimization code. and fuselage weights are impacted by scale. Unique penalty functions are calculated for violated constraints. The wing.may not be used simultaneously. Variable P/W PShaft Description Uninstalled brake powertoweight ratio at design gross weight Uninstalled brake horsepower Table 33 Reciprocating Engine Propulsion Design Variables 3. 17 . and SeaLevel Static (SLS) thrust. The propulsion design variable for jet aircraft are shown below in Table 32. Note that P/W and PShaft may not be used simultaneously. The user can select which constraints are imposed prior to running the code. The thrust is evaluated at the uninstalled SLS maximum thrust conditions. The parametric tail geometry sizes the tail surfaces as a function of tail moment arm. The weight in T/W is the takeoff gross weight of the aircraft. The power is the uninstalled sea level static brake power. Note that the thrust to weight ratio. so the fuselage length reduction is tempered by tail size. Uninstalled refers to the advertised characteristics of the engine that does not account for aircraft installation.
Min. Similarly.Table) Table 35 Mission constraint types The fitness function is a combination of the objective function and the penalty functions.Constraint Mission Complete Mission Constraints Fuel Volume Description Constraint violated if prescribed mission is not completed due to performance limitations or lack of fuel Constraints violated if the set of mission constraints (Table 35) is not fully satisfied over the complete prescribed mission Constraint violated if the required fuel volume is greater than the available allocated volume Table 34 Constraints The mission constraint types are listed in Table 35. If the objective is to minimize the weight. and the weight is the objective function. then the fitness should be the negative of the objective function. then the penalty contribution to the fitness function is the negative of the penalty. A negative fitness with a lower magnitude is considered a greater number. so lowering the negative magnitude is a means of maximizing the fitness. The GA attempts to maximize the fitness.Max) Insufficient thrust for minimum specified rate of climb Insufficient thrust for level flight Dynamic pressure limit (q > qMax) Mach limit (M > MMax) Engine operating limits (type dependent) Altitude limit Weight error (Weights < 0) Lift lower limit (CL < CL. if the violated constraint penalties are represented in terms of weight. Constraint Number 1 2 3 4 5 6 7 8 9 Description Stall (CL > CL. The fitness function is: Fitness = −(Objective + ∑ Penalty ) 18 .
When the objective function is purely a function of design gross weight (WDG). the penalty function takes the form of a quadratic equation with a step function starting when the constraint is violated. Mission Completion The mission completion penalty function is called when the mission completion is less than unity. The penalty parameter is: Param = 1 when a constraint is violated.1 + Param + Param 2 ⋅ PenaltyWt ( ) 19 . the constraint is either satisfied or not satisfied. In other words. The penalty functions for each of the constraints are described below. There is no variation in the degree of constraint violation. The step function ensures that there is an immediate penalty for a violated constraint. Each penalty has a userdefined weighting parameter.Ref The penalty functions are intended to be referenced to the same order of magnitude. Param = MissComp − 1 The penalty function is: Penalty = 0. The variable penalty function is: Param = f (Constraint ) When constraint is violated Penalty = Step + F1 ⋅ Param + F 2 ⋅ Param 2 ⋅ PenaltyWt When constraint is violated ( ) Other constraints take the form of discrete differences. The quadratic equation increases the penalty more aggressively as the magnitude of constraint violation increases. then the objective function is: Objective = WDG WDG. For constraints that vary.The objective function is made to be on the order of 100 (or unity). Discrete constraint penalty functions take the following form: Penalty = F1 ⋅ Param ⋅ PenaltyWt when a constraint is violated.
1 + Param + Param 2 ⋅ PenaltyWt when the constraint is violated. Penalty = 0. and each type is evaluated individually.Type when the margin for a given type of volume is negative. Type VolAvail. ( ) 20 .when the constraint is violated Performance Constraints The performance constraint penalty function is found by: Param = NViolated The penalty function is: Penalty = Param ⋅ PenaltyWt Fuel Volume Constraints The wing volume penalty function is found by: Param = 1 + VolMargin. The type refers to wings and fuselage.
not the likely. However. However. Many trends are based upon the possible. It is possible to accelerate a technology with aggressive funding. this research endeavors to describe what may potentially happen. payloads. priority. This Law has remained notably accurate since 1965. or inspiration. The summation of all technology categories is shown in Table 41. and structures categories. Most technology trends tend to have high levels of uncertainty. since technology trends that endure for more than a few short years are rare.1 Technology Overview Technology modeling into the future is an art. this does not mean that the embryonic technology is suitable for operational applications. technologies may also experience a maturation period prior to garnering sufficient attention to stimulate investment. 21 . quality. not what will. Predicting enabling funding relationships can be as challenging as predicting technology breakthroughs. Nascent technologies may demonstrate impressive performance under laboratory conditions. it is important to capture impacts of technology emphasis and level of design in the predictions. Details of each technology are described throughout this section.Chapter 4 Technology Modeling 4. Transforming the laboratory technology into a fieldready product may result in performance alteration due to practical integration considerations. The individual technologies composing the technology categories were qualitatively ranked for substantiation. To counter the time lag between demonstrated technology and operational capability. and aggressiveness. In light of the inherent limitations associated with divining the future. Long term technology projections contain high uncertainty. avionics/subsystems. Expectations of accuracy become increasingly unrealistic as the time into the future extends. propulsion. Moore’s Law. design. This research attempts to accommodate both considerations. The technologies are grouped into aerodynamics. and the form of this projection is expected to hold true into the coming two decades [OSD 2002]. One of the most famous exceptions. states that the number of transistors on a microprocessor doubles approximately every 18 months.
Table 41 Technology Attributes The ranking standards for each attribute are shown in Table 42. Table 42 Attribute Ranking Description 22 .
An example is a technology factor applied to an input baseline avionics weight. In the case where Year lies beyond the final Year2 date. The technology parameter ratio. The technology phase lag.As can be seen from the above tables. Phase is found through a lookup procedure based on phase lag versus technology level input data.2 Technology Methods Two primary methods of technology modeling are applied to the design code. The procedure is shown graphically in Figure 41 (AB). Method 1. there is only one directly applicable existing technology trend with a high relative chance of success and anticipated high impact on the optimization objective function. (A) (B) Figure 41 Technology Parameter Ratio Model (Method 1) A single technology curve composed of discreet technology versus technology year data points is used. is used whenever the ratio of technologies is multiplied or divided by a baseline value. Phase. 4. These are technology parameter ratios (Method 1) and direct technology parameter lookup (Method 2). The technology curve is shifted to the future by the time amount Phase. The technology for a given lookup year is: (Year Tech(Year) = Tech1 ⋅ RYear −Year1 ) where Year is automatically bound by Year1 and Year2 in the adjusted technology curve. to obtain an adjusted technology curve. the following extrapolation occurs: 23 . is the time difference between availability of the technology and its application.
The baseline characteristic of interest is multiplied or divided by the technology factor. The technology ratio between Year1 and Year2. so the reference year is 1994. For example. For example. The design level ranges from 01. the Global Hawk program began in 1994. is suitable for design technologies that are not ratios. FTech. This lookup procedure is shown graphically in Figure 42. as appropriate. where 0 represents the lowest design level and 1 is the highest design level. and design level 0 represents the equivalent of the maximum phase lag. because design level of 1 equates to the minimum phase lag. 24 . Method 2. If the Global Hawk requirements were evaluated in 2005. The direct technology parameter lookup model. the technology is set to the earliest technology level without further extrapolation. The equivalent of a phase lag is built into the design level. RTech.(Year Tech(Year) = Tech2 ⋅ RYear − Phase −Year2 ) Whenever Year is before the earliest Year1 date. is found by: Tech(Year2 ) RTech = Tech(Year ) 1 The technology factor. then the design year would be 2005. is found by: FTech = Tech(YearDes ) Tech(YearRef ) 1 Year2 − Year1 where YearDes is the design year and YearRef is the reference year of the baseline. the structural design factor of safety as a function of design year and design level is appropriate for a lookup function.
the attributes of these technologies are repeated in Table 43. interference drag. and profile drag. wave drag.3 Technology Modeling 4. This assumption is based on the advancement 25 Quality . Aggressiveness 2 2 2 3 Substantiation Aerodynamics Induced Drag Interference Drag Wave Drag Profile Drag 4 3 3 3 3 2 2 2 Table 43 Aerodynamics Technology Attributes Induced Drag The induced drag technology is modeled as a factor applied to the minimum theoretical induced drag found through the Trefftz plane analysis. The trends are assumed to converge to the ideal as time progresses.3.1 Aerodynamics The aerodynamics technologies are divided into induced drag. For convenience.Figure 42 Direct Technology Parameter Lookup Model 4.
95 1920 Des Lev 0 Des Lev 0. This assumption is justified by the proliferation of CFD codes and design methods. Design Year Induced Drag Multiplication Factor 1. The interference drag multiplication factor is applied to the standard Hoerner interference drag estimates.05 1 0.5 Des Lev 1 1940 1960 1980 2000 2020 2040 Design Year Figure 43 Induced Drag Multiplication Factor Technology Trends Interference Drag Hoerner [1992] interference drag methods are used. Therefore. The general trend is for the interference drag multiplication factors to approach the minimum value as time progresses. Induced Drag Multiplication Factor vs. Hoerner notes that the interference drag of a strut junction may be reduced to 10% of the data presented if properly faired. Due to the significant impacts of the interference drag on vehicle geometry optimization. as described later in the aerodynamics section. the interference drag design level is set to 1 for all cases. The trends are shown in Figure 44. Much of Hoerner’s data and methods originated in Germany during World War II. The assumed technology trends are shown in Figure 43.15 1. 26 .2 1. This trend is assumed to hold for all other interference drag contributors from 1940 through 1995.1 1. Some improvement in aggressive design levels are permitted between 1995 and 2025.of Computational Fluid Dynamics (CFD) in wing design optimization. the starting date is 1940.
The factor for a Boeing 747 is 0. These commercial transport values are used to estimate the stateoftheart trends.6 0.85 0. Wave Drag Technology Factor vs. One noteworthy assumption is that the lowest technology line for 1995 through 2025 is assumed to be consistent with the technology associated with the 747.8 0.5 Des Lev 1 2020 2040 Figure 45 Wave Drag Technology Factor Trends 27 .Interference Drag Multiplication Factor vs.4 0.8 0.95 0.955 [Grasmeyer 1998].7 1960 1980 2000 Design Year Des Lev 0 Des Lev 0. and for a Boeing 777 with supercritical airfoil sections of 1995 technology year is 0.5 Des Lev 1 Design Year Figure 44 Interference Drag Trends Wave Drag The wave drag is modeled through the transonic airfoil technology factor used in the Korn equation [Grasmeyer 1998]. The transonic airfoil technology factor trends are shown in Figure 45.2 1 0. Design Year 1 Transonic Airfoil Technology Factor 0.2 0 1920 1940 1960 1980 2000 2020 2040 Des Lev 0 Des Lev 0.9 0. Design Year Interfererence Drag Multiplication Factor 1.75 0.89.
Profile Drag The profile drag technology is modeled as a multiplication factor to the profile drag.4 0. Table 44 shows the selected airfoils.000.000 3.6 0. The referenced airfoils represent assumed typical stateoftheart for the airfoil design year.000 Airfoil NACA 0009 FX 60100 HQ 2.2 1 0. By the first reference date of 1949. much development and testing had been performed on airfoils 28 . These factors for the selected airfoils are plotted in Figure 46.000 300.000 300.000 3.000 60.000 60.5 47 71 93 95 100 Source Selig [1989] Selig [1989] Althaus [1985] Selig [1989] Selig [1989] Abbot [1949] Somers [192] Table 44 Airfoil Characteristics.8 0. lower speed manned aircraft such as general aviation types.2 0 1900 Re=3M Re=300K Re=60K 1920 1940 1960 1980 2000 Design Year Figure 46 Profile Drag Multiplication Factor Trends The technology trends derived from the available data are highly Reynolds number dependent.3 45. Data was compiled for high performance airfoil sections over a wide time span for three Reynolds numbers covering the majority of the range of interest here.000. The profile drag multiplication factor is the ratio of the inverse of the lift to drag ratios referenced to the starting year condition for a given Reynolds number.59 Clark Y FX 63137 NACA 2415 SM 701 L/D max 26. Airfoil Year 1949 1980 1985 1922 1980 1949 1992 Reynolds Number 60. Profile Drag Multiplication Factor vs. The 3 million Reynolds number case represents the flight regime experienced by small. Design Year Profile Drag Mulitplication Factor 1.
6 1940 Des Lev 0. These codes can realistically model and aid in the design of low Reynolds number airfoils. Design Year 1. had relatively few applications beyond model aircraft.000 range.000. Therefore. The profile drag multiplication factor of 1 represents the state of the art in 1949. a curve approximately splitting the two trends was employed. especially near 60. as a best judgment.8 0. are freely available. low Reynolds number databases available. spurred on by two world wars and 44 years of commercial air transportation. The stateoftheart airfoils in 2025 are assumed to have a 20% improvement over the stateoftheart in 1949 for a given Reynolds number. 29 .4 Profile Drag Multiplication Factor Des Lev 0 1.2 1 0. A single function that is Reynolds number independent is desired for the profile drag multiplication factor. It was not until the late 1970s that the low Reynolds number regime (Re < 300. The presumption is that after the year 1990. high performance boundary layer codes. the poorest airfoil designs employed on UAVs will be the equivalent of the stateoftheart in 1949 for an equivalent Reynolds number. Profile Drag Multiplication Factor vs.5 Des Lev 1 1960 1980 2000 Design Year 2020 2040 Figure 47 Final Profile Drag Multiplication Factor Trends. The ratio of profile drag multiplication factors is applied as a multiplication factor to the calculated profile drag at a given design level. such as XFOIL. highquality. The UAVs under consideration typically operate with wing chord Reynolds numbers near the 300. the lower Reynolds number airfoils.000. By comparison.operating in these Reynolds numbers. Furthermore. which are frequently used by competitive model sailplane enthusiasts and UAV developers. The level of design was also adopted to bound the variation of design. Today. there are multiple.000 to 3.000) received serious attention by technical hobbyists and engineers. Figure 47 shows the resulting assumed airfoil technology trends.
2 Avionics and Subystems Technology trends for various types of avionics are evaluated. Navigation and Control (GNC) avionics includes the INS. Figure 48 and Figure 49 show the GNC technology phase and technology trends. The GNC trend is applied to all avionics except for the air data system and INS. The laminar flow technology impacts are described in the Aerodynamics section. respectively. Navigation and Control The Guidance.Laminar Flow The laminar flow technology affects transition for 2D airfoils and 3D wings with sweep. Data for similar functionality GNC systems from manufacturer specification sheets was used to generate the overall GNC trends. 4. The attributes of these technologies are repeated in Table 45.3. Aggressiveness Substantiation Avionics/Subsystems Autopilot INS Subsystems Communications 3 3 2 4 1 1 3 3 Quality 1 1 1 2 Table 45 Avionics and Subsystems Technology Attributes Guidance. The applied technologies are independent of year. flight computers. 30 . and the formulation takes a different form than other technologies. and varying flight critical sensors. Werrell [1985] noted that a 1970 cruise missile GNC weight is 115 pounds.
respectively.1 1970 1980 1990 Technology Year 2000 2010 Figure 49 GNC System Technology Trends. The trends carry the same functionality forward to current technology. The assumed guidance system technology phase and technology trends are shown in Figure 410 and Figure 411. enabling a miniaturized cruise missile revolution [Werrell 1985]. By 1970.8 1 Technology Level Figure 48 GNC System Technology Phase. Years 10 8 6 4 2 0 0 0. the same functionality system weighed only 29 pounds. Pounds 100 10 1 0.GNC Technology Phase vs.6 0. using manufacturer specification sheets. Technology Year 1000 GNC Weight.4 0. Inertial Navigation System UAV inertial navigation systems (INS) have seen much progress over the past half century. In 1960. Technology Level Technology Phase. a cruise missile INS weighed approximately 300 pounds [Werrell 1985].2 0. GNC Weight vs. 31 .
8 1 Technology Level Figure 410 INS Technology Phase INS Weight vs.4 0.01 1960 1970 1980 1990 2000 2010 Technology Year Figure 411 INS Technology Trends Processors Much of the avionics revolution is made possible through the advancement of processor technology. There is a phase lag between the availability of commercial processors and the application to aircraft. because demand placed on processors increases as processor capability improves. The INS and GNC weight trends are utilized to estimate avionics weights. The DOD UAV Roadmap [2002] has tracked commercial processor speed through time. 32 . Years 5 4 3 2 1 0 0 0. Although a processor performance trend is available for technology modeling within the design code.6 0. though not used to generate the results. The processor speed is not proportional to actual avionics weight. Pounds 100 10 1 0. Technology Level Technology Phase.2 0. This technology trend was adopted directly in this research.INS Technology Phase vs.1 0. it is not utilized here. Technology Year 1000 INS Weight.
The mass fraction of the empty weight less subsystems is assumed to be 0. are used for subsystems technology modeling. but competing subsystems technology information suitable for a model is unavailable. respectively.Tech Numbers representative of the Boeing 777 are used as an estimate for the 1995 mass fraction numbers. the subsystems weight multiplication factor must equal 0.05. Years 5 4 3 2 1 0 0 0. but the total synthesized aircraft weight reduction. the anticipated total gross weight reduction due to subsystems contributions is estimated at 9% [Defense and Aerospace 1995]. The mass fraction of the 1995 subsystems is assumed to be 0. however. when the subsystems weight multiplication factor is equal to 1.6 0. The 1960 subsystems weight multiplication factor is assumed to be 1. − Subst − MFSubs ⋅ FSubs . or 10% greater than the 1995 subsystems weight.8 1 Technology Level Figure 412 Subsystems Technology Phase 33 . which present Defense and Aerospace study [1995] results. This technology trend is assumed to be applicable to UAVs in absence of better subsystems weight technology models. The fuel mass fraction is assumed to be 0. Technology Level 6 Technology Phase.The demands placed on processors typically increases faster than the improvements in processor speed [Noor 1997]. transonic transports. Subsystems Trends found in Noor [2000]. A very simple sizing relationship for a given aircraft class is: WTO 1 = WPayload 1 − MFFuel − MFE .1.4. is not the subsystems weight reduction.2 0. The assumed subsystems technology phase and multiplication factor trends are shown in Figure 412 and Figure 413. Subsystems Technology Phase vs. since subsystems technology has improved between these dates. The GNC weight trends are more representative of attainable weight reductions.4 0. In order to achieve a 9% reduction in takeoff gross weight due to subsystems contributions. 42. All UAVs considered here are much smaller than transonic transports.75. This reduction. The subsystems contribution is found by a simple mass fraction analysis. For longrange.
It is difficult to predict the advancement of communication satellites through 2025. and more efficient usage of bandwidth will act to relieve the communications challenges facing UAVs. the antenna takes up considerable volume and drives the shaping of the fuselage. No direct impacts on aperture sizing are provided.Subsystems Weight Multiplication Factor vs. an assumption was made that for equivalent collection functionality. if so equipped. 34 . In light of the OSD predictions. The assumption that there will be no communication satellite or related UAV antenna technology advancement is bolder than assuming that technology will progress. This relationship is shown in Figure 414. the dish antenna diameter on UAVs will decrease to 50% of the 1994 values by 2010 and to 25% of the 1994 value by 2025. preprocessed payload data. These are largely driven by the satellite constellation that the UAV is communicating with. as seen on the Global Hawk and Predator.2 1 0. The gain requirements are in turn driven by the desired data rates.8 0.6 1960 1970 1980 1990 Design Year 2000 2010 2020 Figure 413 Subsystem Weight Technology Trend Communications The most significant communications system impact on UAV design is the antenna for satellite communications. and beam width requirements. The DOD UAV Roadmap [OSD 2002] predicts that data compression. is driven primarily by the antenna gain requirements. It is even more difficult to translate the communication satellite capabilities into UAV antenna diameter. In addition to the weight and power of the antenna and gimbals. This trend affects the Global Hawk and Predator families only. The diameter of dish antennas. radiated power. Design Year Subsystems Weight Multiplication Factor 1. This is largely driven by satellite project funding.4 1. It is assumed that no technology phase applies to this assumed trend.
2 0 1990 1995 2000 2005 2010 2015 2020 2025 Technolodgy Year Figure 414 Satellite Communications Dish Antenna Diameter Trends 4. The weight growth can be substantially higher than this. Nearly every aircraft experiences weight growth in the development process as the aircraft goes from early design to operations.3. Technology Year 1. The typical weight growth from project goahead to first flight is 7% [SAWE 1996].9% weight growth from contract award to the delivery date [SAWE 1996]. The F111 experienced 17% growth from goahead to the end of the first phase research and development. the C5A experienced only 2. respectively.2 Comms Antenna Diameter Ratio 1 0. For convenience.3 Design The only technology attribute in the design technology category is weight growth. Aggressiveness Substantiation Design Weight Growth 4 3 Quality 2 Table 46 Design Technology Attributes Weight Growth The overall vehicle weight growth design technology is assumed to follow the trend presented in Figure 415. The assumed lower and upper bounds from 1930 through 1990 from design year to operations are 3% and 20%.4 0. and 23% growth over the first five years after goahead [SAWE 1996]. the assumed weight growth upper and lower bounds are assumed to change to 1% and 15%. By 2025.6 0. respectively. By comparison.Comms Antenna Diameter Ratio vs. the attributes of this technology are repeated in Table 46.8 0. The technology trend 35 .
Aggressiveness 1 2 3 Substantiation Payloads SAR Payload Small EO/IR Large EO/IR 1 2 2 1 2 2 Table 47 Payload Technology Attributes 36 Quality .assumes that the combined impacts of improving integrated mass properties tools and evolving systems engineering practices containing requirements creep will limit the weight growth encountered during development. the attributes of these technologies are repeated in Table 47. Long Wave Infrared (LWIR). and MidWave Infrared (MWIR) are incorporated. The source of all payload technology trends is the 2000 DOD UAV Roadmap [OSD 2001].3. For convenience.5 Des Lev 1 1950 1970 1990 2010 2030 Design Year Figure 415 Weight Growth Technology 4. Technology trends for Synthetic Aperture Radar (SAR).4 Payloads Only a small subset of potential payload types is considered for this research. Weight Growth Technology 20% Weight Growth to Operations 15% Des Lev 0 10% 5% 0% 1930 Des Lev 0. ElectroOptical (EO).
The high technology level phase lag is zero. Years 6 5 4 3 2 1 0 0 0. The relative performance per weight is used for SAR payload scaling.4 0. Note that the relative performance to weight improves much faster than the relative performance.8 1 Technology Level Figure 416 Payload Technology Phase SAR Payloads The SAR payload technology trends are shown in Figure 417.6 0. Payload Technology Phase vs. SAR Relative Technology vs. because it is possible to develop a payload and a UAV concurrently. Technology Year Relative Technology Metric 10000 Rel Perf/Wt 1000 100 10 1 1980 Rel Perf 1990 2000 2010 2020 2030 Technology Year Figure 417 SAR Payload Technology Trends 37 .2 0.All payloads are assumed to follow the payload technology phase relationship shown in Figure 416. The two trends are the payload relative performance and relative performance to weight. Technology Level Technology Phase.
This assumption may be conservative if the EO/IR payload exhibits similar behavior to the SAR.2 1. respectively. the MWIR data was extrapolated to match the first EO technology date.4 1. and the small EO and small MWIR payload trend were combined for the small EO/IR trend. The EO/IR payload weight per performance is considered constant.3 1.15 1.25 1.1 1. so the technology trends were averaged. The resulting curves for large and small EO/IR payloads are shown in Figure 419 and Figure 420. The large EO and large MWIR payload trends were combined for the large EO/IR trend.ElectroOptical / Infrared (EO/IR) Payloads (A) Figure 418 EO/IR Payloads (B) The DOD UAV Roadmap has performance technology trends for large and small ElectroOptical (EO) payloads and large and small MidWave Infrared (MWIR) payloads.35 1. In both cases. Most EO and Infrared (IR) payloads are combined into single units with an assumed even relative weight and functionality division. Large EO/IR Technology 1. The relative performance metric is the ratio of ground resolved distances. Two examples of EO/IR balls are shown in Figure 418 (AB). of the initial capability over the new capability.05 1 1970 1980 1990 2000 2010 2020 2030 Relative Performance Technology Year Figure 419 Large EO/IR Payload Performance Technology Trend 38 . a measure of imaging performance.
5 Propulsion The basic propulsion technology trends were derived directly from data provided in the DOD UAV Roadmap [2002]. Both jet engines and internal combustion engine trends are included. 4.3.Small EO/IR Technology 8 Relative Performance 7 6 5 4 3 2 1 1990 1995 2000 2005 2010 2015 2020 2025 Technology Year Figure 420 Small EO/IR Payload Performance Technology Trend. Aggressiveness Substantiation Propulsion Jet Propulsion Reciprocating Engines 1 1 2 2 Quality 1 1 Table 48 Propulsion Technology Attributes 39 . The attributes of these technologies are repeated in Table 48.
Jet Engine Technololgy Phase vs.4 0. Technology Level Technology Phase. The jet engine technology phase is based on the assumed time from engine technology development to fielding in an aircraft.6 0. Years 15 10 5 0 0 0. so appropriate technology trends must be applied. The jet engine technology phase and trends are presented in Figure 422 and Figure 423.2 0.8 1 Design Level Figure 422 Jet Engine Technology Phase 40 . respectively. An example of a jet engine intended for UAV application is shown in Figure 421.Jet Engines Figure 421 UAV Jet Engine The HALE family considered here uses jet engines for propulsion.
5 2 1. The internal combustion engine technology phase is based on the time from development of internal combustion technology and fielding in an aircraft. Note that the lowest technology level phase is 50 years. respectively. because many UAV and general aviation engines use internal combustion engines with highly mature technologies.5 3 2.5 1 0. An example of an internal combustion engine intended for UAV applications is shown in Figure 424. Technology Year 4 3.Jet Technology Factors vs.5 0 1980 Technology Factors T/W fact SFC fact 1990 2000 2010 2020 2030 Technology Year Figure 423 Jet Engine Technology Trends Internal Combustion Engines Figure 424 UAV Internal Combustion Engine The tactical and MAE UAV classes considered here utilize internal combustion engines. The internal combustion technology phase and technology trends are presented in Figure 425 and Figure 426. 41 .
5 0 1940 P/W Fact SFC fact 1960 1980 2000 2020 2040 Technology Year Figure 426 Internal Combustion Engine Technology Trends 4.5 2 1. Years 50 40 30 20 10 0 0 0.6 0.4 0. The attributes of these technologies are repeated in Table 49.8 1 Design Level Figure 425 Internal Combustion Engine Technology Phase Internal Combustion Engine Technology Factors vs. Technology Level Technology Phase.5 1 0.6 Structures Two approaches are used for estimating structures technologies . 42 .Internal Comustion Engine Technology Phase vs.3.Factor of Safety (FOS) and direct structural weight reductions. Technology Year Technology Factors 2.2 0.
5 Factor of Safety 2 1. Design Year 2.5 1 0. The 2025 factor of safety for high technology UAVs is assumed to be 1.5 1 1940 1960 1980 2000 2020 2040 Design Year Figure 427 Factor of Safety Technology Trends Structural Weight Factors The second structural technology factor relates to structural weight directly. which predicts aircraft weight reduction due to various technologies from 1995 to 2020. the factor of safety is assumed to be reduced due to improved structural analysis methods and improved manufacturing quality control. A standard UAV minimum factor of safety is assumed to hold constant until 2005. By 2025. Factor of Safety vs.15.5 to 1. For long range transonic transports. the anticipated total gross weight reduction due to structural 43 Aggressiveness Substantiation Quality . This was done to account for the absence of a pilot and the associated safety considerations.25 as late as 1994.5 0 1920 0 0.Structures Structural Factor Factor of Safety 2 3 3 2 1 2 Table 49 Structures Technology Attributes Factor of Safety The factor of safety for UAV structures was reduced from 1. Noor [2000] provides relevant results of Defense and Aerospace [1995] study. Figure 427 shows the factor of safety technology trends. This trend is then assumed to become more aggressive into the future. but for different reasons.
4. The structural technology phase and technology multiplication factors are found in Figure 428 and Figure 429. In order to achieve a 24% reduction in takeoff gross weight due to structural contributions.2. The structural contribution is found by a simple mass fraction analysis similar to that performed on the subsystems. A very simple sizing relationship for a given aircraft class is: WTO 1 = WPayload 1 − MFFuel − MFE . This technology trend is assumed to be applicable to UAVs. respectively. This reduction. The 1960 structural weight multiplication factor is assumed to be 1. − Struct − MFStruct ⋅ FStruct .contributions is estimated at 24%. Technolgy Level 6 Technology Phase.1. when the structural weight multiplication factor is equal to 1. Structural Technology Phase vs. is not the structural weight reduction. The structural weight multiplication factor and factor of safety are applied independently. but the total synthesized aircraft weight reduction. or 10% greater than the 1995 structural weight.Tech Numbers representative of the Boeing 777 were used as an estimate for the 1995 mass fraction numbers.25. since advanced finite element structural design was introduced between these dates.8 1 Technology Level Figure 428 Structural Technology Phase 44 . the structural weight multiplication factor must equal 0. Years 5 4 3 2 1 0 0 0.6 0. The fuel mass fraction is assumed to be 0. The mass fraction of the 1995 structure is assumed to be 0. The mass fraction of the empty weight less structure is assumed to be 0.4 0.2 0.765. however.
6 1950 1960 1970 1980 1990 2000 2010 2020 2030 Design Year Figure 429 Structural Weight Multiplication Factor Technology Trend 45 .2 Structural Weight Multiplication Factor 1 0. Design Year 1.Structural Weight Multiplication Factor vs.8 0.
The selected method employs conic sections applied across multiple fuselage stations. The complex geometry definition takes general and specialized forms. The parametric airfoil generation uses the NACA 4 and 5 series airfoil equations.1 Wings and Tails To streamline the code.2 Geometry Definition The geometry definition is used to define aircraft geometry in detail. horizontal tail. and graphical generators all require specialized geometry evaluations. The goal is to minimize the quantity of data needed to describe a wide range of fuselage geometries. The wing geometry equations presented in Appendix A apply to all three surface types unless specified otherwise. 5. The specialized forms are essentially customized paneling that supports aerodynamics and structures analysis. 5.2. The general form will be discussed first. and vertical tail definitions are combined. since all are aerodynamic surfaces with almost identical geometric properties.1 Geometry Overview Several geometry definitions are required for various analyses within the AV design code.2 Fuselage A robust yet simple fuselage geometry formulation is applied. This approach permits the greatest flexibility. The user selects which two are input. However. The NACA parametric airfoil method is available to all applicable subroutines that require airfoil geometric information. and not the analysis procedures. these parameters are not independent.2. root chord. then the remaining two can be determined through a set of equations. the wing. It is necessary to know the wing span. The airfoil coordinates may be input directly from a designated airfoil coordinate file or generated parametrically. The airfoil coordinate file approach is applied only to the graphics file generator. A parametric geometry model is employed to facilitate the optimization process. The aerodynamics. Conic sections 46 . 5. area and aspect ratio. If any two are known.Chapter 5 Geometry 5. The airfoil coordinate files take the form of the UIUC airfoil coordinate database data files to increase utility. structures/weights.
000 0.414 is an ellipse.414 ……. Three zcoordinate data points are defined for each xcoordinate. ρt(j) rhob 0. the ycoordinate moves out the right wing. Figure 51 shows the Predator nose region. z/Lb. Each fuselage file data contains geometry descriptions in the following format: x/L 0.003 ……. A conic curve may be described by three points and a shape parameter. Composite aircraft often have complex curvature that does not require conic sections.000 …… z/Lm. and 0. because conic sections describe shapes that 2dimensional sheets may attain through simple wrapping. z/Lt. The first is a midline. The EO/IR ball is the sphere under the Predator fuselage in Figure 51. The ρ parameters define the fullness of the fuselage cross section such that the surface between adjacent fuselage segments is a conic section.003 …… y/Lf(j) rhot 0. Note that lines of intersection are not shown in these AutoCADTM grid drawings.414 0. x/Lf(j) z/Lm 0. Figure 51 Predator Fuselage Nose Region 47 .f(j) z/Lb 0. which defines the widest portion of the fuselage for a given section. and the EO/IR ball has a conic parameter of 0. The second is the upper surface of the fuselage. yet the flexibility in geometry description still makes conic sections a useful geometry analysis tool.005 ……. and the zcoordinate is positive in the updirection. ρb(j) The xcoordinate moves from the nose to the tail. The upper conic parameter ρ for the main fuselage is 0.000 0. All x.000 0.003 ……. 1 is a rectangle.35.414 …….414 0. The single ycoordinate data point defines the maximum width of the fuselage. A ρ value of 0 corresponds to a straight line (triangle). and z data are normalized to the fuselage length.000 0.f(j) z/Lt 0. y. and the third is the lower surface of the fuselage.f(j) y/Ls 0.000 0.414. The ρ is defined separately for the upper and lower surfaces.are extremely useful for metal aircraft.
3 for the entire upper surface. The propeller chord and twist distribution geometry used for graphics generation and the detailed propeller analysis models are not linked.25 0. …. this does 48 .25 0.75 for the center fuselage.p is the reference chordwise location for the wing blade twist.05 …. The conic parameter is 0. However.3 Propeller The propeller input is very similar to that of the wing except there are many segments.0 0.08 …. The radial position r/Rp goes from 0 to 1.. The selected conic parameter is 0. The file has the following form: r/R 0.25 0.. The first input is the name of the propeller geometry file. β is the blade twist angle relative to the rotational plane.p(j) All dimensions are normalized to the propeller radius. r/Rp(j) c/R 0. giving a boxy appearance.p(j) x/cβ. 35. β(j) Sweep nsweep ntwist 0. Figure 52 X45A Geometry Figure 53 shows the Dragon Eye fuselage representation.08 0. x/cΛ. Figure 53 Dragon Eye Fuselage Geometry 5.25 … …. …… Λp(j) x/cΛ. x/cβ.Figure 52 shows the X45A geometry as represented by this geometry definition.p is the reference chordwise location for the blade sweep.2. c/Rp(j) Beta 35. 0. 0.
Pusher propellers are applied to the parametric geometry for propellerdriven aircraft. Geometry sizing laws must be established. because both the Shadow 200 and Predator use pusher propellers. 5. optimization requires that geometry components. Figure 54 Propeller Integration on Generic Fuselage 49 . 5. the engine nacelle is located on top of the fuselage such that the rear of the nacelle is located at 95% of the main fuselage length.3 Parametric Geometry The analysis of a known aircraft with a fixed geometry lends itself well to detailed geometry inputs described above. the propeller is centered at 96% of the fuselage length. For example. The only negative outcome is that the graphical output of the chord and twist distribution do not match. respectively. The geometry selected supports both jet and propeller configurations. Simply making the wing geometry and fuselage length design variables does not guarantee the resulting optimized geometry solution is reasonable. tails and bodies. However. the tails and wings should remain attached to the fuselage when so configured.not affect the results because the propeller radius is consistent throughout the code and the detailed propeller design methods are not directly utilized here. such as wings.1 Parametric Fuselage A parametric fuselage formulation is used for all optimization cases considered. for configuration similarity to the Global Hawk. For the propeller case. Figure 54 and Figure 55 show how propellers and jets are applied to the generic fuselage.3. For jets. move together in a sensible manner.
The parametric fuselage permits blending with other bodies. Figure 56 Wing Fairing Geometry Two of the three vehicle classes have radomes to house dish antennas for satellite communications. The wing root leading edge will shift forward relative to the fuselage as wing sweep increases. The maximum conic parameter of the two bodies is applied for samehalf upper or lower surfaces. The mostoutside dimension at a given point for two bodies is selected as the final outer geometry for the resultant body. A 50 . and a single fuselage geometry that represents the radome geometry impacts is not possible. A fairing body with a length twice the wing root chord and centered about the wing is blended with the main fuselage. The resultant blended fairing geometry can be seen in Figure 56. The proportions of the radome relative to the fuselage vary. The wing root leading edge is always defined as the origin. This holds true for upper and lower heights and the width. The wing fairing is one example of fuselage blending. regardless of sweep in order to keep a reasonable tail moment arm for tail sizing.Figure 55 Jet Integration on Generic Fuselage The aerodynamic center of the wing is fixed to 48% of the main fuselage length. 48% appears to be a reasonable number based on visual inspection and from experience with early sizing attempts. The wing root leading edge rests at the bottom of the fuselage at the point of intersection.
parametric radome body is applied. Figure 58 shows the EO/IR ball geometry. avg 1 − x / LAC . as will be discussed shortly. The relative location of the ball is inspired by the Predator configuration and based on visual inspection. Rather than creating a constraint with a penalty function for the optimizer. The diameter of the ball is found through the EO/IR payload calculations. First. the fuselage length is: L fuse = 4 ⋅ cWing . The ball is fixed at 10% of the fuselage length and approximately 20% of the ball height is submerged in the main fuselage. the fuselage minimum length is 35% of the wing span. Figure 57 Radome Impacts on Fuselage Geometry If unconstrained. but not blended with.Wing The first minimum fuselage length protects the fuselage minimum length for high aspect ratio wing cases. and the second minimum protects the fuselage length for low aspect ratio wing cases. the optimizer tends to minimize the fuselage length such that volume constraints are satisfied. the fuselage length is automatically set to the minimum length if the input fuselage length is too low. This raises the possibility of creating an infeasible solution for tail sizing. which is parametrically linked to the dish diameter. Second. the main fuselage. A fuselage minimum length constraint is necessary. Figure 57 shows the radome geometry impact on the fuselage. and the fuselage length must be greater than the greatest of the two. 51 . There are two minimum fuselage length definitions. An EO/IR ball is attached to.
The nacelle length to diameter and overall length relative to the engine is based on visual inspection of appropriate proportions.TE S w ⋅ MAC w ⋅ (sin (ΓV −Tail ) + cos(ΓV −Tail )) = 0 − ACFunc ⋅ SV −Tail ⋅ cos(ΓV −Tail ) ) where ACFunc is the distance from the Vtail root chord aerodynamic center to the trailing edge normalized by the square root of with Vtail area.3. The secant method is employed to solve for tail area such that: SV −Tail − TVCH ⋅ (L AC −VT .TE. First.1 times greater than the jet engine length to diameter ratio. The nacelle length is assumed to be 1. the dihedral angle of the Vtail is found by: TVCV ⋅ bw ΓV −Tail = tan −1 TVC ⋅ MAC H w The length between the wing aerodynamic center and the Vtail trailing edge. A closed form solution to the tail sizing for a fixed tail trailing edge location is not forthcoming. because all three calibration cases use variants this tail configuration.8 times greater than the jet engine length. Conventional tails or other alternative tail configurations could also be applied in a similar fashion. so a convergence approach is used. LACVT. The nacelle length to diameter ratio is assumed to be 1. is found such that the Vtail trailing edge is located at 93% of the fuselage length.2 Parametric Tails The parametric tail geometry uses a tail volume coefficient methodology applied to a Vtail. partially described earlier. Vtails are used here. is another body that is attached to the main fuselage but not blended. 5. 52 .Figure 58 EO/IR Ball Geometry The jet engine nacelle.
This tail sizing approach permits the tails to automatically scale with the wing and tail moment arm. The vertical separation combined with upwardcanted Vtails ensures that the lift distribution methods will not have artificial interference leading to singularities. but sufficient commonality exists for a single geometry definition subroutine to handle all three cases. The second method panels the entire wing by blending segments. reference parameters are defined. and Trefftz plane analysis. The front two faces of the triangles are xyz1 and xyz2. Figure 59 Virginia Tech Forestry UAV 53 . The reference area for the other methods is the wing area of the main wing. respectively.4 Wing Lift Distribution Paneling Three lift distribution methods are employed: vortex lattice method. First. so the reference wing area and span is for each individual wing. The rear triangle points are xyzm. The lifting line theory analyzes wings sequentially rather than simultaneously. which automatically has vertical separation from the wing. Each of the methods require specific geometric paneling. The first method panels each segment individually. The first paneling method was used. There are two primary methods of paneling the wing. The total average chord for the Trefftz plane analysis is the ratio of the first wing planform area to the first wing’s span. Note that the panels on the inboard portion of the main wing are evenly spaced and the outboard segment has sinusoidal spacing. All are described further in the Aerodynamics section. lifting line theory. Further discussion of these methods can be found in Appendix A. 5. The tail root vertical height is fixed to the maximum width location on the fuselage at the attachment location. Figure 59 and Figure 510 show the Virginia Tech Forestry UAV and its wing paneling for lift distribution methods.
Properties at the root and tip of the panel are calculated. Each skin is composed of straight line segments connecting disretized points on the surface.Figure 510 Wing Aerodynamic Paneling for Virginia Tech Forestry UAV 5. The coordinate system used here is the same as the aircraft coordinates. The length of each element is found by: Lelt (i ) = (X pt (i ) − X pt (i + 1) ) + (Z pt (i ) − Z pt (i + 1) ) 2 2 The midpoint of each element is found by: X mid (i ) = X pt (i ) + X pt (i + 1) 2 Z pt (i ) + Z pt (i + 1) 2 Z mid (i ) = The angle of the element from the xaxis in the xz plane is found by: 54 . The wing box for each panel begins after a specified leading edge chordwise distance and ends at the control surfaces. if any are present. The properties under consideration are described below. The structural contours for the skins are then determined using the NACA airfoil generator. The NACA airfoil coordinate generator calculates the 2D coordinates of the upper and lower surface of the skin.5 Wing Structures Paneling The code shares many wing paneling methods for aerodynamics and structures. and the average value is used for the panel calculations. The wing is broken up into multiple discrete panels.
The wing box centroid is found by: nelt ∑ X mid (i ) ⋅ Lelt (i ) = i =1 LSurf . Bottom 55 . Bottom The centroid calculations neglect wing thickness because the model assumes that the wing skin thickness is uniform.t = ∑ elt i =1 12 ⋅ Sin(θ elt (i ) ) ⋅ Cos (θ elt (i ) ) + Lelt (i ) ⋅ ∆X cent (i ) ⋅ ∆Z cent (i ) Top . Bottom nelt L (i )3 I XZ .tot = ∑ Lelt (i ) i =1 Top .θ elt (i) = Tan −1 Z pt (i + 1) − Z pt (i ) X (i + 1) − X (i) pt pt The total length of the wing box surface excluding the front and back walls is found by: nelt LSurf . Bottom X cent Z cent The X and Z distance from each point is found by: ∆X cent (i ) = X cent − X mid (i ) ∆Z cent (i ) = Z cent − Z mid (i ) The area moments of inertia divided by the skin thickness are found by: nelt L (i )3 2 2 I XX . Bottom nelt ∑ Z mid (i ) ⋅ Lelt (i ) = i =1 LSurf .tot Top .t = ∑ elt 12 ⋅ Sin(θ elt (i ) ) + Lelt (i ) ⋅ ∆Z cent (i ) Top .tot Top . Bottom i =1 nelt L (i )3 2 2 I ZZ .t = ∑ elt i =1 12 ⋅ Cos (θ elt (i ) ) + Lelt (i ) ⋅ ∆X cent (i ) Top .
Shear web contributions are neglected for the purposes of the calculations. The shear flow divided by thickness is found by: nelt qShear . Root (k ) ) + (Ycent .Tip (k ) − X Cent .Tip (k ) − YCent . Root (k ) Γcent (k ) = Tan −1 cent . the average panel characteristics must be calculated. The structural sweep of the panel is found by: X (k ) − X cent .t = ∑ Lelt (i ) ⋅ ∆Z (i ) i =1 Top The above calculations were performed on both the root and tip sections of the wing panel.Tip (k ) + LSurf .Tip (k ) − YCent . k is the index for the structural panels.Tip ( k ) − Z cent . Next. For the geometry calculations. The shear web is not included. 56 . The coordinates of the panel centroid are simply the average of the root and tip coordinates. The skin thickness corrected shear flow is the average or the root and tip values. Root ( k ) The structural span of the panel is found by: bStruct (k ) = (X cent .Tip Z cent .The shear flow at the centroid divided by thickness considers only the contribution of the wing skins. Root (k ) Λ cent (k ) = Tan −1 cent . Root (k ) ) + (Ycent . Root (k ) ) 2 2 2 The total surface area of the skins for each panel is found by: S Pan (k ) = (L Surf . Root ( k ) The centroid dihedral of the panel is found by: X (k ) − X cent .Tip ( k ) − Ycent . Root (k ) ) 2 ⋅ bStruct (k ) The results of the structural paneling are shown in Figure 511 and Figure 512. The skin thickness corrected area moments of inertia for the panel is the average of the root and tip values. The total thickness of the panel wing box is the average of the root and tip thicknesses.Tip Y cent .
The graphics file generator has the ability to evaluate a wide variety of configurations. 57 . Figure 513 shows one unconventional configuration. The graphics generating capabilities are inspired by the work of Libeau [1999].6 Graphics File Generator The graphics file generator outputs data exchange format (DXF) drawing files that can be accessed by Computer Aided Drawing (CAD) software. Not all of this functionality is utilized in this research. The purpose of the graphics file generator is to provide nearly photorealistic 3D renderings of the designs. The graphical outputs provide a means of rapidly evaluating the configurations for fundamental flaws. where tables and plots become cumbersome. The graphics also provide an effective means of communicating the behavior of the optimization results.Figure 511 Vehicle Structural Paneling Example Figure 512 Wing Tip Paneling Detail 5. The graphics file generator automatically outputs the geometry of the vehicle of interest.
Typically.Figure 513 Example Unconventional Configuration The wing. This is necessary to graphically represent tiltwing V/STOL configurations as shown in Figure 514. nacelle. any new aircraft concept can be graphically represented within 2 hours. Figure 514 TiltWing Graphics Option The rotation feature and other capabilities facilitate rapid drawings of new concepts. and propeller rotation options allow the listed components to be rotated such that the incidence is modified about some point Prot. 58 .
Early conceptual design work does not require the level of detail necessary in early preliminary design. and consists of the tire.1 Aerodynamics Overview The aerodynamics code models a large variety of aircraft shapes in the subsonic and low transonic flight regimes over a wide Reynolds number range. The methods also vary in robustness and computational efficiency. and it does not require exact definition of characteristics such as wing twist. Early conceptual design focuses on simple sizing. The compressible correction is zero until the critical Mach number is reached. Both simple and complex methods are employed with varying degrees of accuracy. depending on the analysis objectives. The problem is decomposed into an incompressible solution with an additive compressible correction. Lift is not corrected for compressibility. Gear drag is calculated in the extended and retracted positions. and miscellaneous drag.Chapter 6 Aerodynamics 6. therefore.000. with no compressibility effects on lift or drag. The profile drag is the additive combination of the friction and pressure drag on aerodynamic surfaces such as wings and tails. The various approaches permit different types of questions to be answered. fuselage drag. The aerodynamics code utilizes a number of methodologies to evaluate the aerodynamic characteristics of relatively arbitrary vehicles. door. profile drag. landing gear drag. The primary product of the aerodynamics code is a drag polar for use in the mission analysis subroutine.000 to very large numbers on the order of 20. All methods must operate in Reynolds numbers ranging from less than 100. thickness and camber distribution or tail incidence. as appropriate. Induced drag contains both traditional drag due to lift and some trim drag. interference drag. Early preliminary design generates much more detailed information and. Interference drag consists of drag at intersections of the wings and bodies. This research focuses on both conceptual and early preliminary design aerodynamics methods. computational efficiency. after which the drag rises with Mach number according to the Korn equation [Grasmeyer 1998]. strut. and drag reductions due to reduction of wetted areas at the intersections. The incompressible drag is composed of induced drag. 59 .000 and Mach number between zero and the low transonic region where the drag rise begins. must model the impacts of detailed inputs. and flexibility. and cavity contributions. The incompressible solution is assumed to be equivalent to Mach zero at a given Reynolds number. There is no consideration for MachReynolds number interactions. which reduces the number of aerodynamics tables that must be generated. aerodynamic surface induced drag.
Re) CD(α. Aerodynamics interacts with propulsion and other systems. Product CL(α. and induced drag. Propulsion interactions are captured as drag penalties in the propulsion section. The lift methods are quite detailed to provide detailed lift distribution information to determine local stall. trim requirements. Propulsion effects on lift and pitching moment are neglected. though these influences will be described in the aerodynamics chapter. induced drag can also include profile drag due to lift and profile trim drag. The inputs for the aerodynamics code include vehicle geometry and paneling.Re) CDw(α. The lifting surface profile drag is another significant drag driver that is given much attention. Cooling drag is captured as a drag penalty in the propulsion analysis during mission execution. The primary products are shown in Table 61. The MISSPARAM subroutine estimates the minimum and maximum flight Reynolds number 60 . drag polar array size parameters. and technology data. method selections. flight regime limitations. 6. The primary output is a drag polar and an additive compressibility correction array. usable total lift coefficient. Subsequent sections will discuss the methodologies used in the subroutines in more detail. The drag polar consists of angle of attack and Reynolds number dependent lift and drag arrays and a suitability array.M) Description Total lift coefficient array Total drag coefficient array Permissible data point flag array Total wave drag coefficient array Table 61 Primary aerodynamics product list.The various methodologies book keep drag in different ways.Re) Allow (α. aerodynamics parameters. and a drag rise array corresponding with the Mach array. Much information is required prior to running the AEROC aerodynamics analysis routine. where these components are subtracted from the profile drag. Airfoil data input and consderable geometry definition occurs beforehand. The suitability array determines if the drag array element is usable by considering elevator deflection and local stall constraints. The philosophy adopted for this research is to place the greatest emphasis on the largest contributors to the aerodynamic behavior of interest. For example. The compressibility correction array consists of a Mach array starting with the critical Mach number.2 Aerodynamics Code Architecture This section describes the aerodynamics code architecture and its components at a high level.
If either the VLM or LLT are selected. which then outputs the pitching moment about the center of gravity. Once trimmed. The positive and negative stall angles of attack for all panels are found in subroutine CLMINMAX. respectively. and the lift distribution. which was described in the Geometry section. then the lift coefficient is input. The retracted and extended landing gear drag contributions are calculated in subroutine GEARDRAG. the profile drag for each panel is calculated by one of three methods in CDPCALC. The CGLOC subroutine estimates the aerodynamic center based on the geometry and locates the center of gravity based on the aerodynamic center and static margin. The zero lift angle of attack for the panels is determined in subroutine AIRFOILI. An elevator position and angle of attack. The first step is to generate the final paneling used for the lift distribution methods. Once the inputs are properly conditioned. the AEROC subroutine is called by EVAL. are sent to subroutine CM. If the vortex lattice method is used. among other inputs. then the elevator position to trim the vehicle at a given angle of attack is determined through the secant method by setting the pitching moment about the center of gravity equal to zero. Figure 61 AEROC code Architecture Figure 61 shows the AEROC code architecture. The interference drag is calculated in subroutine INTDRAG. The wave drag is calculated for a given lift distribution and Mach number in WAVEDRAG. then the velocity influence coefficients are calculated up front with subroutine VLMUVW and INFLVPM. The fuselage drag is calculated in subroutine FUSEDRAG. 61 . induced drag. If the Trefftz plane analysis is selected. and angle of attack are output.per unit length based on the flight envelope boundaries.
as described in the lift methodology section of this chapter. CM. is called. 6. The airfoil 2D pitching moment subroutine. These pitching moments are used to help determine how to trim the aircraft. CMPROF. Finally. The pitching moment caused by the lift moment arm about the center of gravity is calculated in subroutine CMLIFT. The pitching moment due to flap deflections is determined in subroutine CMFLAP.3. at which local stall first occurs and induced drag.3 Lift Methodology 6. First.1 Lift Methodology Overview The lift distribution of a configuration must be determined to evaluate the total lift coefficient. The lift distribution information is also necessary for determination of profile drag and loads for structural analysis. FLAPETA. flap efficiency parameters are calculated in subroutine.2 CG Location Determination 62 . either the LLT or VLM subroutines are called to determine the lift distribution as a function of angle of attack and elevator deflection. CL. Then.3. 6.Figure 62 Pitching Moment Subroutine Architecture Figure 62 shows the architecture of the pitching moment subroutine. the fuselage pitching moment is calculated in subroutine CMFUSE.
The Trefftz plane analysis calculates the minimum induced drag trimmed lift distribution for a given planform geometry without determining how the distribution was generated. wing + (hn − 0. The first two methods take a detailed geometry definition. and tail volume coefficients. Prandtl Lifting Line Theory 2. vortex lattice method (VLM). as shown in Table 62. The fuselage is broken into a number of horizontal strips perpendicular to the Xaxis. including camber effects (α0L shift and elevator deflections) and washout.The center of gravity location must be known in order to estimate the trimmed lift distribution. The origin of this coordinate system is the front of the main wing root chord. The end equation for the Xcoordinate of the center of gravity is: X CG = X NP − sm ⋅ c where the neutral point for the aircraft is defined as: X NP = X AC . The derivative of the fuselage pitching moment with respect to angle of attack is found using methods described in Nelson [1989]. Trefftz Plane Analysis Applied Here No No Yes Table 62 Lift Distribution Methods 63 . wing mean aerodynamic chord. The aircraft neutral point is found using standard methods found in Etkin [1996] and other stability and control texts. Method 1.3 Lift Distribution Methods Three lift distribution methods are employed in this code: Prandtl lifting line theory (LLT). 6. The elevator deflections are adjusted at a given angle of attack to trim the vehicle. Vortex Lattice Method 3. and X is positive moving towards the rear of the aircraft. The key inputs to the center of gravity location are the vehicle geometry and the static margin.3. The Trefftz plane analysis was applied to generate all results for this research.25) ⋅ c The wing aerodynamic center is found though standard multisegmented trapezoidal wing methods. and a Trefftz plane analysis. The geometry definition subroutines calculate the Xcoordinates of the aerodynamic center of all lifting surfaces.
In the method used.6. j )][x(i)] = [b(i)] where n = 2 ⋅ j −1 π ⋅ c(i ) ⋅ n A(i.3. c. This method allows varied chord.1 Prandtl Lifting Line Theory The wing lift distribution is determined by using Prandtl's line theory as presented in Davenport [1998]. The following matrix formula is solved: [A(i. j )]−1 [b(i)] The total lift coefficient for the wing is: CL = AR ⋅ π ⋅ x(1) and the induced drag is 64 . The method provided in Bertin and Smith [1989] describes only a single taper ratio directly. is found through linear interpolation across the semispan. Note that in effect the integer 2N1 replaces : in the above equation.3. sweep and dihedral cannot be accounted for and it is assumed that the wing is tapered about the quarter chord. odd ∑ ∞ where θ = cos −1 (− y / s) υ is divided up into N segments between 0 and π/2 radians. and n assumes odd integer values from 1 to 2N1. camber and twist distributions along the span. The matrix formula takes the form: πc 4s (α + α twist − α 0 L )sin θ = π ⋅ c ⋅ n + sin θ An sin( nθ ) 4s n =1. The chord at the given υ. j ) = sin (n ⋅θ (i ) ) ⋅ + sin (θ (i ) ) 2⋅b w b(i ) = π ⋅ c(i ) 2 ⋅ bw ⋅ (α + α twist (i ) − α 0 L (i ) ) ⋅ sin (θ (i ) ) The coefficients x(i) are determined by the matrix operation: [x(i)] = [A(i.
and the tail contribution to induced drag is underpredicted. and wm) induced by the circulation by ensuring that there is no flow passing through each control point m.3. though this is not explored further here. The LLT can only analyze a single wing at a time. spanwise and vertical velocities (um. because the upstream wing has trailing vortices. there is no possibility for tail calculations to fail due to singularities induced by the other surfaces. The inherent limitations of the LLT are that the tail effectiveness is overpredicted. The downstream surface is always more affected by the upstream surface than vice versa.odd ∑ n ⋅ (x(n) / x(l )) 2 8 ⋅ s 2 N −1 ⋅ ∑ x(n) ⋅ sin (θ (i ) ) c n =1. Configurations with multiple surfaces can be analyzed simultaneously. Although a selectable option in the code. only one panel per chord length is used. the LLT method was not used to generate the results for this research. The 3D formulation determines the streamwise. vm. Sweep and dihedral can now be modeled.3. Because trailing vortices from upstream wings are not considered. so downstream downwash effects on tails can not be evaluated directly.C Di = where e= 1+ The section lift coefficients at υ are: C l (i ) = CL π ⋅ AR ⋅ e 2 1 2 N −1 n =3. 6. As in the example provided in Bertin and Smith [1989].odd The tip lift coefficient is always zero.2 Vortex Lattice Method The vortex lattice method (VLM) offers the ability to evaluate more complex configurations than the LLT. which creates inaccuracies. A wing downwash model that creates a new tail effective local angle of attack distribution would improve the tail lift predictions. The tail is modeled as a new wing without an upstream wing. This can be shown by: − u m ⋅ sin(δ ) ⋅ cos(φ ) − v m ⋅ cos(δ ) ⋅ sin(φ ) + wm ⋅ cos(φ ) ⋅ cos(δ ) = −U ∞ ⋅ sin(α − δ ) ⋅ cos(φ ) 65 . The 2D and 3D vortex lattice methods of Bertin and Smith [1989] are used directly in the code. so υ = 0 is not included in the A matrix.
The 2D formulation increases the risk of panel interference. δ. The 2D formulation neglects dihedral effects and vertical separation of panels. but a direct relationship between δ and α0L cannot be established for both conventional and reflexed airfoils. the velocities are made nondimensional by U: and δ is replaced by α0Lαtwist. The local lift coefficient is found by: 66 . α0Lαtwist. The parametric geometry model used for optimization automatically ensures that these singularities do not occur. The elements of the 2D and 3D A(i. is replaced by the zero lift angle of attack minus local incidence. j )][x(i)] = [b(i)] . This yields some inaccuracies. so it is important to ensure sufficient vertical separation between tandem aerodynamic surfaces. because the airfoil geometry is not known for many cases. The 2D equation becomes: wm ⋅ cos(δ ) = −U ∞ ⋅ sin(α − δ ) As with the 3D case. j )]−1 [b(i)] The 3D code is likely to yield unrealistic results if the trailing vortex from an upstream panel approaches a downstream panel control point. Additionally. This is accomplished through surface vertical separation. By eliminating a dimension. and they will not be reproduced here. Also. The user must control this through inspection for arbitrary configurations. The right hand side of the equation was modified further to account for the fact that the angle of attack has less direct influence on the lift of an airfoil as the local dihedral approaches vertical. The matrix represents the influence of the bound and trailing vortices of all panels on all control points.The angle of the control point at zero angle of attack. U: is not necessary to determine the lift coefficient distribution. the right hand side of the equation is modified to: b(i ) = − sin (a ⋅ cos(φ (i ) + α twist (i ) − α 0 L (i ) )) The x(i) coefficients are determined by the matrix operation: [x(i)] = [A(i. Using the matrix form: [A(i. since a change of δ in the term (αδ) should have the same effect on local lift as a change in α0Lαtwist. δ and α0L are intuitively interchangeable on the right hand side of the matrix equation.j) matrix are presented in Bertin and Smith [1989]. the problem is greatly simplified and computation time is shortened. so all velocities are normalized by U:.
The equivalent nondimensional velocity in the Trefftz plane is determined by the matrix b(i).C l (i ) = The total vehicle lift coefficient is found by: CL = ∑ i =1 N 2 ⋅ x(i ) c(i ) 2 ⋅ C l (i ) ⋅ c(i ) ⋅ ∆y (i ) ⋅ cos(φ (i )) Sw where the factor 2 accounts for ∆y being summed over only one semispan by assuming symmetry in the lift distribution. The induced downwash influence coefficient matrix Ai(i. The downwash on the bound vortex is half that of the associated downwash infinitely downstream. j ) ⋅ b( j ) 2 The local induced drag. is found by: C di (i ) = C l (i ) ⋅ ε (i ) The total induced drag is found by: C Di = ∑ i =1 N 2 ⋅ C di (i ) ⋅ c(i ) ⋅ ∆y (i ) Sw As before. 67 . This is similar to the approach taken in the Trefftz plane analysis. Cdi(i). The Trefftz plane can be thought of as the conditions associated with the ideal flow at a location infinitely downstream of the lifting surfaces.j) matrix inverse does not need to be evaluated in this operation. but it was not used to generate the results for the research. [Ai (i. since x(i) for this system was previously determined.j) is determined for a position 100 main span lengths downstream of the origin. which is described later. is found by: ε (i ) = tan −1 ∑ j =1 N Ai (i. the factor of 2 represents symmetry. The downwash angle. The induced drag is found by evaluating the downwash in the Trefftz plane and then relating it back to the downwash at the wing surface. The VLM analysis is available to the user. ε. j )][x(i)] = [b(i)] Note that the Ai(i.
The means of determining the trim requirements are handled by the lift distribution methods. There are. Methods like LLT and VLM become more beneficial when further vehicle definition becomes available. Grasmeyer's general methodology is employed here. This makes the Trefftz plane approach ideal for conceptual design. airfoil zerolift pitching moments. Flap + CM . In other words. The trimmed condition occurs only when the summation of the pitching moments about the vehicle center of gravity is equal to zero.0 L + CM . Fuse. several disadvantages. no angle of attack information is provided to or generated by this method directly. Third. CM = CM . However. no iterations of elevator deflections are necessary.3. The trimmed flight condition must be known for performance estimation. however. Second. The primary purpose of the aerodynamics model is to determine the trimmed drag polar for use in the fuel burn calculations in the performance model. the twist and camber distribution (if it were determined) for one lift condition may not provide the optimal lift distribution at a second condition. Unlike LLT and VLM. and pitching moments due to fuselages and nacelles. Grasmeyer [1998] discusses the details of this method in detail. or elevatoraffected lift distribution on a flying wing. Despite its limitations. The pitching moment contributors are the lifting moments. the method is optimistic by assuming that a single vehicle can generate optimal lift distributions at more than one lifting condition without variable geometry. counteract all other contributions to attain the trimmed condition.3 Trefftz Plane Analysis The Trefftz plane analysis offers the advantage of automatically determining the optimal trimmed lift distribution for minimum induced drag at a given CL. 6. the method does not determine how this optimal lift distribution is attained or if it is even possible to achieve.4 Pitching Moment Both the LLT and VLM provide lift distributions and induced drag estimations for a given configuration at a given angle of attack and elevator deflection. First. flap pitching moments. there is no guarantee that the configuration is trimmed at the given condition.3. Nac The pitching moment due to lift for each panel is determined by: 68 .3.CL + CM . The Trefftz plane analysis is used to generate all results of this research. the Trefftz plane analysis is ideal for quickly analyzing and assessing the optimized results of a lifting system with little definition.6. Note that the empennage lifting moments for a conventional configuration.
flap ∂Cl (i ) ⋅ ∆Cl . Due to the order in which the aerodynamics subroutines are called.40 ⋅ flap (i ) − 0.CL = ∑ 2 ⋅ CL (i ) ⋅ i =1 N c(i ) ⋅ ∆y (i ) xCG − xc / 4 (i ) ⋅ ⋅ cos(φ (i ) ) Sw macW The zerolift pitching moment of the airfoil sections is determined by: C M .CM . The reference line used to determine fuselage angle of attack is the zeroincidence fuselage reference. The zerolift fuselage pitching moment estimation approach is the element method used by Nelson [1989]. fuse ⋅ α ∂α Fuselages ∑ The derivative of the fuselage pitching moment with respect to angle of attack was discussed previously.259 + 0. the aircraft zero lift angle of attack is unknown when this fuselage pitching moment contributor is invoked. fuse = ∂C CM 0. flap (i ) = 2 ⋅ π ⋅ δ f (i ) ⋅ η f ⋅ τ (i ) where ηf is found through interpolation of data found in McCormick [1995].0 L = ∑ 2 ⋅ C M . fuse + M . flap ∂CL c c (i ) = −0. The pitching moment is: CM . The flap/elevator pitching moment increment is estimated by: ∂CM . and τ is determined by an empirical equation from the same source.0 L (i ) ⋅ i =1 N c(i ) ⋅ ∆y (i ) ⋅ cos(φ (i ) ) Sw The flap pitching moment is estimated by a series of empirical equations developed from data of several flap types found in McCormick [1995].141 ⋅ flap (i ) c c 2 The flap/elevator 2D lift increment due to deflection is: ∆Cl . Note that fuselage and nacelle pitching moment contribution methods are treated identically in this code. so the typical fuselage and nacelle subscript is covered by the fuselage subscript alone. 69 . so the zero lift angle of attack is assumed to be zero. The profile pitching moment contribution due to flap deflection is: CM . flap (i ) ⋅ c(i ) ⋅ ∆y (i ) ⋅ cos(φ (i ) ) Sw The pitching moment due to fuselages and nacelles is determined by methods found in Nelson [1989]. flap = ∑ 2 ⋅ i =1 N ∂CM .
the solution is infeasible. A more suitable model is as follows: 70 . The above formulation works reasonably well for aircraft with allturbulent or fixed transition boundary layers. the only means of trimming the vehicle is through flap/ elevator deflections. and an error flag is set to indicate the condition.4 Drag Methodology 6. No poststall evaluations are performed. which is usually held constant at the average Reynolds number. No flap pitching moments are included. Similarly. The Trefftz plane analysis requires a pitching moment input prior to calculating the trimmed lift distribution. e. but it may vary with geometry. and the trimmed lift distribution solution ensures that the total pitching moment about the center of gravity is zero. The span efficiency factor. so the maximum lift coefficient occurs when the first wing panel is stalled. Though deflections create profile pitching moments directly. if the local Cl is greater than Cl. This creates a pitching moment due to lift.max. Note that wave drag contributions are not under consideration for the purposes of this discussion. The zerolift profile pitching moment and fuselage pitching moment coefficients are input into the method. Qseg is the interference factor. This formulation requires that all liftdependent drag sources are included in the second term on the right hand side of the first equation. and Sseg is the wetted area of the body or aerodynamic surface of interest. CD0 is also considered to be independent of lift. The elevator deflection that creates a zero pitching moment about the CG is found via the secant method.4.1 Drag Methodology Overview A traditional conceptual design wing drag polar equation takes the following parabolic form: 2 CL CD = CD 0 + π ⋅ AR ⋅ e where CD 0 = 1 S ref ∑C d 0 seg ⋅ FFseg ⋅ Qseg ⋅ S seg Recall that FFseg is the shape form factor. but it is not suitable for many practical applications. is usually considered to be independent of lift. CD0 accounts for the profile drag. 6.For a given angle of attack. If the required elevator deflection falls outside the permissible bounds. an error flag is set to indicate the local stall condition. their greatest contribution comes from redistributing the lift on the surfaces.
Initially. low and high Reynolds numbers.000. because it is a major drag contributor. the parabolic nature collapses.CD = CDprof + where C Dprof = 1 S ref 2 CL π ⋅ AR ⋅ e ∑C d . The pressure drag coefficient is also a function of lift coefficient and Reynolds number.000). Figure 63 shows data for the NACA 0009 airfoil at very low. at a Reynolds number of 60. the drag polar behavior becomes erratic and no suitable formulation could be derived.min drag ) + b ⋅ (C l − C l . an Excel spreadsheet was developed to help formulate a series of parametric drag polar functions. seg ⋅ Qseg ⋅ S seg and Cd . prof = Cd . This equation worked well for high Reynolds numbers (Re>3.000. Finally. The transition location strongly affects the friction drag. and n are found from regression analysis and may be dependent upon Reynolds number. The upper and lower transition locations are a strong function of lift coefficient and Reynolds number. press The profile drag is readily obtainable from airfoil wind tunnel test results. At low Reynolds numbers (Re<300.4. 71 . especially at low Reynolds numbers. Note that the parabolic features of the drag polar become much more pronounced as the Reynolds number decreases. Significant attention is given to profile drag. prof . At very low Reynolds numbers (Re<100. t / c ) + ∆C d 0 + a ⋅ (C l − C l . Using a realistic profile drag will yield an aircraft drag polar that generally does not have a parabolic form at all Reynolds numbers. camber and thickness.2 Profile Drag Overview The profile drag is composed of both friction and pressure drag contributions. 6. no simple formulation with acceptable accuracy is possible for most airfoils. a. The NACA 0009 is a very predictable airfoil compared to most that were investigated. frict + Cd . The data is based on Abbot and Von Doenhoff [1959] and the UIUC low Reynolds number wind tunnel tests [Selig 1995]. The desired equation would take the following form: C d = C d 0 (Re.000). b.min drag ) 2 n where equations for ∆Cd0.000).
5 CL 1 1. the SeligDonovan 7032.5 0 0. This airfoil. The subroutine must work for all Reynolds numbers 72 . it became clear that a different approach was necessary.03 CD Re=100K Re=200K Re=300K 0.000. SD 7032 Wind Tunnel Data 0.05 0. is relatively predictable at Reynolds numbers above 200.02 Re=3M Re=9M 0.02 Re=300K 0. Selig Donnovan SD7032 at Low Reynolds Numbers After expending considerable effort pursuing a parametric function.5 CL 1 1. The data is based on the UIUC low Reynolds number wind tunnel tests [Selig 1995].5 Figure 64.01 0 0 0.NACA 0009 Wind Tunnel Data 0.04 0.01 0 0.04 Re=60K 0.5 Figure 63 NACA 0009 at Several Reynolds Numbers Figure 64 is more representative of most airfoils in the lower Reynolds number range.03 CD Re=60K Re=100K Re=200K 0. The shape of the drag polar at lower Reynolds numbers differs greatly between airfoils.05 0.
Ideally. camber and 73 . CMc/4 camber t/c Figure 65 Airfoil 3D Interpolation Space Traditional multidimensional interpolation would require all but one of the independent parameters to remain constant while the dependent variable is found for the nonconstant independent parameter. camber and pitching moment. Profile Drag Method 1. Parametric Airfoil Method 3. Assume that. The data is interpolated for a fixed Reynolds number and incompressible flow.1 Airfoil Interpolation Unfortunately. The parametric airfoil method applies parametric relationships for the pressure drag and friction drag of several airfoils. The airfoil interpolation and extrapolation methods can be applied to the direct airfoil data method and the parametric airfoil method. The direct airfoil data method interpolates from databases of airfoil data. Compressibility effects are described separately. the drag coefficient must be determined for an airfoil of a certain thickness. Figure 65 shows a graphical representation of the problem. Direct Airfoil Data Method 2. Finally.encountered by several classes of UAV. Three methods of determining the profile drag were developed for this research.2. the simplified airfoil method uses simple relationships to relate profile drag to equivalent flat plate skin friction drag. aerodynamic data is not always available for an airfoil with the desired thickness.4. characteristics of the arbitrary airfoil should be bounded by those of the reference airfoils. Sacrificing fidelity at small UAV Reynolds numbers is not acceptable. Simplified Airfoil Method Applied Here No Yes No Table 63 Profile Drag Methods 6. A method was developed to determine the drag polar characteristics for an arbitrary airfoil given data from a variety of other airfoils. as shown in Table 63. for a given Reynolds number and lift coefficient.
i C m . ref . The zerolift angle of attack. First. The goal of the methodology applied here is to find a way of interpolating between relatively few reference airfoils. fact ⋅ ( cambmax C m . and their values for the target airfoil must be bounded by the reference airfoils surrounding it. Assuming regular parameter increments. For the above example.25. then the factor for all four is 0. An interpolation in one parameter requires two reference data points. For a given airfoil drag polar.i ) z i = C m. The volume of airfoil data required to perform such an interpolation for all Reynolds numbers of interest would be difficult to obtain within the scope of this research.i ) t / c max (cambt arg et − cambref . The sum of the factors is always equal to unity. The last three characteristics are independent parameters.pitching moment. If the radius between a reference and target airfoil is zero. The target airfoil maximum lift coefficient is the sum of the products of the factors and reference airfoil maximum lift 74 . Thus. and the target airfoil is equidistant from all. a reference 'radius' is determined between the characteristics of the target airfoil and those of the reference airfoils. the airfoil characteristics are transformed into weighted lengths as follows: xi = t / c fact ⋅ y i = camb fact ⋅ (t / c t arg et − t / c ref . The factor for this reference airfoil is equal to 1 and all others are zero. then the two airfoils are identical. maximum lift coefficient and drag coefficient for each airfoil are found via the airfoil lookup function. the drag coefficient for that airfoil is found by interpolating first by lift coefficient (for two bounding Reynolds numbers) to get two drag coefficients. t arg et − C m.max ) These lengths are then converted into equivalent radii: ri = xi2 + y i2 + z i2 The algorithm takes a series of sequential weighted factors and penalties based on the equivalent radii and the number of preceding cases. a twoparameter interpolation would require four data points. the number of required reference airfoils is 2n. These two drag values are then interpolated between the two bounding Reynolds numbers. Assume that there are four airfoils. Using regular parameter increments. where n is the number of independent variables. there must be exactly eight reference airfoils for three independent parameters if regular parameter increments are assumed.
the initial weighting factors and initial penalties for the remaining airfoils are assessed: fact beg (i ) = 1 − (i − 1) ⋅ r (i ) (i − 1) ⋅ r (i ) + req (i − 1) penbeg (i − 1) = 1 − req (i − 1) (i − 1) ⋅ r (i ) + req (i − 1) Where req(i) is the smallest equivalent radius up to reference airfoil i. For this case.coefficients. the factor for polar 1 and 2 is about 0. First. Figure 66 shows the interpolation plane. The lookup lift coefficient is scaled from each reference/target airfoil maximum lift coefficient ratio.25 and 0. the factor for each airfoil is: 75 . the initial weighting factors for airfoils 1 and 2 relative to one another are assessed: fact beg (1) = 1 − r (1) r (2) . This procedure may be performed for any number of reference airfoils greater than 1. Next. the factor penalty is assessed: factpen(i ) = ∏ penbeg ( j ) j =2 i Finally. respectively.75. fact beg (2) = 1 − r (1) + r (2) r (1) + r (2) Next. Clmax 2 Polar 2 Polar Target CL Polar 1 Clmax 1 CD Figure 66 Drag Polar Interpolation Diagram for Two Airfoils The procedure for determining the interpolation factors is relatively straightforward. The target airfoil drag coefficient at a given lift coefficient is the sum of the products of the factors and reference airfoil drag coefficients.
fact (i ) = factpen(i ) ⋅ fact beg (i ) The summation of all the factors always equals unity. Other data is obtained from offline XFoil [Drela 2000] airfoil design and analysis software outputs. 6. Again.2. The extrapolation formula is as follows: C d = C d (α = α min dat ) + [C d (α = −α ) − C d (α = −α min dat )] 76 . camber and pitching moment. The drag bucket is not assumed to be symmetric about the α=0o axis.neg = 2 ⋅ C l α = 0 o − C l max ( ) and the prestall extrapolated lift coefficient is: C l = C l (α = α min dat ) − [C l (α = −α ) − C l (α = −α min dat )] where αmindat is less than 0o and may correspond to a positive lift coefficient. Assuming that the behavior of the lift curve is symmetrical as well. which is described later. the stall angle is assumed to be symmetrical.4. which merely shifts the lift curve up and down but does not impact the stall angles. The predicted transition locations and pressure drag terms exhibit no undesirable behavior under extrapolated conditions. The parametric airfoil interpolation subroutine permits lift and drag extrapolation at negative lift coefficients without modification. The directdata airfoil interpolation subroutine presents more challenges than the parametric case. with the change in slope opposite in sign on either side of the axis. It has been determined from analysis of wind tunnel data that the negative stall occurs at an angle of attack equal in magnitude and opposite in sign to that of the positive stall. but the minimum drag coefficient is assumed to lie near α=0o. Much of this data comes from UIUC lowspeed wind tunnel test databases [Selig 1989 & 1995]. This effect is independent of camber. However. The current approach is to use data for airfoils designed to operate at a given Reynolds number rather than use a single airfoil over the entire Reynolds number range. These airfoils span the range of thickness. the maximum negative lift coefficient is: C l max. most airfoils considered have a substantial nonstalled lifting region at negative lift coefficients for which no data is available. The stall angle of attack is assumed to be symmetric about the α=0o axis. The airfoil database has a small population of airfoils.2 Airfoil Extrapolation The airfoil interpolation subroutines focus on positive lift coefficients below stall. This assumption is quite suitable in practice.
PostStall Lift Behavior 1. Deg 100 200 300 400 (A) (B) Figure 68 Curve Fit for 360Degree Angle of Attack Sweep 77 . Degrees 90 180 270 360 Drag Coefficient 2 PostStall Drag Be hav ior 1.dat α CD (A) (B) Figure 67 Angle of Attack Symmetry Diagram A report by Sheldahl and Klimas [1981] provides the lift.5 Angle of Attack.5 1 1. and pitching moment data of symmetric airfoil sections through 0 < α < 180o at low Reynolds numbers. The results are shown in Figure 68 (AB). and are insensitive to Reynolds number and thickness. α CL αmin.dat α αmin. the results can be applied to a full 360o angle of attack range.5 0 0 0. It was found that the poststall behavior may be modeled as modified trigonometric functions. The numerical data was downloaded and analyzed.5 Angle of Attack.The lift and drag airfoil data behavior as functions of angle of attack are shown in Figure 67 (AB). By symmetry. A regression analysis was used to help determine a curvefit to the poststall data.5 Lift Coefficient 1 0. drag. Only lift and drag are considered here. respectively.5 1 0.dat α αmin. Unlike the prestall angles of attack.dat αmin.5 0 0 0. the poststall lift and drag are functions of angle of attack alone.
stall ( ( ) ) π α − α stall 1 + sin ⋅π − o 2 5 ⋅ 2 The lower stall function takes the same form.0617 ⋅ ( sin (2 ⋅ α ) ) 1.8332 ⋅ ( sin (α ) ) 1. the flexibility afforded by this methodology is considered ample justification for any modeling inaccuracies of a phenomenon that is inherently rife with uncertainty.0617 ⋅ ( sin (2 ⋅ α ) ) Otherwise C l = −1. To have the airfoil prediction code jump from the pre.3548 After the onset of stall. It is therefore desirable to create some smoothing function in a transition region. The transition function for the upper stall is as follows: C l = C l .The drag coefficient is: C d = 1. However.to poststall values without any blending would create a discontinuity that may prove troublesome for some design problems. Results for the directdata interpolation subroutine are presented in Figure 69 (AB). The poststall subroutine is applied to both the directdata and parametric airfoil interpolation subroutines. It is recognized that some airfoils have an abrupt stall and that a transition function may not be the best representation for those cases. The poststall data covers only symmetrical airfoils. the lift coefficients decrease and the drag coefficients increase relative to their prestall values.3548 1. Note that the drag is on a logarithmic scale. 78 . A sinusoidal function was selected for a transition region of 5o angle of attack. The drag transition function takes the same form as the lift transition function. stall + C l α = α stall + 5 o − C l . and the effects of camber are not known at this time.6729 If sin(2α) is greater than or equal to 0. then the lift coefficient is: C l = 1.
61 11. Angle of Attack 2.000 Airfoil S1223 SD8020 S823 SD7032 MH46 MH64 S1223 SD8020 S823 SD7032 MH46 200.49 3.150 0.85 79 Cm0 0.150 0.2.10 21.005 0. Angle of Attack 1 40 20 0 20 40 Lift Coefficient vs.1 0.00 9.000 t/c (%) 11.5 1 0.099 0.099 0.67 0.290 0. In other words.5 2 Angle of Attack.85 1.4. the airfoil set changes with reference Reynolds number.0100 0.290 0.49 3.3 Direct Airfoil Data Method The direct airfoil data method interpolates airfoil lift and drag data taken from an airfoil database.95 11. The database contains airfoils designed for a specific Reynolds number.10 21.000 0.5 60 40 0 20 0.00 9.67 0. Reynolds Number 100.35 Camber (%) 8.35 8.00 2. Deg Figure 69 Example Drag polar and Lift Curve of an Airfoil Through a Large Angle of Attack Sweep 6.01 0.93 10.00 2.66 2.95 11.5 2 Lift Coefficient 1.0100 Source XFOIL Selig Selig Selig XFOIL XFOIL XFOIL Selig Selig Selig XFOIL .66 2. Deg 20 40 Drag Coefficient 0.5 0 1 1.000 0.001 Angle of Attack.Drag Coefficient vs. The current airfoil database contents are listed in Table 64 and Table 65.93 10.91 8.
010 0.0100 0.67 0.000 500.010 0.35 8.91 2.005 0.150 0.099 0.005 0.050 0.00 1.000 MH64 S1223 SD8020 S823 SD7032 MH46 MH64 S823 MH46 MH64 MH78 MH20 NACA 2412 NACA 4412 8.85 1.02 12.0575 0. 0.6 2.49 3.49 2. 4.91 2.00 12.61 21.150 0.35 8.91 8.1065 XFOIL XFOIL Selig Selig Selig XFOIL XFOIL Selig XFOIL XFOIL XFOIL XFOIL XFOIL XFOIL Table 64 Low Reynolds number airfoil database.10 21.00 9.00 2.1 2. 80 .93 10.005 0.00 11.95 11.61 14.000 0.61 11.300.4 9.66 2.85 1.290 0.
single camber. most of the Reynolds numbers in the second method are not at the design condition.000.000. This method differs from the direct airfoil data interpolation in another regard. The list of airfoils for which empirical equations are written is provided in Table 66. such as standard versus natural laminar flow airfoils.00 12. The variety of airfoils is important to ensure that the target airfoil is bound within the parameter space.000 t/c (%) 11.00 2.0575 0.050 0. Despite this limitation.00 12.0575 0. This method divides the drag into skin friction and profile drag components.00 0.91 2. The first method uses the data for airfoils that may be designed for the Reynolds numbers of interest.000 Airfoil MH46 MH64 MH78 MH20 NACA 0012 NACA 2412 NACA 4412 NACA 0012 NACA 2412 NACA 4412 NACA 0012 NACA 2412 NACA 4412 NACA 0012 NACA 2412 NACA 4412 3.00 12.1065 Source XFOIL XFOIL XFOIL XFOIL XFOIL XFOIL XFOIL XFOIL XFOIL XFOIL XFOIL XFOIL XFOIL XFOIL XFOIL XFOIL Table 65 High Reynolds number airfoil database. 81 . the empirical equations are very useful for describing the behavior of airfoils as a function of Reynolds number.005 0. The empirical equation method models the behavior of a given airfoil over a very wide range of Reynolds numbers. 0.2.0575 0. Some of the provided airfoils represent different technologies. 0.4 9.000. Equations were generated for airfoils of symmetrical.4 Parametric Airfoil Method The parametric airfoil method uses empirical equations to generate smooth and adaptable airfoil lift.4.00 0. 4.1065 0.000.00 12.010 0. 0. Since airfoils are designed for a relatively small Reynolds range.00 12.1065 0.00 2. The direct data airfoil method was not used to generate the results of this research.85 1.Reynolds Number 1. and drag data. 4. Cm0 0.00 2.00 2.00 12.6 2.010 0.00 12.0575 0. which facilitates the application of technology corrections to each component. angle of attack.02 12.000 20.000. 6.000 to 20.00 Camber (%) 2.1065 0.00 0.61 14. 4.000.00 12.00 12.00 0. and reflexed camber sections. 4. All equations are based upon XFOIL [Drela 2000] outputs over a Reynolds number range of 60.000 9.35 8.00 12.1 0.00 12. The accuracy of the direct airfoil data interpolation is only as good as the number and type of airfoils in the database.
103 Table 66 Empirical equation airfoil database.max c dp = c dp 0 + ∆c dp 2 ⋅ (cl − cl .bot Lower surface transition location x / ctr .00 12.top = x / ctr . where the difference between airfoil data and the calculated data is minimized up to stall.bot π ⋅ sin ⋅ cl 2.00 4.top Upper surface transition location x / ctr .00 Cm0 0. Second level equations are functions of the first level equations and the lift coefficient.00 2. The first level equations for the upper surface transition are as follows: 82 .top 2 cl − cl1.min d ) + c db 2 6 Pressure drag where cl − clb1 π 1 + sin c − c ⋅ 2 ⋅ π − 2 lb1 lb 2 = c db 0 ⋅ 2 c db Laminar separation bubble drag The factors and exponents given symbols on the right hand side of the second level equations are functions of only Reynolds number for a given airfoil.0525 0. a simple set of equations having a similar form can be used to model the behavior of a wide range of airfoil parameters for many airfoil sections.66 3. The second level equations are as follows: cl − cl1.top π ⋅ cos ⋅ cl 2 . For a given airfoil.bot .max ntr . The factors and exponents described by first level equations are determined through an optimization routine in Microsoft Excel.bot = x / ctr .bot − cl1.14 12.95 11.bot 2 ntr . first level equations are functions of Reynolds number only.00 Camber (%) 3.015 0. Quite unexpectedly.25 0.Airfoil SD7032 MH18 NACA 0012 NACA 2412 NACA 4412 t/c (%) 9. The forms of the first and second level equations were selected from visual inspection of the behavior of the data.min d ) + ∆c dp 6 ⋅ (cl − cl .top − cl1.00 12.099 0.top .000 0.
bot ⋅ sin ⋅ Re ref 2 ncl1.max = 1. max 0 + ∆x / ctr .top ⋅ sin ⋅ Re ref 2 n Re π cl max cl max = cl max 0 + ∆cl max ⋅ sin ⋅ Re ref 2 Re π = cl max r 2.bot . The first level equations for the lower surface transition are as follows: x / ctr .bot cl 2.top .bot = cl max r 2 ⋅ cl max Re π = ntr . The first level equations for the pressure drag are as follows: 83 .bot 0 + ∆cl 2.top cl1.top .0 cl1.bot ⋅ sin ⋅ Re ref 2 n Re π ntr . max l max r 2 cl max r 2 nntr . max Re π = x / ctr .tr . max ⋅ sin ⋅ Re ref 2 n Re π cl1. the optimization of the second level upper surface transition location formulation involves twelve independent variables.bot Re π = cl1. since the maximum transition location is set to 1.bot n Re π cl 2.bot 0 + ∆cl1.top 0 + ∆ntr .bot ⋅ sin ⋅ Re ref 2 The lower surface transition location formulation optimization involves nine independent variables.0 for all Reynolds numbers.0 + ∆cl max r 2 ⋅ sin ⋅ Re ref 2 cl 2.top Excluding the maximum lift coefficient.bot = ntr .top = cl1.top ⋅ sin ⋅ Re ref 2 nc nx / c .bot ntr .x / ctr .bot 0 + ∆ntr . top ntr .top .top . which is evaluated in a separate optimization.top 0 + ∆cl1.bot = cl 2.
bub 2 nclb 2 Re π clb 2 = clb 20 + ∆clb 2 ⋅ sin ⋅ Re ref .bub 2 nclb1 Re π clb1 = clb10 + ∆clb1 ⋅ sin ⋅ Re ref . min d 0 + ∆cl . involves thirteen independent variables. but merely large deviations from the polynomial drag growth characteristics of the pressure drag equation. The first level equations for the laminar separation bubble drag are as follows: ncdb 0 Re π cdb 0 = cdb 00 + ∆cdb 0 ⋅ sin ⋅ Re ref . since the optimizer will otherwise exploit the parameters in areas where the bubble drag is not relevant. 84 .cdp 0 Re π = cdp 00 + ∆cdp 0 ⋅ sin ⋅ Re ref 2 ncdp 0 100. Re ⋅ Re n Re π cl min d cl . min d ⋅ sin ⋅ Re ref 2 n Re π ∆cdp 2 ∆c dp 2 = ∆c dp 20 + ∆∆c dp 2 ⋅ sin ⋅ Re ref 2 n Re π ∆cdp 6 ∆c dp 6 = ∆c dp 60 + ∆∆c dp 6 ⋅ sin ⋅ Re ref 2 The optimization of the pressure drag formulation. excluding the bubble drag. nclb1 = nclb 2 = 1 and ∆c db 0 = −c db 00 All parameters for the bubble drag are selected via manual inputs.bub 2 Typically. min d = cl . The bubble drag term does not fully capture the bubble drag.000 + ∆cdp .
the values are held fixed at: param = param0 + ∆param The reference Reynolds number for all first level equations except for the bubble drag is 20 million. the airfoil interpolation subroutine interpolates the first level lefthandside (LHS) parameters versus the more numerous righthandside (RHS) parameters. Do i=1.The above first level equations are applicable for a Reynolds number range of 60. Due to the nonlinear nature of the first level equations. The friction drag is found using standard laminar and turbulent flat plate skin friction equations found in Raymer [1992].000 to the reference Reynolds number. The overall skin friction coefficient for a surface that transitions from laminar to turbulent flow is: c f = x / ctr ⋅ cf lam + (1 − x / ctr ) ⋅ c fturb The airfoil skin friction coefficient is the summation of the upper and lower skin friction coefficients. The following pseudocode shows the interpolation process: Param = 0. The total drag of the airfoil is given by: c d = c dfrict + c dp where the pressure drag is found directly from the above second level equation. The bubble drag reference Reynolds number is dependent upon the characteristics of the airfoil. 85 . After the reference Reynolds number is exceeded.nAirfoils Param = Param+LHS(i)*Fact(i) End Do Where nAirfoils i =1 ∑ Fact (i) = 1 The interpolated LHS of the first level equations are used to evaluate the second level equations.
4 0.2 0 0. Correlation is quite good for this case as well.5 CL 1 1. The lines without markers represent the curve fit. Lift Coefficient Comparison Between XFOIL and Empirical Equation Matching for the NACA 2412 Airfoil Figure 610 (AB) show the correlation between the XFOIL and parametric boundary layer transition locations as a function of Reynolds number and lift coefficient. Figure 611 The Pressure Drag Comparison Between XFOIL and the Empirical Equation Matching for the NACA 2412 Airfoil Figure 611 shows the correlation between XFOIL and parametric pressure drag coefficients over a large range of Reynolds numbers and lift coefficients.NACA 2412 Upper Transition 1.8 0.5 0 0. Agreement is quite good for most Reynolds numbers.6 0.5 0 0. 86 .2 1 1.5 0.5 CL (A) (B) Figure 610 Upper and Lower Transition Location vs. and the lines with markers represent XFOIL data.6 0.5 Re x/c Transition Re 1 0.2 1 NACA 2412 Lower Transition 1. The laminar separation bubble drag occurs at lower Reynolds numbers.2 x/c Transition 0.2 0 0.4 0.8 0.
5 2 Parametric Airfoil Interpolation 0. (A) (B) Figure 613 Comparison Between XFOIL and Wind Tunnel Data.020 0. Second.035 0. the interpolation of first level equation results in method 2 has some nonlinear effects in the second level equations.Direct Airfoil Data Interpolation 0.025 CD 0. Third. The magnitude of the minimum drag is in close agreement. 87 .020 0.015 0.040 0.015 0.5 1 CL 1.5 2 (A) (B) Figure 612 Comparison of Drag Polars Between the Direct Data and Parametric Airfoil Interpolation Methods Figure 612 (AB) shows the difference in drag polars for a general airfoil between the direct airfoil data interpolation (method 1) and the parametric airfoil interpolation (method 2).000 0 0.010 0.005 0. method 1 uses wind tunnel data at low Reynolds numbers that has a stronger pressure drag rise than XFOIL data.030 0. the database for method 1 is presently populated by more airfoils at low Reynolds numbers. This is caused by three factors.035 0.010 0. First.000 0 0.005 0. but the profile drag grows more slowly with method 2.025 CD 0.030 0.040 0.5 1 CL 1.
XFOIL does capture the laminar separation bubble effects on pressure drag. 6. though the behavior does differ. the simplified airfoil method was used to satisfy the profile drag needs. but it was not used to generate any results for this research. but the wind tunnel data does not necessarily correspond to the conditions encountered during flight.Fact. Figure 613 (A) shows the correlation for a Selig Donnovan SD 7032. is found using: 88 . as can be seen on the SD 7032 graph in Figure 613 (A). In general. TR2D. The 2D transition multiplication factor. While the detailed airfoil methods were under development.4.2. Ncrit is set to 9 for all XFOIL runs. 6. This method is available within the code. All wind tunnel data for Reynolds numbers of 300. This is close to the more traditional methods employed in conceptual design. The high Reynolds number runs for the NACA 0009 are from the NACA tests as reported in Abbot and Von Doenhoff [1959]. Both series of tests describe the tunnel conditions as 'low turbulence'. a simplified airfoil method was developed.Figure 613 (AB) show the correlation between XFOIL and wind tunnel data for cambered and symmetrical airfoils.5 Simplified Airfoil Method The two detailed airfoil methods both require a database and a large number of calculations. Due to these considerations. using curve fits to airfoil data and XFOIL results. In order to reduce memory requirements and to speed computation time.6 Laminar Flow Technology The impacts of 2D laminar flow due to design level and laminar flow influence on transition are covered. K. According to the XFOIL manual [Drela 2000]. the eN turbulence model corresponds to an average turbulence wind tunnel when Ncrit = 9.4. the XFOIL results tend to underestimate the drag measured in a wind tunnel. Correlation suffers with increasing camber. The parametric airfoil method was used to generate all results for this research effort. and the parabolic parameter. and Figure 613 (B) shows results for a NACA 0009. Cl0. even if the results do not match the wind tunnel data most closely.2. The airfoil profile drag takes the form: Cd = Cdfrict ⋅ FF + K ⋅ (Cl − Cl 0 ) 2 Special functions were created for the minimum drag lift coefficient.000 and below come from Selig's low speed wind tunnel tests at University of Illinois or Princeton [Selig 1989]. Decreasing the value of Ncrit does improve the correlation with the available wind tunnel data. A Ncrit value of 9 is low compared to the flight conditions encountered by some aircraft types.
incline ) CDint. For other airfoils. The interference drag of a straight wing perpendicularly intersecting a flat wall with no lift is estimated by Grasmeyer [1998] as: CD int. The slope of the curve is: Slope = 9 ⋅ (1 + 0.FACT is a multiplication factor to account for improvements possible with a good design. Fact = 1 + 2 ⋅ (TechLam − 0.TR2 − D . Fact = 1 − Slope ⋅ Λ c / 4 ⋅ 60 π This factor is applied to the profile drag just like the 2D factor. wall + CD int.5) The TR2D. Hoerner [1990] suggested that 0. The effect of sweep on transition is modeled by a curve fit to the laminar flow versus sweep technology curve used by Grasmeyer [1998].3 Interference Drag The interference drag model is built upon that used by Grasmeyer [1998] on the SBW project.4. Future research could investigate laminar flow improvements relative to a baseline airfoil through experimental or CFD efforts to further develop this function.0003 ⋅ S ref 3 ( ) The interference drag increment due to lift is estimated by Grasmeyer [1998] as: 89 .5 − TechLam )) 15 The transition multiplication factor due to sweep is: TRSweep . 6. wall c2 = 0.Fact transition multiplication factor behavior described by this equation is assumed in the absence of other models. which is a modified version of the Hoerner [1990] method.8 ⋅ t / c − 0.4 ⋅ (0. The transition factor is multiplied by the upper and lower x/c transition locations for the parametric airfoil model. The total interference drag is: CD int = CD int FACT ⋅ ∑ (CD int.1 is a suitable value. the ratio of the flat plate skin friction drag for an assumed fixed transition location multiplied by the transition multiplication factor and the flat plate skin friction drag for an assumed fixed transition location is multiplied by the profile drag. lift + CD int. sweep + CD int. The interference drag is broken down into several components that can be summed.
CD int. lift c2 = 0 . Detailed geometry inputs are required for this formulation. Fuselage overall size never enters into the formulation.0015 ⋅ θ incline ⋅ ( ) Sc 2 ref The Hoerner [1990] formulation as modified by Grasmeyer [1998] has limitations. An algorithm determines which wing segments intersected each fuselage. The height of the intersection along the fuselage is translated sideways 90 . The interference drag of two intersecting wings is not included in this formulation. the spanwise point on the wing panel that intersects the fuselage and the height along the centerline of the fuselage are determined.1 ⋅ C ⋅ S ref 2 l The interference drag increment due to wing sweep relative to the flat wall is estimated by Grasmeyer [1998] as: CD int. sweep = − 0. θincline θincline Figure 614 WingFuselage Interference Geometry Determining the interference geometry is not trivial. a fuselage length equal to the chord length will produce the same interference drag as a fuselage a hundred times this length.00006 ⋅ θ incline + 0. incline = 0. In other words. Figure 614 shows the basic wingfuselage interference geometry. The sweep interference drag formulation is suspect because it produces negative drag coefficients at moderate sweep angles.0009 ⋅ Λ c / 4 ⋅ ( ) c2 S ref The interference drag increment due to inclining the wing relative to the flat wall is estimated by Grasmeyer [1998] as: 2 CD int. The flat wall assumption may not hold for fuselages of small diameter.00018 ⋅ Λ2c / 4 + 0. but the overall accuracy is questionable. Once the intersection is confirmed.
to the right and left. 91 . and straight lines between these points are assumed to represent the fuselage adequately. This factor is always less than unity. The sphere and slender fuselage formulations are smoothly blended together. 6. The form factor equation presented in Raymer [1992] for circular cross sections is applied. The difference between the wing dihedral angle and the slopes of the fuselage walls is assumed to be the inclination angle. if the wing passes through the fuselage. frict + CD . This area is divided by the total planform area of the given surface to provide a profile drag multiplication factor. frict ⋅ FFFuse ⋅ (∑ Penalty ) ⋅ S wet . If the wing ends in the fuselage. Fuse S ref where the pressure drag is accounted for by the form factor and the penalty factors. Fuse = CD . then the average of the right and lefthand side inclination angles is used.4. A number of fuselage points are defined within the conic section formulation. press ) ⋅ S ref 1 S ref For a slender fuselage this formulation becomes: CD . but the wing dihedral angle is preserved. Otherwise. Fuse = (CD . then the inclination angle is taken only for the relevant side. The general equation for fuselage drag is: CD . The fuselage cross section may be circular or have squared edges at the higher fineness ratios.4 Fuselage Drag The fuselage drag subroutine calculates the drag of generic bodies ranging from slender fuselages to spheres to flat plates normal to the freestream. so it can be considered a constructive interference drag. The above geometry calculations are performed on the leading and trailing edges. The amount of span covered by the fuselage for the leading and trailing edges are averaged and multiplied by the chord to estimate the amount of covered wing planform area.
ρ (rho).2 0. Figure 615 shows the penalty factors. This data is presented in Figure 616.2 0.4 Rho 0. provided drag penalties for cross sections ranging from ovals to rectangles.6 1. Penalty Factor 92 . The volumeaveraged ρ value for the fuselage is used to find the penalty through interpolation of the data.414. X/L Figure 616 Fuselage Drag Maximum Diameter Location Penalty Ritchie [1968] provides drag penalty data associated with the location of the maximum equivalent diameter location of the fuselage.6 0 0.8 1 Figure 615 Fuselage Drag Cross Section Penalty Not all fuselage sections of interest are circular in cross sections. This data was modified to correspond with the conic parameter.6 0.4 0. The ρ value associated with a circular cross section is 0.8 0.4 Penalty Factor 1.Fuselage Drag Cross Section Penalty 1.2 1 0.8 1 Maximum Diameter Location. Fuselage Drag Maximum Diameter Location Penalty 9 8 7 6 5 4 3 2 1 0 0 0.6 0. The penalty can be quite high for fuselages with the maximum equivalent diameter located too far forward or too far aft. Ritchie [1968].
so the wing would sweep. The drag divergent Mach number is found for a number of panels through the following equation: M DD (i ) = cos(Λ c / 2 (i ) ) κa − t / c(i ) Cl (i ) − cos (Λ c / 2 (i ) ) 10 ⋅ cos3 (Λ c / 2 (i ) ) 2 where the airfoil technology factor. The artificial wing sweep justified setting the factor equal to unity at all times. The aerodynamic center is fixed on the fuselage in the parametric geometry. the sphere and flat plate are blended together.Unfortunately.41 when below a Reynolds number of 400. Above a L/DFuse value of 2. the wing root would shift forward to align the fairing with the radome of some configurations. A flat plate is assumed to have a drag coefficient of 1. which is based on the Korn equation.87 for a NACA 6Series section and 0. sphere ⋅ S frontal S ref where the friction and pressure drag are accounted for in CD. κa. Between 1 and 0.18 referenced to the frontal area regardless of shape. The selection of the airfoil technology factor is described in the technology modeling chapter.15 when above. the solution is linearly blended between the slender fuselage and sphere. the slender fuselage formulation is used. The three cases are blended together as a function of length to diameter ratios.4. the thickness location function was exploited by the optimizer to artificially drive sweep. 6. Note that the sphere drag coefficient is referenced to the sphere frontal area.1 M crit (i ) = M DD (i ) − 80 1/ 3 93 . so the ratio of the sphere frontal area to the wing reference are is applied to be consistent with other drag contributions.5 Wave Drag The wave drag formulation is the same method used by Grasmeyer [1998]. Because the wing was affixed to the fuselage by a parametrically blended wing fairing. The drag coefficient for a sphere is found by: CD . The critical Mach number is defined found by: 0.sphere. fuse = CD .000 and 0. The sphere drag coefficient is 0.95 for a supercritical section. The wave drag is the realworld driver for wing sweep. Between 2 and 1. is estimated by Grasmeyer [1998] as 0.
12 and 0.25 ⋅ N tire / strut ⋅ N strut ⋅ Dtire ⋅ Wtire S ref where the size of the nose and main gear may be different. This geometry is passed to the gear drag subroutine prior from the complex geometry definition subroutines.If the flight Mach number is above the critical Mach number. Note that 0. The landing/takeoff condition has the gear fully extended. based upon factors derived from Hoerner [1990].25 is a typical frontal area based drag coefficient. gear = ∑ (CD . streamwise doors.35.6 Landing Gear Drag The landing gear drag is broken into two phases of flight: cruise and landing/takeoff. the fuselage wave drag is not incorporated into the formulation. wave (i ) = 20 ⋅ (M − M crit ) and the total wave drag coefficient is: CD . wave (i ) ⋅ S segment S ref 3 If the flight Mach number is below the critical Mach number. if the configuration has retractable gear. The cruise condition has all the gear in the full retraction position. and body cavities. strut + CD . The landing gear drag coefficient is defined as: CD . The gear drag is composed of combinations of tires. The above drag coefficient is then modified by a multiplication factor for rear fairing. the impacts of fuselage wave drag in the lower transonic region are considered to be unimportant in the determination of cruise Mach number. struts. Therefore. The wing wave drag is assumed to become a major driver at a lower Mach number than the fuselage. depending on tire type. Hoerner [1990] shows a variation in drag coefficient between 0.4. cavity ) The tire drag is estimated for the nose and main gear by: CD . then the wave drag is zero. door + CD . or wheel pants cases. 94 . wave = ∑ Cd . Since the wave drag increases so rapidly with Mach number. forward fairing.tire = 0.tire + CD . 6. then the local wave drag coefficient is: Cd .
If the gear is permanently extended. with the critical Reynolds number centered about 3x106. The gear doors are assumed aligned with the flow. The equivalent flatplate turbulent skin friction on the door is calculated on both sides to yield the drag coefficient. strut.0083 ⋅ N strut ⋅ Wdoor ⋅ Ldoor S ref where the value for the nose and main gear well cavities may be different. then the retracted drag coefficient is zero and the extended drag includes the tire. the gear bay cavity drag is calculated for the extended gear case. assuming natural transition.4. then both drag values include the strut and tire. but not the gear well cavity or the gear door. For retractable landing gear. the total drag associated with cooling is found by: U DRam = mdot Ram ⋅ U ∞ ⋅ 1 − e U ∞ The ram air mass flow rate (kg/hr) [Sloan 1985] is found by: mdot Ram = 3. 6. then gear door drag is calculated for the extended gear case. and cavity. the engine and systems require a cooling mass flow.cavity = 0. and the simple airfoil method is employed for calculating the drag.The strut drag is treated as a circular cylinder perpendicular to the freestream or as a faired strut. The retraction scheme can be different for the nose and main gears. The drag coefficient is: C D . The circular cylinder drag coefficient varies with Reynolds number. If retractable landing gear is used. Regardless of the cool air source. The faired strut assumes a t/c of 0. If a gear is retractable. The total drag for the up and down position depends on the gear configuration.33. The cooling can be accomplished from an environmental control system (ECS) driven by bleed air from the engine or by a ram air ECS system. The drag of the open gear bay cavity is found in Hoerner [1990].59 ⋅ Qcool Tcomp − Tair 95 .7 Cooling Drag The cooling drag associated with the engine and systems is book kept as in the propulsion section. For ram air ECS systems. door.
the heat load is in Watts. 96 . the factor acts as a simultaneous minimum drag shift and increases the parabolic behavior of the drag polar. The cooling drag for an air cycle machine ECS system is assumed to be the same as a ram air ECS system for the purposes of this code. and the heat load is in Watts.where the temperatures are in degrees Celsius. A 5% miscellaneous drag is used for all cases. the following is attained: DRam = 0.2 due to pressure losses in the ECS system. is assumed to be only 0.4. 6.0012686 ⋅ Qcool U ⋅U ∞ ⋅ 1 − e U Tcomp − Tair ∞ where the drag is in pounds. the temperatures are in degrees Fahrenheit. To put the ram cooling equation in convenient units. A userinput miscellaneous drag factor can be applied to scale the drag polar. and the velocity is in knots. so it scales the minimum drag value and the drag due to lift by the same ratio. This factor is a direct multiplier. In other words. A fuelcooled ECS system. like the one used on Global Hawk. The velocity recovery. is assumed to have no drag penalties. Ue/U∞.8 Miscellaneous Drag There are inevitably many drag contributors that are not accounted for in the above formulations.
which then calls 97 .Chapter 7 Propulsion 7. The engine deck consists of the following arrays: Altitude(altmin:altmax) Velocity(altmin:altmax.velmin:velmax. The jet thrust to weight and specific fuel consumption are assumed to stay constant. turboprops. velocity. The primary output of the propulsion code is an engine deck. such as being outside the permissible altitude. velmin:velmax. The threedimensional thrust. to a specified maximum altitude for the engine deck. Figure 71 shows the top level propulsion code architecture. ACMAIN calls EVAL. throtmin:throtmax) Allow(altmin:altmax. throtmin:throtmax) TSFC(altmin:altmax. The engine deck has a common format. meaning that many engine characteristics are independent of scale. ACMAIN calls INFILES prior to any optimizations on analyses. The optimizer freely scales the engines to meet the demands of a given solution. and throttle setting. The burden falls on the propulsion methods to generate this single format deck. velmin:velmax) Thrust(altmin:altmax. The allowable array indicates whether or not a given point in the engine deck violates a propulsion constraint. velocity. Scaling engines is a bold assumption. The initial propulsion data is retrieved from PROPDATIN which is called by INFILES. The twodimensional velocity array varies with altitude and velocity setting. which is used by the performance analysis. Each engine type has specialized analysis methods.1 Propulsion Overview The propulsion code evaluates the performance of turbofans. The engine is scaled as a rubber engine. Thrust Specific Fuel Consumption (TSFC) and allowable arrays vary with altitude. Despite the realworld limitations. The thrust specific fuel consumption is in jet engine format. The piston engine power to weight ratio and brake specific fuel consumption remain constant. because the performance code must adapt to multiple different engine types without modifying its formulation. 7. throtmin:throtmax) The onedimensional altitude array varies from the minimum altitude. typically sea level. or thrust bounds. This may not always be reasonable due the high cost of engine development and the typically fastpaced UAV development schedule. and pistonprops. velmin:velmax.2 Code Architecture The propulsion code is broken into multiple subroutines based upon propulsion type. This implies that the appropriate engine is either available or can be produced in time to meet the UAV development schedule. the rubber engine assumption is applied here.
The first step is to calculate the propeller diameter in PROPDIAM using empirical formulas for a specified design condition. The jet engine thrust and fuel consumption characteristics at each altitude. then the propeller is designed in PROPDES. If a complex propeller design method is chosen. electricprop. and throttle are found via the TURBORATIO subroutine. The engine power and fuel consumption characteristics are found by PISTALT for piston engines and by TURBORATIO for turboprops. PROPULSION can call the propulsion category subroutines: GASPROP for piston and turboprops. The engine deck loop is then entered. The TURBOFAN subroutine creates engine decks for turbojets and turbofan propulsion systems. The JETON detailed jet ondesign performance subroutine is integrated but was not used here. Two propeller analysis methods are employed. and TURBOFAN for turbojets and turbofans.3 Propeller Methods Common propeller design and evaluation methods are used. and turboprop propulsion systems all utilize propellers. Figure 71 Propulsion Code Architecture GASPROP combines pistonpropellers and turboprops because of a high degree of commonality.PROPULSION. velocity. The propulsion system is evaluated at each point. because pistonprop. and thrust levels (throttle) vary. The propeller efficiency is found by PROPEFF if the disk actuator model is used or by PTHRUST if the complex propeller analysis method is used. The atmospheric conditions are found by standard atmosphere routine. The tabular engine deck is then calculated. A simple disk actuator method provides corrected 98 . STDATM. 7. velocity. A loop is entered where the altitude.
The Predator and Shadow 200 designs appear to fit within the application of the diameter estimation. The blade element model is fully integrated into the code. Power 12 Propeller Diameter. Unfortunately. it is only utilized for determining supportable correction factors for the faster disk actuator model. Therefore. The Raymer [1993] formulation is used to aid in the scaling. Both of these cases use twoblade propellers. D = DRe f ⋅ D func (BHPRef D func (BHP ) ) 99 .theoretical efficiencies with little computation time. Propeller Diameter vs.1 Propeller Diameter Sizing Raymer [1992] provides empirical equations for sizing propellers as a function of maximum engine power and the number of blades. These equations are intended for manned aircraft with two or three blades. The propulsion code has the ability to size propeller diameters based on empirical formulas. UAV designs frequently make more extensive use of gearing than manned aircraft to achieve large propeller diameters with high relative efficiency. the Raymer methods are used when the propeller diameter is not specified. Feet nb=2 nb=3 Data Pts 10 8 6 4 2 0 0.3. and is available for use. the blade element model is computationally expensive relative to the disk actuator model. 7.01 0. Figure 72 shows results of the Raymer empirical equations. A detailed blade element model designs propellers and provides highly fidelity propeller analysis outputs. Horsepower Figure 72 Raymer Propeller Sizing Equation Results The second method for sizing the propellers is scaling from known propellerengine combination. Despite the limitations.1 1 10 100 1000 Engine Power. Appendix B describes the blade element methodology in detail.
The disk actuator model is a theoretical method of predicting the ideal efficiency of a propeller. The relevant thrust equation is: T = mdot ⋅ (Ve − V ) = mdot ⋅ ∆V For a propeller. the disk actuator model cannot predict static thrust without significant modifications. Important limitations of the model are that viscosity. To remedy this deficiency. The static case is not considered in this research.nonideal 100 . Importantly.ideal ⋅η p .ideal = 2 ⋅V 1 = ∆V Ve − V +1 2 ⋅V With the above equations. airspeed (except static case). inescapable practical inefficiencies are neglected in the disk actuator model.3. the thrust equation may be written in terms of the propeller disk area. The theoretical efficiency is: η p . Details of the blade element model can be found in Appendix B. or thrust condition. propeller hubs. and other relevant aspects of propellers are neglected. the ideal efficiency can be found for a given propeller diameter at any altitude. Ap: T= 1 ⋅ ρ ⋅ A p ⋅ ∆V ⋅ (2 ⋅ V + ∆V ) 2 Solving for the velocity difference yields: − 2 ⋅V + 4 ⋅V 2 + ∆V = 2 8 ⋅T ρ ⋅ Ap The exit velocity. Details of the method are well documented [Mattingly 1987] and will not be reproduced here. Corrections were applied to the theoretical model to provide more realistic predictions. and 2) a blade element model.2 Disk Actuator Model Two propeller performance models were developed for this effort: 1) a disk actuator model. However. Ve. nonuniform flow characteristics. Only the disk actuator model was applied here to generate the results.7. is simply the sum V+∆V. The disk swept out by the propeller is assumed to act uniformly on the air that passes through it. a nonideal efficiency correction is included. This powerful result is achieved at little computational expense. The total propeller efficiency of a nonideal propeller is: η p = η p .
where the nonideal efficiency ranges from approximately 8595% based on comparisons with the blade element model. Therefore. A comparison is made in Figure 73 between the blade element analysis and the ideal propeller using the disk actuator model. The ideal efficiency increases as the thrust level decreases over the constant propeller disk area. The propeller efficiency is a function of the nonideal propeller efficiency corrections and the ideal efficiency from the disk actuator model. The initial value of ideal efficiency is set to unity for the starting conditions. The engine shaft power to produce a given amount of thrust is: Pshaft = T ⋅V η p ⋅η gear When the flight velocity is zero. 101 . the blade element design code comes within 710% of the ideal theoretical efficiency. and the power absorbed by the propeller cannot be realistically determined by the disk actuator theory. The ideal efficiency plotted below corresponds to the thrust output at each given advance ratio. Note that the ideal efficiency neglects viscosity and other considerations. The propeller thrust can be calculated as a function of shaft horsepower: T= Pshaft ⋅η p V The thrust is a function of the propeller efficiency. The ideal efficiency is found iteratively. A nominal value of 90% is applied. the propeller efficiency is zero. A 90% efficiency factor is applied to the disk actuator model based on these results. In the sample case. the thrust is a function of itself.
Advance Ratio 1 0.9 Efficiency 0. the method applies to rotary engines as well.95 0. This helps aircraft that fly in hot.4 PistonProp Engines The pistonprop method generates an engine deck for piston engines that drive a propeller.8 0.65 0. A normally aspirated engine relies upon ambient pressure and suction to bring air into the cylinders.55 where the subscript h indicates the conditions at flight altitude and 0 indicates sea level conditions. and performance degrades as altitude increases. where the pressure of the intake air is increased by mechanical means.8 1 Predicted Ideal Figure 73 Comparison Between Predicted and Ideal Propeller Efficiencies 7. humid 102 . The GaggFerrar [Stinton 1995] equation describes this behavior: FBHP (σ ) = (1 − σ ) BHPh =σ − BHP0 7.4 0.2 0.6 0.7 0.Ideal and Predicted Efficiency vs. Although the word piston is used. This type of engine generates its maximum power at sea level.75 0.85 0. The density ratio is defined as: σ = ρh ρ0 The GaggFerrar equation is used for all rotary and reciprocating engines in this code. Engine power can be restored by means of turbo or supercharging.6 Advance Ratio J 0.
Mises [1945] wrote that the typical engine specific fuel consumption ranged from 0. The applied technology trends for reciprocating engines are found in the technology chapter. so the potential improvements made possible by these devices were not investigated here.conditions or at high altitudes. Neither Predator nor Shadow 200 are known to include turbo or superchargers. Figure 74 Density Ratio Impacts on Brake Specific Fuel Consumption 103 .117. The situation has hardly improved since 1945 due to the domination of jet engines.065 and nσ is 1. Figure 74 shows the specific fuel consumption as a function of density ratio for a typical naturally aspirated internal combustion engine. The specific fuel consumption for a piston engine is defined as the ratio of fuel flow to the power. It can be assumed that he was speaking of fourstroke engines for manned aircraft applications.55lb/HPhr. The density altitude function FBSFC(σ) may be applied directly to the sea level BSFC of a known engine: BSFCh = FBSFC (σ ) ⋅ BSFC0 The above equation is used for both reciprocating and rotary engines in this code.450. The density altitude function FBSFC(σ) is estimated by: FBSFC (σ ) = σ ⋅ (1 − C ) (σ nσ − C ) where C is about 0. and is usually given in the units lb/HPhr.
min )) ⋅ FBHP (σ ) ⋅ η gear The power available to the propeller is assumed to be equal to the power absorbed by the propeller. Unfortunately. 7. not the thrust. only the jet engines as a whole. The power available to the propeller is: P = (P0 . This power is passed to either the disk actuator or blade element analysis codes to find the associated thrust and overall propeller efficiency. it is fruitless to try to surmise how the technology improvements are divided up into multiple components. The end result is an engine deck composed of specific fuel consumption versus thrust level.1 PistonProp Engine Deck The throttle setting on a piston engine regulates the engine power directly. Complex design methods and simplified offdesign performance estimation codes were written for each engine type. The Brake specific fuel consumption is: BSFCh = BSFC0 ⋅ FBSFC (σ ) η p ⋅ η gear The engine deck for fueled engines must be written in terms of thrust specific fuel consumption and not brake specific fuel consumption. Simple offdesign relationships were established to evaluate how gross engine characteristics vary with operating conditions. but it was not used here. 104 . The objective of this research is to estimate the performance of UAVs and to predict the impacts of technology trends on future UAV designs. altitude and Mach number.4. and turboprops. max − P0 . Because of this.7. The sensible approach is to apply the gross technology trends to simple methods.5 Turbojets and Turbofans Turbopropulsion covers turbojets. turbofans. The ondesign analysis is available to the code. The conversion for piston engines is as follows: TSFC = BSFCh ⋅ V ⋅ 6080 3600 ⋅ 550 where V is in knots. technology trends are not available for many jet engine components. min + throt ⋅ (P0 . only the thrust value. and BSFCh is in pounds mass per pounds force per hour. The engine power is not stored in the engine deck.
SLS ⋅ a The maximum sea level static installed thrust is provided in an input file or is calculated within the code based upon a thrust to weight ratio or other parameter.65 for a bypass ratio of 2. and 0. 0.1 Turbojet/Turbofan Engine Decks The looping structure for the turbojet/turbofan engine deck model is similar to the pistonprop.262 ⋅ M − 0. First.5 ( 1. The bypass ratio at a given altitude and velocity is found by: TSFC = TSFCrat ⋅ TSFCSLS 105 . There are altitude. The throttled thrust is found by: T = (Tmin + throt ⋅ (Tmax − Tmin )) The thrust specific fuel consumption is found using an equation modified from Grasmeyer’s [1998] SBW code.7 a = 0. This equation is calibrated to work well for high bypass ratio jet engines.245 ⋅ M − 0.0.88 + 0.5 )⋅ σ 0.4 )⋅ σ 0. The minimum installed thrust in flight is assumed to be 5% of the maximum thrust.7. The throttle loop regulates the thrust.0.25 ⋅ M − 1. velocity.4704 ⋅ (TSFC ref + 0.7 The actual thrust ratio is found through a linear interpolation between the thrust ratios based on bypass ratio.0.6 a = 0.4021 ⋅ M ) TSFCref The reference thrust specific fuel consumption is 0. though it also works well for low bypass ratio engines and turbojets: TSFCrat Temp = Temp SL 0.5 for a bypass ratio of 7. The reference value used in the above equation is interpolated within these values by bypass ratio.85 for a bypass ratio of 0. The maximum thrust at a given altitude and velocity is found by: Tmax = Tmax. and throttle loops.5.4 ( 3 ) for high bypass ratio (assumed bypass = 7) for low bypass ratio (assumed bypass = 2) for high bypass ratio (assumed bypass = 0) ( 1.2 ⋅ σ 0.907 + 0.568 + 0. the maximum thrust at the given altitude and Mach number is determined using modified Mattingly [1987] equations: a = 0.
The best installed sea level static thrust specific fuel consumption is provided in an input file.
106
Chapter 8 Size, Weight, and Power
8.1 Size, Weight, and Power Overview
A systems approach to AV design is essential, since Size, Weight, and Power (SWaP) of all components comprising the aircraft have significant impacts on the overall sizing. The size, as considered here, refers to the component volume. Weight considers both component weights and installation weights. Power contains both power consumed by the component and the cooling power that must be dissipated. These considerations must be addressed for every component on the aircraft whether it falls under structures, avionics, subsystems, propulsion, payload, or fuel categories. A flexible means of evaluating SWaP is provided.
8.2 Generic Systems
8.2.1 Generic Systems Overview
The systems approach to aircraft design is essential for UAV applications. There simply is no way to contend with such a wide range of vehicle sizes and missions using empirical methods alone. The systems method is a core strength of the design code. Consideration of SWaP must be given to every component that enters the air vehicle. Component and type attributes have been assigned to every component that enters the aircraft. The categories and components comprising the categories are shown in Table 81.
Avionics Autopilot INS GPS Processor Camera ADS Recorder ATC Subsystems FCS ECS Communications Electrical System Fuel System APU Structures Generic Struct. Wing Tails Fuselage Nacelles Landing Gear Propulsion Generic Prop. Recip. Engines Jet Engines Payload Generic Payload EO/IR SAR
Table 81 Component Categories and Types
Multiple options are provided for SWaP. In general, the user may input values directly or select from multiple calculation options. Many component types have specialty input subroutines, and typespecific methods can be entered if available. Though complex, this approach is intended to maximize flexibility over a wide range of problems where 107
information is often difficult to obtain. Default information for all component categories are input through the defaults file, which may be edited by the user.
8.2.2 Generic Systems Weights
The design code is intended to offer multiple means of calculating weights to enhance flexibility. Weight is divided into two categories: component weight and component installation weight. Standard methods for calculating the component weights include direct weight input, weight as a fraction of the empty weight, and weight as a fraction of the takeoff gross weight. When specialty methods are available and selected, the code calls the specialized weight subroutines for the components. Default component weight categories for all component types are set in the code, but these may be edited by the user. Standard methods of calculating the installation weights include direct input, a default fraction of the uninstalled weight, and input fraction of the uninstalled weight. The default installation weight factor is a function of the component category and level of design. Default component installation weight categories for all component types are input through the defaults file, which may be edited by the user. The default installation weight factor is found by: FWt , Install = FWt , Install , Max + FDesign ⋅ (FWt , Install , Min − FWt , Install , Max ) where the installation factor is set by component category. Typically, the minimum installation weight factor is 0.15 and the maximum is 0.30 in this research. SAWE [1996] indicates that the installation weight factor usually ranges from 0.250.5, but this range was modified as stated to attain sensible calibration results. The installation weight factor is independent of technology year. Installation weight does not apply to fuel, since the fuel system handles all fuel installation considerations. The default structural installation weight factor is set to zero. The installation weight is found by: Wt Install = WtUninstalled ⋅ FWt , Install An important aspect of the component weights estimation is determining the location of the weight. In particular, weight in the wing will provide inertia relief that affects wing weight if the piecewise linear beam model is utilized. The location also affects volume margin, as described below. The component location is determined by the user while setting up the input files. A subroutine guides the inputs to ensure the user selects only existing wings, tails, fuselages, and nacelles.
108
8.2.3 Generic Systems Size
Size, represented here as volume, considerations are very important for many UAV applications. As with manned aircraft, many components on an AV compete for similar locations on the vehicle. A vehicle sized as a point mass may be smaller and lighter than a vehicle that is designed taking volumetric details into account. Volumetric considerations constrain the design in a realistic manner. There are multiple, userselectable means of calculating component volumes. The user may input volume directly, calculate volume from an input density, default density, or enter dimensions of various shapes. The density methods allow the user to input component density directly or for default densities to be used. In limited cases, there are component specialized volume calculations performed in the component weights procedures. These specialized volumes will override densitybased volume calculations, but not dimensional inputs or direct volume inputs. The specialized volume calculations are described along with the weight methodology in the weights chapter. The design code has options for full or partial external volumes. External volumes may occur for antennas and payloads, though almost all components may be mounted externally. The ratio of internal to external volume is input by the user. Structures are exempt from standard volume calculations, since the volume of the structural outer mold line is calculated separately. The dimensional inputs for volume calculations permit multiple components and multiple shape types to be input by the user. This is necessary because a single functional item may consist of multiple individual elements. For example, most EO/IR ball payloads consist of the sphericalcylindrical EO/IR ball and separate rectangular driver electronics. Shapes used for calculations include rectangular boxes, spheres, hemispheres, and cylinders. The default installation volume is a function of component category and level of design. Currently, all default category volume installation factors are a minimum of 0.1 and a maximum of 1.0 of the uninstalled volume. The number used is a linear interpolation between these values with the design level as the interpolation variable. Installation volume factors are independent of technology year. The recommended method is direct user input of installation factors. For the default case, the volume installation factor is found by: FVol , Install = FVol , Install , Max + FDesign ⋅ (FVol , Install , Min − FVol , Install , Max ) The default installation volume is found by:
109
VInstall = FVol , Install ⋅ VUninstalled The code calculates the volume margin for various elements such as wings and bodies. The outer mold line volume is calculated in the geometry subroutines. All bodies and wings have volume calculated. Volumetric decrements to the outer mold line volume due to structures thickness, substructures, routing, and other considerations are estimated. The uninstalled and installation volumes of all components allocated to a given wing or body are summed. The difference between the available volume and the volume of the contents is the volume margin. Fuel may be applied to this margin if fuel is allocated to the element. Violated volume constraints are output, as described in the Optimization section.
8.2.4 Generic Systems Power
The power of all components are divided into mission phases. This enables power sequencing for flight performance fuel burn calculations and sizing of electrical and ECS subsystems. The user also has the option of specifying that the component is powered at all times. The minimum, nominal, and maximum power are either input by the user or calculated. The power level is selected either for each phase of flight or for the total mission. The calculated power is based on power per weight of the component. The design code has the ability to calculate minimum, nominal, and maximum power for some component types. These methods are described for the applicable components in the weights chapter. The performance impacts of power consumption are covered in the mission performance chapter. The nominal power is applied to all mission segments here.
8.2.5 Generic Systems Thermal Management
Component cooling is covered by the code, but component heating is not evaluated. The required cooling of a component is based on the power consumed. This varies for radiating and nonradiating types. Radiating components include RF communications antennas. For nonradiating types, the cooling required is equal to the power consumed. For radiating components the cooling required is calculated by: C = P ⋅ (1 − η Rad ) where ηRad is the radiating efficiency of the component.
110
Roskam[1990] – General Dynamics Commercial Transport FCS Weight Equation 3. nominal. Custom Method Table 82 Flight Control Systems Weight Methods Applied Here No No No No No No Yes 111 .3 Specialized Size.1. and Power Methods The specialized weights methods are selected by the user when building the system files. The flight control surfaces are not included in this category. These methods are described for the applicable components in the weights chapter.1 Subsystems 8. Multiple flight controls weight methods are integrated into the design code. and maximum cooling loads for some components. The flight control methods and associated functional dependencies are shown in Table 82 and Table 83. Raymer[1992] Fighter FCS Weight Equation 5. respectively.3. In addition to the standard weight methods. Often multiple methods per component type are available. Roskam[1990] – Cessna General Aviation FCS Weight Equaiton 2.3. 8. and equipment required to support them. This chapter describes all available methods. Raymer[1992] Transport FCS Weight Equation 6. Weight. Roskam[1990] – General Dynamics Bomber RCS Weight Equation 4.The design code has the ability to calculate minimum. Method 1. Raymer[1992] General Aviation FCS Weight Equation 7. The performance impacts of cooling are covered in the Performance and Aerodynamics sections. 8. the user has the ability to use specialized weights methods when available.1 Flight Control System (FCS) The flight control system includes the flight control actuators.
the mechanical functions of retractable landing gear could be linked. The custom flight control system equation developed is: WFCS = 0.mass 1 2 3 4 5 6 7 Yes  Yes Yes Yes Yes Yes Yes Yes Yes  Yes Yes Yes Yes  Yes Yes Yes Yes Yes  Yes  Yes Yes Yes Yes Yes  Yes  Table 83 Flight Control System Weight Method Functional Dependencies Method 1 has a severe limitation in that it requires a crew to be present. is set to one. then the equation outputs zero weight. In order to preserve some functionality. This method is not applied to the cases considered. The basis of the equation is the assumption that the actuators and other FCS weight contributions are proportional to the dynamic pressure and the total control surface area.1. NCrew NFCS NMech NSurf MMax qMax Veq. If no crew is present. 112 . Method 7 is based on rough calculations performed on model aircraft electric actuators. there is no need for cabin environmental control. This is justified because most of the FCS weight is related to crew interfaces. max ⋅ SCont ⋅ N FCS 2 8. Method 2 accounts for mechanical functions beyond flight control surfaces. the number of crew.Max SCont LOverall WOverall WDG NUlt Iyy.00002 ⋅ Veq .3. Unlike manned aircraft. The equations have been appropriately modified to reflect this consideration. Method 3 requires that the overall length and width is calculated. The maximum width is typically the wing span. the number of mechanical functions. because it is a function of design gross weight and number of flight control systems only. Many of the listed methods have very complex interactions among many variables that limit applicability beyond their intended use. There is no airspeed dependency. The environmental control system weight methods and associated functional dependencies are shown in Table 84 and Table 84. is set to zero. The overall length is the distance from the farthest forward point on the vehicle to the farthest aft.2 Environmental Control System (ECS) The environmental control system weight includes all equipment necessary to thermally manage the aircraft. Method 4 has potential for broad applicability. NMech. respectively. For the purposes of this code. which is largely the reason for elimination here. Ncrew. If integrated later.Meth.
gimbal. and therefore signal to noise ratio. Roskam[1990] – General Dynamics LowSubsonic Aircraft ECS Weight Equation Table 84 Environmental Control System Weight Methods Applied Here No Yes Method 1 2 WAntiIce Disabled WAvionics Yes Yes Table 85 ECS System Weight Method Functional Dependencies The antiicing factors in Method 2 have not been applied. A rough approximation of the volume occupied by the dish is based on estimations from drawings of Global Hawk. the antenna gain. The correlation was poor. 8. The requirements for power to the antenna decrease with improved gain for a given data rate and frequency.3 Communications System The communications system SWaP varies significantly across different applications. is: 2 WDish = 3 ⋅ DDish The factor of 3 was found through an analysis of terrestrial dishes. which has a 4foot dish diameter. which includes the dish. For many cases. these values are entered directly by the user through the generalized methods described earlier. and the data rates and 113 . The assumed dish volume equation is: VolDish D = 120 ⋅ Dish 4 3 For a given frequency. Roskam[1990] – General Dynamics Fighter ECS Weight Equation 2.1. and actuators. because none of the UAVs considered include antiicing systems. Furthermore. but the factor of 3 represents highly transportable terrestrial satellite communications dishes. The assumed dish weight equation. No link analysis is performed to evaluate these impacts. One specialized method is applied for dish antennas typically used for satellite communications. It is assumed that the selected antenna power input is proportional to the dish diameter.Method 1. the frequency and desired data rates are not uniform across currently fielded UAV satellite communication solutions.3. improves with increasing diameter. because the effective radiated power in the desired direction improves.
LElec. Method 1. batteries. is: PDish = 250 ⋅ DDish Further refinement of satellite communications systems SWaP would be beneficial to future UAV research efforts.frequencies are selected as appropriate to meet the aperture size.3. The electrical system weights methods and associated functional dependencies are shown in Table 86 and Table 87. Roskam[1990]Cessna General Aviation Electrical System Weight Equation 2. 8. This value. requires an estimation of the overall length of the electrical system. respectively. The maximum power is calculated for all avionics 114 . is assumed to be 50% of the overall vehicle length. Roskam [1990] – General Dynamics Navy Fighter Electrical System Weight Equation 6. Roskam[1990]General Dynamics Bomber Electrical System Weight Equation 5. Multiple empirical equations have been included in the code. generators. Raymer [1992] Transport Electrical System Weight Equation 7.1. Roskam[1990]USAF Electrical System Weight Equation 3. Raymer [1992] General Aviation Electrical System Weight Equation Table 86 Electrical System Weight Methods Applied Here No No No No No Yes No Method 1 2 3 4 5 6 7 WDG Yes  WFuels Yes Yes Yes Yes Yes WAvion Yes Yes Yes Yes Yes PMax Yes  LElec Yes  NGen Yes  Table 87 Electrical System Weight Method Functional Dependencies The Raymer [1992] transport electrical system weight equation.4 Electrical System The electrical system includes all elements of the electrical system including. based on bestjudgment. and the power distribution elements. converters. The assumed dish power equation. Method 2. but not limited to. Roskam[1990]General Dynamics Transport Electrical System Weight Equation 4.
1. NTanks. respectively. This method is used because it is a function of system power. and other components. but this was disabled since none of the UAVs under consideration have external fuel tanks.1. The parameter VolInternal/VolTot is set to one. Method 1. fuel pumps.5 Fuel System The fuel system contains the fuel tanks.6 Auxiliary Power Unit (APU) The user may either enter in the APU weight directly or use the APU weight estimation method. Most of the methods are adaptable to external fuel tanks.3. 115 . Several fuel system weight empirical equations have been integrated into the code. Raymer [1992] General Aviation Fuel System Weight Equation Table 88 Fuel System Weights Methods Applied Here No Yes No Method 1 2 3 VolTank Yes Yes Yes VolInternal/VolTot NTanks Yes Yes Yes Yes Yes Yes Tmax Yes  TSFC Yes  NEng Yes Yes Table 89 Fuel System Weight Method Functional Dependencies The Raymer [1992] fighter fuel system weight equation originally contained provisions for protected fuel tanks. is assumed to be two for the purposes of this code. fuel lines. Raymer [1992] Transport Fuel System Weight Equation 3. 8. not avionics and fuel system weights.and most subsystems prior to the electrical system subroutine call.3. The fuel system weight methods and associated functional dependencies are shown in Table 88 and Table 89. The total number of tanks. The estimation method is an attempt to relate APU weight from the power required of it. Raymer [1992] Fighter Fuel System Weight Equation 2. The tank volume is simply the fuel weight divided by the fuel density for the purposes of these calculations. Factors for this consideration were disabled. Fuel volume constraint checking is handled in other subroutines. 8.
GPS. recorders. which is between typical values of 2stroke and 4stroke engines. The maximum values X2 (=1) and Y2 are also input. INS. so this method was not utilized.3. 8. None of the cases considered are believed to contain an APU. all avionics were input directly. these methods were not employed in the final evaluations of the cases of interest.2 Avionics Avionics are given substantial consideration in this research. With the exception of the applied ADS method.1 General Avionics Functions The weight. It is unlikely that an effective.It is assumed that the power required of the APU is 50% of the maximum power experienced by the aircraft. design philosophy.5) and Y1 (=1) are input. and ATC avionics. cameras. and is heavy for a turboshaft.2. though the methods are not mature.5 pounds per horsepower. The APU is assumed to weight 1. and technology. The avionics suites of the vehicles under consideration are simply too divergent and casespecific to be captured in generalized formulations. For example. Though providing some utility. A general methodology for avionics weight was attempted. volume and power functions all use a common form. The common equation is: Y = Yo + a ⋅ X n Xo (=0) and Yo is input. The weight of a GPS antenna is within typical weight calculation error levels for a manned aircraft. Avionics suites are highly dependent on the mission.3. General avionics methods were attempted for autopilots. The avionics methods are shown in Appendix C for completeness. because they can be a dominant weight driver. 8. The mid values X1 (=0. a small component like a GPS antenna can weigh as much as an optical payload on a very small UAV. vehicle class. The values of n and a are found by: Y −Y ln 2 0 Y −Y 1 0 n= X − X0 ln 2 X −X 0 1 a= Y1 − Y0 ( X 1 − X 0 )n 116 . processors. generalized avionics methodology is possible.
8 0.2 Air Data System (ADS) Air data systems are necessary to provide the autopilot with angle of attack. The factors are shown in Table 810. and dynamic pressure.05 ⋅ Veq .000 feet.02 ⋅ Veq .5) 1. Two types of air data systems are considered: ADS booms and integral ADS. The ADS weights resulting from these methods appear to be reasonable for the purposes of this research. or if Y1 is greater than or equal to Y2.2. The ADS method is intended to make the ADS estimated weight of the correct order of magnitude for a wide range of UAV AVs. Future research could adjust the factors of these equations based on regression analysis of a custom ADS database. the ADS is assumed to have a heater. The weight of the ADS with heater is assumed to be: WADS = 1. The typical ADS boom weight is assumed to be: WADS = 0. The weight and volume installation factors are the same as for boards and boxes.8 0.3. 117 . pressure altitude.2 ⋅ WADS The ADS weight is corrected for ruggedness.0 Maximum (1) 1.4 Table 810 Air Data System Factors The weight and volume are initially calculated by using the boards and boxes formulas. 8.2 1. considering that there are no alternative methods available. The ADS weight equations used here are based upon bestjudgment assumptions. and miniaturization.0 1. depending on the autopilot.0 1. max The typical weight of an integral ADS system is assumed to be: WADS = 0. Driver Ruggedness Performance Miniaturization Minimum (0) 0. max If the maximum flight altitude is greater than 15.2 1. Only 25% of the ADS volume is internal. performance. angle of sideslip.6 Mid (0. then linear interpolation is used.If Y0 is greater or equal to Y1.
All wing weight methods are available. tandem wings and other types may be evaluated.1. Raymer [1992] General Aviation Wing Weight Equation No 4.1 Piecewise Linear Beam Model 118 .tot Applied Here 1. The semianalytic method is a piecewise linear beam model. transport.3 Structures An attempt has been made to provide multiple options for the various structural weights.wing Tech Yes Yes Yes Yes Yes Yes Yes Method Table 812 Wing Weight Method Functional Dependencies 8. The level flight loading at the design gross weight condition of each wing is found by: LoadWing (i ) = WDG ⋅ SW (i ) SW . 1 2 3 LoadW Yes Yes Yes NUlt Yes Yes Yes Sw Yes Yes Yes ARw t/cAvg Yes Yes Yes Yes Yes Yes λeq Λc/4 Yes Yes Yes Yes Yes Yes Sc/Sw q WFuel. but the piecewise linear beam model is believed to have the most general applicability to small AV designs. and general aviation wing weights.3. The new piecewise linear beam structures methods do not have reference UAV cases to draw upon.1 Wing Multiple methods for calculating wing weights have been incorporated into the design code. There are options for three parametric weights and one semianalytic method. so fidelity may be low.3. Numerous manned aircraft equations have been utilized. The code can handle multiple wings performing lifting functions. but other customized options have been introduced. The wing weight methods and associated functional dependencies are shown in Table 811 and Table 812.3.3. However. 8.8. so canards. Raymer [1992] Fighter Wing Weight Equation No 2. Raymer [1992] Transport Wing Weight Equation No 3. respectively.3. this is often better than using equations for manned aircraft outside their applicable range. The parametric weights are modified versions of Raymer’s [1992] fighter. The piecewise linear beam model was applied to all cases in this work. Piecewise Linear Beam Yes Table 811 Wing Weight Methods Meth.
8.1. The load distribution is assumed to be triangular. The lift load distribution method assumes an elliptical distribution. The shear and moment are corrected for the structural wing weight of outboard panels.Tot π bStruct . these methods could be adapted in future versions of the code. These are allocated to the wing by the standardized systems subroutines. The wing may carry nonstructural components such as avionics and subsystems.The wing and tail structure is analyzed using the piecewise linear beam model.Tot The overall shear load and bending moment due to lift and nonstructural weight are found by numerical integration from the tip to the root. with the maximum weight at the root. is allocated externally. Piecewise Linear Beam Load Estimation The load allocated to each wing is assumed to be proportional to the ratio of the area of the wing to the total wing area. The lift load for each panel is found by: Lift (k ) = LoadWing ⋅ bStruct (k ) 4 2 ⋅ YRun (k ) ⋅ ⋅ 1− bStruct . and bStruct. However.3.Tot bStruct . since only cantilever wings are considered. The wing weight is calculated from tip to root.Tot.3.Tot where YRun is the distance of the midpanel centroid from the root along the path of centroids. The method applied here is more limited in scope. The methods were inspired by publications from the StrutBraced Wing (SBW) research at Virginia Tech [NaghshinehPour 1998]. The total weight of these nonstructural components. which provide inertia relief.2 Piecewise Linear Beam Wing/Tail Weight Estimation The thickness of the bending material for the panel is found by: 119 . WNonStruct.Tot is the total structural span of the wing along the path of centroids. The nonstructural weight distribution is found by: WNon − Struct (k ) = Wt Non − Struct .Tot ⋅ 2 ⋅ bStruct (k ) 2 ⋅ YRun (k ) ⋅ 1 − bStruct . The total load carried by the wing is found by: LoadWing = N ult ⋅ WDG ⋅ SW SW .Tot The design code currently does not calculate the wing load distribution based on the lift distribution methodologies used in the aerodynamics.
t Bend (k ) = MomBend (k ) ⋅ I xx . The total bending material weight of the panel is found by: WBend (k ) = ρ Matl ⋅ t Bend (k ) ⋅ S Pan (k ) The total panel area moment of inertia is found by: I xx (k ) = I xx . the total shear material weight will be comparable. The shear flow is found by: qShear (k ) = qShear . The area moment of inertia for shear material calculations is identical to that found through the wing bending material calculations.t (k ) ⋅ σ ult TBox (k ) 2 If this value is less than the minimum gage value. so the contribution of shear material to the area moment of inertia and the shear flow is small. then the shear material thickness is set to the minimum gage value.t (k ) ⋅ t Bend (k ) The shear material weight assumes that the majority of the structural material is contained in the wing skins. then the thickness is set to the minimum gage thickness.Tot = Nseg k =1 ∑ 2 ⋅W Shear (k ) 120 . The shear web is assumed to be a single web which occupies the entire maximum thickness of the panel.t (k ) ⋅ t Bend (k ) The shear material thickness is found by: t Shear (k ) = Shear (k ) ⋅ qShear (k ) I xx (k ) ⋅ τ ult If the shear material thickness is greater than the minimum gage thickness. The shear material weight of the panel is found by: WShear (k ) = t Shear (k ) ⋅ TPan (k ) ⋅ bStruct (k ) The bending and shear material weights are summed across the wing to provide the total wing weight: WBend .Tot = Nseg k =1 ∑ 2 ⋅W Bend (k ) WShear . It is assumed that if multiple webs are used.
Tot+W Shear .000 Wing box weight 30.Tot The wing box is only part of the wing weight. Macci [1995] developed a semianalytical method for estimating wing weight. Other weights components include: • Attachments for engines.1 0 0 Wing box multiplication factors Attach fact CS Second fact Rib fact 10.3 0. WCS.8 0.000 Figure 81 Macci Wing Factors The average wing box multiplication factor value for each weight type was determined. Secondary = 0.6 0.9 0. to give the following weight estimation equations: WAttach = 0. the results of the analysis are helpful in determining the weight of wing components.5 0. are much more scarce.2 0. undercarriage. and miscellaneous items.Secondary • Rib Weight. WAttach • Leading and trailing edge control surfaces and secondary structure. Figure 81 shows the resulting multiplication factors applied to the wing box for other wing structural elements. manned or unmanned. While this method is not directly employed here.000 20. Detailed weight statements for aircraft.4 0.114 ⋅ WBox The overall wing weight is: 121 . WRibs Empirical equations for these components are difficult to find.303 ⋅ WBox WCS .The wing box weight is found by summing the total bending and total shear weights: WBox =W Bend .7 0. Wing box mulitplication factors 1 0.500 ⋅ WBox WRibs = 0.
Additional weight penalties will be incurred if a stiffness based wing design is necessary. Also.2 Tails The tail weight accounting includes all tail structures as a single weight. especially flutter. landing loads. so the piecewise linear beam model was employed. this may reflect reality because of limitations of processes to manufacture variable thickness skins at a given spanwise location. aerolastic effects. Although conservative.3. it will provide a reasonable estimate across all vehicles where no applicable competing methods exist. The hope is that. The tail weight methods are shown in Table 813. All of the wings are assumed to be designed for strength rather than for stiffness. The thicknesses are calculated at the points of maximum bending and shear stresses for a given wing segment. which simplifies the calculation of multiple tail arrangements.3.3. 122 .3 Piecewise linear beam model considerations The piecewise linear beam model contains both conservative and optimistic assumptions. drag loads. All the configurations under consideration have Vtails. The piecewise linear beam model and Raymer [1992] transport and general aviation tail weight methods are included. No single parametric wing weight equation is applicable across the wide range of vehicles considered in this research. which means that all other points have more thickness than required. The piecewise linear beam model is applicable to horizontal tails and Vtails. The horizontal tail weight methods functional dependencies are shown in Table 814. No direct consideration of aerolastic effects are considered here. raillaunch shock loads.WWing = WBox + WAttach + WCS . drag moments. despite the beam model limitations. or axial loads. One conservative assumption made in the model is that the wing skin bending material and shear material thicknesses are uniform for a given segment. These penalties are not applied in this code.1. this will vary from case to case. offnormal lift load vectors. can be a major driver in high aspect ratio wings. While these considerations may partially cancel one another. 8. The vertical tail weight methods functional dependencies are shown in Table 815. Secondary + WRibs 8.3. This model does not account for wing torsion.
Raymer [1992] General Aviation Tail Equations 3. the tail weight is calculated as a combination of horizontal and vertical tail weights when using parametric equations. The Vtail weight is calculated without these manipulations when applying the piecewise linear beam model. Piecewise Linear Beam Table 813 Tail Weight Methods Meth. 8.V Yes Yes q Yes Tech Yes Yes Table 815 Vertical Tail Weight Method Functional Dependencies For Vtails (also known as butterfly tails). The fuselage weight methods and associated functional dependencies are shown in Table 816 and Table 817. Raymer [1992] Fighter Fuselage 2. For many of the smaller UAVs considered.3.3 Fuselage Three fuselage methods have been adopted from Raymer [1992].3. Raymer [1992] Transport Fuselage 3. 1 2 WDG Yes Yes NUlt Yes Yes SH Yes Yes ARH Yes Yes LH Yes λH Yes Yes Λc/4. Raymer [1992] Transport Tail Equations 2.Method 1. Method 1. respectively.H Yes Yes Applied Here No No Yes Sc/SH Yes  q Yes Tech Yes Yes Table 814 Horizontal Tail Weight Method Functional Dependencies Meth. All other geometry parameters such as taper ratio. sweep. The Vtail weight is found by: WV −Tail = WHT ⋅ Cos (Γ ) + WVT ⋅ Sin(Γ ) 2 2 The aspect ratio for the vertical tail part of the calculation is assumed to be half of the input aspect ratio. these manned aircraft equations are not working within the intended applicable bounds. and thickness to chord ratio are identical. WDG 1 Yes 2 Yes NUlt Yes Yes SV Yes Yes ARV Yes Yes LV Yes  λV Yes Λc/4. Raymer [1992] General Aviation Fuselage Applied Here No No Yes Table 816 Fuselage Weight Methods 123 .
but this is not always included in the nacelle. 1 2 3 WDG Yes Yes Yes NUlt Yes Yes Yes Swet. The equation is dependent upon the engine controller weight. Given the limitations of the analysis and large spread of results. The parachute mass fraction method is a bestjudgment assumption. as an assumption. The Shadow 200 can use a parachute recovery system. and Shadow 200 in a spreadsheet. since many UAVs may not have tails attached to the fuselage or even have tails at all. The tail geometry imposes limitations on the design.16%. tail location on the fuselage.Meth. is assumed to be 75% of the maximum dynamic pressure. 124 . The nominal cruise dynamic pressure.5 Launch and Recovery Gear Multiple empirical methods exist for estimation of landing gear weight [Raymer 1992.3. WEngcont. is approximated as 60% of the fuselage length. 8. as a weight index to the conventional landing gear method. The fuselage is assumed to be unpressurized. 8.3.26%. but these methods typically have landing gear geometry functional dependencies. Nevertheless. is set to unity. The minimum landing gear mass fraction was 3.3. skid landing. KNG. 4% was selected as the basic landing gear mass fraction applied to all conventional landing gear in this research. so no pressurization weight penalties are incurred. The Global Hawk and Predator use conventional landing gear only. so the associated correction factor.28%. or parachute recovery. The average mass fraction of the landing gear relative to the design gross weight as evaluated was 4. and cabin pressurization. LTail. This method requires special information on the nominal cruise dynamic pressure.Fuse Yes Yes LTail Yes L/DFuse Yes* Yes Yes LFuse Yes Yes  q Yes Wing Geom Yes  Tech Yes Yes Yes Table 817 Fuselage Weight Method Functional Dependencies The Raymer [1992] General Aviation fuselage equation was used for all cases. The parachute recovery is considered here. Parachute recovery gear is assumed to weigh 2% of the design gross weight. Roskam 1990]. The engine controller weight was set to 10% of the engine weight for the purposes of these calculations. and the maximum was 5. the length of the tail from the wing. The pylon mounted option is not used. q.4 Nacelles The Raymer [1992] transport engine nacelle weight was applicable to both pylon and other arrangements.3. Recoveryonly gear can take the form of nonrecoverable (crash landing). Raymer [1992] landing transport landing gear weights equations were applied to the Global Hawk. Predator.
8.56 ⋅ PEng Wid Eng = 1.8.3.5 ⋅ H Eng The engine volume is found by: VolEng = LEng ⋅ Wid Eng ⋅ H Eng The power draw and cooling of the engine are set to zero for the standard component accounting. Jet engine weight – thrust to weight method The thrust to weight jet weight calculation method is: WJet = Jet engine weight – Roskam method TJet T / WJet 125 . These are calculated under the mission performance.15 + 0. This is calculated by: WEng = PSpecific ⋅ PEng The overall engine dimensions are found by Roskam [1990] 6cylinder engine formulas modified to have application beyond general aviation engines: LEng = 1.4.2 Jet Engines 0. A specialized means of calculating the jet engine volume was employed.4 Propulsion 8.7 ⋅ Eng + 0.00134 ⋅ PEng 200 P = 1.1 Reciprocating Engines The engine calculation method multiplies the specific power and the engine power to obtain the engine weight.00156 ⋅ PEng 200 PEng A simple thrust to weight method and a parametric method are used to determine the engine weight.4.3.3.
Unfortunately.Ref The jet engine volume is then found by: VolJet = Jet engine power and cooling π 4 LJet ⋅ DJet 2 126 . A limited engine database for turbojets and turbofans was used to determine the length/diameter2 parameter. The jet engine diameter equation is: D Jet = TJet (TJet / D 2 )Ref TJet TJet. Jet engine weight – default rubber engine method This method uses a default reference engine to calculate the engine weight. there is little correlation between bypass ratio and the parameter.Ref ⋅ SF Jet engine volume calculations For the thrust to weight and Roskam [1990] engine weight methods. This method relates the jet engine weight to the thrust level and bypass ratio. the engine diameter is found by using a reference thrust/diameter2 parameter.The Roskam [1990] jet weight calculation method is available in the code. The default rubber method uses the length and diameter methods described earlier. and there is significant scatter. there was little correlation with aspect ratio. This method was not used to generate the results of this effort. and diameter. length. The thrust scale factor is found by: SF = The jet engine weight is found by: W Jet = W Jet. Similar to the thrust/diameter2 parameter. A nominal value of 500 lb/ft2 was used for the purposes of these simple calculations. The jet engine length equation is: L Jet = D Jet ⋅ (L / D ) Jet.Ref The engine length is found by a length/diameter parameter. The engine database was used to select an appropriate value. and the data has significant scatter.
127 . The impacts of engine cooling are calculated in the performance model. These are covered in the mission performance chapter.The jet power consumption and cooling are set to zero in the component calculations.
These segments can represent climb. and the second is suitable for an objective function contributor. The figures of merit include range efficiency. specified altitude limits. The performance code is designed to evaluate the fuel burn of the AV as it flies the mission profile and to evaluate associated constraints. climb efficiency. The code evaluates multiple mission segments. During each segment of flight. ranging from direct evaluations of a specified mission to limited mission optimization. The mission analysis evaluates a wide array of mission types with an emphasis on flexibility and robustness.1 Overview The performance model predicts aircraft fuel burn and optimizes flight trajectories. and maximum rate of climb. descent. Several options exist for evaluating the mission. specified Mach limits. The first possibility lends itself well to a mission completion constraint. cruise. specified dynamic pressure limits. The individual segments are further broken into subsegments.Chapter 9 Flight Performance 9. and engine operating limits. Second. First. Simple performance methods are combined in multiple ways to solve complex performance problems without the need for external flight trajectory optimization. 128 . several constraints are evaluated. standard or arbitrary mission profiles can be evaluated either forwards or backwards to determine the percentage of the mission that could be completed. loiter or combinations thereof. a loiter segment can be established in the middle of a mission profile with sufficient endurance to use all of the available fuel. Two possibilities exist for the evaluation of fuel burn. insufficient thrust for level flight. These velocities are optimized such that no constraints are violated. insufficient thrust for prescribed rate of climb. endurance efficiency. These include stall. The performance code has the ability to optimize flight velocities.
PERFORM calls the necessary subroutines to evaluate the fuel burn and mission constraints. the aerodynamics and propulsion tables must be generated. The header of the mission input file has the following format: 129 . MISSOPT has the ability to drive the evaluation of a normal onedirection mission or an optimized centered mission. PSEGSEG calls WTFINSEG multiple times to determine the fuel burn or performance and any remaining unknowns. weights. PSEGSEGO calls PSEGSEG multiple times to determine optimal velocities that do not violate constraints. the main performance subroutine. ACMAIN initially calls PERFORMIN to read and interpret the mission input data files. PERFSEG subdivides the mission segment into subsegments and calls either PSEGSEG for prescribed velocities or PSEGSEGO to determine optimized velocities. With either option. along with others. Once this is done. PERFORM. first calls MISSOPT to organize the mission. the mission parameters must be input. and the weights must be determined. and aerodynamics subroutines.2 Performance Code Architecture Figure 91 Performance Code Architecture Figure 91 shows the performance code architecture. geometry. to generate the data necessary to drive PERFORM. MISSOPT calls PERFSEG for every mission segment. ACMAIN then calls EVAL directly or through the optimizer. The PERFORM subroutine drives the performance evaluation. Prior to calling PERFORM.9. EVAL calls the propulsion.
This does not include the climb to altitude.initial altitude (if contig_flag=2). 3=climb) 20.6=h/f) 150. 2=endur inpt. and takeoff.4=const) 30000. and nseg is the total number segments in the mission. nmi 1. hrs vm_flag . 2=reverse.99 4 8 miss_type (1=nonopt. 3=center) ra_en_flag (1=range inpt. and a value of 2 indicates that the mission is centered and that all the fuel will be depleted during the midmission loiter. or from a centered mission.2=ROC. ft 500. ft altroc_flag (1=fin alt.2=M.5=h/t.3=cruise. If climb is the primary objective. Rseg. Range.4=E. backwards from landing (at the empty weight) to takeoff. rangeseg (if ra_en_flag = 1). nsegseg is the number of subsegments each segment will be divided into. A centered mission must follow a centered mission profile described later.Test performance input file 2 0. ff_taxi is the fuel fraction that is used during warmup.3 mach (if vm_flag = 2) contig_flag (1=contiguous. vel (if vm_flag = 1). alt2 (if altroc_flag = 1). Endurance. is input if endurance is the primary objective. endurance.3=R. then climb between two specified altitudes will be achieved either with maximum rate of climb or maximum fuel efficiency. taxi. 2 = not contig) 0.velocityMach flag (1=V. alt1 . kts 0. Eseg. The toggle flag ra_en_flag indicates whether range. A value of 1 indicates a normal onedirection mission. The remainder of the mission input file represents the individual mission segments. ft/min The selection flag fwd_flag indicates whether the mission is flown forward from takeoff (at design gross weight) to landing. 130 . is input if range is the primary objective. endurseg (if ra_en_flag = 2). roc (if altroc_flag = 2). or climb is the primary objective of the mission segment.2=R spec) ff_taxi nsegseg nseg The toggle flag miss_type specifies the type of mission to be flown. Each segment has the following format: 3 1 1 3 1 1 segment number fwd_flag (1=fwd.
The altitudes must be input directly when the segment is not contiguous. The product may not hold constant well if the altitude change is significant. depending on which information is known a priori. The toggle flag contig_flag indicates whether the mission segment is contiguous with an adjacent mission segment. 9. The prediction accuracy can be improved by breaking each mission segment into subsegments and introducing corrections. The Breguet range equation is modified to account for thrust values differing from the ideal. Either the initial or final altitude may be set to the appropriate altitude of the adjacent segment when contiguous. The options for optimized flight velocities include the velocities associated with best range. Many missions are flown at constant altitude or velocity. depending on the direction in which the mission is evaluated. The selection flag altroc_flag indicates whether the final altitude or constant rate of climb is input.The selection flag vm_flag indicates how the flight velocity will be determined.1 Performance Equations The heart of the performance modeling for fueled aircraft begins with the Breguet range equation: R= V ⋅ L / D Wi ⋅ ln W TSFC f This form of the Breguet equation requires that the product V*L/D/TSFC remains constant. 131 . creating inaccuracies in the Breguet range equation estimation. endurance. where constraints may be violated.3.3 Performance Methodology 9. Climbs differing from the cruise climb can also introduce inaccuracies. R = ∑ Rseg The equations that follow may be calculated in several different orders. The flight velocity or the Mach number may be input directly. The flight velocity may be optimized such that no constraints are violated. or most efficient climb. rate of climb. which necessitates a cruiseclimb.
The Reynolds number per unit length. The required thrust becomes: Treq = The lift coefficient. ρ. is found by: Tmult = The required thrust. is found via the standard atmosphere subroutine. is found by: Re / L = ρ ⋅V µ where the viscosity and density are found via the standard atmosphere subroutine. 132 . Treq. CL. M . is found via a standard atmosphere subroutine. L/D. is simply: CL = W 1/ 2 ⋅ ρ ⋅ V 2 ⋅ Sw ROC ⋅W + D V Treq D The air density. The speed of sound. is found by: 1 ROC Treq = + ⋅W L/ D V The lift to drag ratio. Re / L) where. Re/L. is replaced by the ratio of weight over drag. a. M = V a or V = M ⋅a depending on whether Mach number or velocity is known beforehand. The drag coefficient is found by the aerodynamics table lookup subroutine: C D = C D (C L .Rseg L V ⋅ W D = ⋅ ln i Tmult ⋅ TSFC W f where the thrust multiplication factor. Tmult.
M . and V is in knots. is found via the engine deck lookup subroutine. This method is a simple approximation for the equivalent drag resulting from power extraction. When this relationship is rearranged and corrected for units. the initial and final weights can be input and the endurance output. and is assumed to be applicable to jet engines as well. Tmin and Tmax. the final weight is found by: 133 . then there will be power generation efficiency. are also output from the engine deck lookup subroutine. For example. For a forwardcalculated mission segment. the segment range and initial weight can be input and the final weight output. The overall drag including all considerations becomes: D= CD ⋅ W + DPow + DCool CL The weight calculations can occur in different orders. The approach lends itself well to reciprocating engines. is: Pow = D ⋅ V If a generator supplies this power.The specific fuel consumption. depending on the available inputs and required outputs. Or. ηGen. In general. assuming thrust is equal to drag. TSFC. The cooling drag estimation is described in the aerodynamics chapter. power extraction equivalent drag is found by: DPow = 0. TSFC = TSFC (alt .43687 ⋅ Pow V ⋅ηGen where DPow is in pounds. The relationship between drag and power. The time taken to complete the subsegment range is: E seg = Rseg V The performance modeling must also capture the impacts of cooling drag and power extraction. respectively. The mission performance code is structured such that multiple combinations of inputs and outputs may be accommodated. Treq ) The minimum and maximum thrust values at the given altitude. Pow is in Watts.
Wf = e i Tmult ⋅ R seg ⋅TSFC C V⋅ L CD W For a reversecalculated mission segment. the initial weight is found by: Tmult ⋅ R seg ⋅TSFC C V⋅ L CD Wi = W f ⋅ e 9.max 4) V > ρ / ρ SL 5) V > M max ⋅ a 6) Treq < Tmin 7) alt > alt max Stall constraint Minimum rate of climb constraint Insufficient thrust for level flight constraint Dynamic pressure constraint Mach constraint Minimum available thrust constraint (not set yet) Maximum altitude constraint Figure 92 shows potential flight envelope constraints.3. Figure 92 Flight Envelope Boundaries 134 .2 Performance Constraints There are multiple flight envelope constraints: 1) V < Vstall 2) ROC avail < ROC min 3) D > Tmax Veq .
Point 1 is created halfway between points min and mid.3. and point 2 is created halfway between points mid and max. The case where the best figure of merit lies on the end will be discussed later.3 Flight Velocity Optimization To optimize the velocities. The right and left adjacent points become points min and max. Depending on which constraints are violated. The initial maximum velocity is the minimum of the dynamic pressure limit and the Mach limit. Figure 93 shows how the search is initiated. then the point with the best figure of merit becomes the point mid. The initial minimum velocity is the stall speed. there are several figures of merit: FOM = FOM = Rseg Wi − W f E seg Wi − W f Range efficiency Endurance efficiency Maximum climb rate Climb efficiency FOM = ROC avail ROC ⋅ E seg FOM = Wi − W f A search algorithm finds the optimal velocity for the given figure of merit in the constrained flight envelope. the search begins. the minimum and maximum velocities fall outside the constraint boundaries. respectively. the minimum velocity is incrementally increased or the maximum velocity is incrementally decreased until the boundaries fall within the feasible region.9. The process continues through several 135 . If the point with best figure of merit does not lie on the end. In this case. Figure 93 Velocity Search Initialization Once all the velocities fall within the feasible region. In a trade between speed and robustness. a bisection search algorithm is used.
The cruise climb altitude change is found by: alt 2 = alt1 + 10. 9.000 feet. 136 . For forwardcalculated cruise climb segments. The reference altitude change is 10. the altitude used for calculations is the starting altitude. The process is different when the best figure of merit is an end point. the fuel burn is first calculated without any rate of climb. When a cruise climb profile is selected. The difference in fuel burn for the level flight and cruise climb conditions is found by first estimating the difference in thrust required for cruise climb: ∆TClimb = ROC ⋅W V Next. In the case where the best figure of merit is point min. The initial cruise altitude for the second through final subsegments is simply the altitude associated with the best cruise condition of the predecessor segment. The 80% factor is applied to account for typical reductions in available rate of climb as altitude is increased. The optimal cruise velocity is found for each new subsegment. reversecalculated cruise climb segments use the final altitude.3. Similarly.000 ⋅ Wf ⋅ − 1 ρ (alt1 + 10. The average altitude is output. The velocity associated with the best figure of merit at the end of the search is the final velocity selected. point mid becomes point max and point 1 becomes point mid. the cruise climb rate of climb is set to 80% of the available rate of climb. If the cruise climb rate of climb exceeds the 80% of the available rate of climb at the given velocity.000) − ρ (alt1) Wi ρ (alt1) The rate of climb associated with the cruise climb is estimated.4 Climb Performance The cruise climb altitude change is assumed equal to the altitude change required to keep the lift coefficient constant. The final subsegment altitude is found through linear interpolation or extrapolation. The process continues with one or more of the processes discussed until the maximum number of iterations is reached. the difference in fuel burn for the cruise climb is estimated by: ∆WFuel.iterations.Climb = ∆TClimb ⋅ SFC ⋅ E Seg Note that ESeg is the endurance of the subsegment. Points 1 and 2 are generated halfway between these points as described above.
The performance code will output the violated constraint value if the required rate of climb is not attainable. then the rate of climb will be set to the maximum rate of climb. The mission analysis code can evaluate the climb performance without a prescribed range or endurance. avail = (Tmax − D ) ⋅ V ⋅ 6080 W 60 For descent. The two desired climb modes are maximum rate of climb and most fuel efficient climb. if the required rate of descent exceeds the minimum rate of climb. exceeds the maximum altitude. and the range and endurance will be output. the minimum rate of climb (positive rate of descent) is: ROCmin. Even if a rate of climb is input. The associated velocities are found via the outer loop figure of merit search procedure described earlier. then the final altitude is set to the maximum altitude. avail = (Tmin − D ) ⋅ V ⋅ 6080 W 60 The rate of climb may be reset for missions segments that permit flexibility in the rate of climb. Unlike the flight velocities. then the rate of climb is set to the minimum rate of climb. it may not be achievable. For most fuel efficient climb. alt2. The flight velocity may be optimized for any figure of merit desired. If the required rate of climb exceeds the maximum rate of climb. the altitude calculations do not optimize altitudes such that no constraints are violated. For a forward evaluation: alt 2 = alt1 + ROCseg ⋅ Eseg ⋅ 60 For a backwards evaluation: alt1 = alt 2 − ROCseg ⋅ Eseg ⋅ 60 If the initial and final altitudes are specified. The initial and final altitudes must be input. For the maximum rate of climb. Similarly.A constant rate of climb may be specified for a mission segment. maximum throttle is used. The maximum available rate of climb is: ROCmax. a search procedure is initiated with the following figure of merit: 137 . then the rate of climb is found by: ROCseg = (alt 2 − alt1) Eseg ⋅ 60 If the final altitude.
The endurance and range are then found by: Eseg = alt 2 − alt1 ROCseg ⋅ 60 and Rseg = Eseg ⋅ V The initial or final subsegment weight. is then found via the modified Breguet range equation. depending on evaluation direction.FOM = ROCseg Wi − W f The initial and final weights are evaluated with the Breguet range equation with the segment range missing. 138 . thus eliminating the range and endurance dependency.
Arbitrary missions are much more flexible.3. The mission completion criterion is: misscomp = 1 − W f − Wi Wi 139 . The arrows indicate the direction of flight. then the time on station will be negative.9. 9. The order of calculation is the same as the order of segments for forward calculations or in the opposite order for backwards calculations. Figure 94 Standard Mission Profile Figure 94 shows the standard mission profile used for longrange missions.6 Mission Completion Criterion The mission completion criterion is a metric for determining if the air vehicle with the available fuel load can complete the desired mission. These can contain up to 20 segments and the compiler integer limit for subsegments per segment.5 Mission Evaluation Mission profiles can be either standard or arbitrary. The underlined numbers show the order of calculation for a centered mission.3. The centeredstandard mission completion criteria depends on the available fuel at the loiter segment. The mission completion criterion is handled differently for centeredstandard and arbitrary mission profiles. The boxes identify the mission segment name and the flight order. If the loiter segment’s final weight is greater than the initial weight. The standard mission profile is useful because it permits an evaluation of available time on station for a given fuel load.
Otherwise. max Denom ∆alttot ⋅ 0.75 ⋅ Veq . then an error flag is output along with the amount of the segment that is completed. The range. Then each distancenormalized parameter is weighted. the relative weightings are found: Rwt = Rtot Denom Ewt = Etot ⋅ 0. and this may be an objective function. the mission completion criterion is found by: misscomp = W fuel W fuelb A different procedure is used when there is insufficient fuel to complete the arbitrary mission. The mission analysis code keeps track of three different types of mission segment: range. All segments that follow 140 . The time on station is a direct result of the available fuel. endurance. if the loiter segment initial weight is greater than the final weight. endurance. and climb. The arbitrary mission completion criterion evaluation method depends on if the AV completes the mission. the relative weighting of each metric must be assessed. the common denominator is found: ∆alttot ⋅ 0. Although all three parameters are known for each segment. First. and climb are calculated: Rtot = ∑ Rseg For all rangetype segments For all endurancetype segments For all climbtype segments ∆alttot = ∑ ∆alt seg Etot = ∑ Eseg Next. only one per segment is tracked for the mission completion criteria. and climb are all put in terms of distance. max Denom = Rtot + Etot + ROCmin ⋅ 60 Second. endurance.75 ⋅ Veq .75 ⋅ Veq . the mission completion criteria is unity. First. If the AV completes the mission from either the forward or backward analysis direction. the total range. max ROCmin ⋅ 60 alt wt = Denom If the air vehicle depletes the fuel during any mission segment.
There is a running summation of the completed and partially completed segments.are not completed. The mission completion parameter for noncompleted arbitrary missions is: misscomp = Rwt ⋅ Rcomp Rtot + Ewt ⋅ Ecomp Etot + alt wt ⋅ ∆altcomp ∆alttot 141 .
Installation weight design level 4.2 Calibration Methodology Design code calibration is a challenging endeavor. It is highly unlikely that a design algorithm will provide sensible results without adequate tailoring of inputs and methods. design gross weight. Volumetric closure is a function of component volumes. Factor of safety design level (if not specified) 2. Vehicle Geometry 2. It is possible to reach a point calibration through many different approaches.1 Calibration Overview Calibration Design code calibration is essential to ensuring the validity of a new design analysis. Structural materials characteristics 6. 10. aerodynamics performance. Component SWaP or SWaP method inputs 4. The mission completion is a function of weights. In this effort. structural loading. Fuel fraction input 5. The calibration process starts with the following inputs: 1. flight envelope. The design gross weight is a function of empty weight and the fuel weight. If the empty weight is higher than the reference value. Subsystems parameters and characteristics 7. payloads. The primary outputs that must be matched are empty weight. and the selfoptimized mission profile. the technology design impacts may be assessed with a higher degree of confidence. propulsion performance. The empty weight is affected by the avionics and subsystems. and numerous other inputs. fuel volume and available volume. fuel capacity. Structural materials characteristics (if appropriate) 142 . Avionics and subsystems characteristics (if appropriate) 5.Chapter 10 10. and mission completion. Mission profile The vehicle design code is run with all available inputs. Once calibrated. Weight growth design level 3. The present design code was calibrated for the three UAV classes under consideration. Loads and flight envelope parameters 3. then several items may be adjusted: 1. volumetric closure. there are more code inputs than available calibration point references.
Global Hawk started as an Advanced Concept Technology Demonstrator (ACTD) with many goals. Many modifications have occurred to improve the system. Uninstalled engine characteristics (if unknown) 2. As with nearly any aircraft program. the available performance numbers represent the estimated performance of the vehicle as built.1 Global Hawk Description The Global Hawk is the first and only operational strategic high altitude UAV.1. Aerodynamics design levels If volumetric constraints are violated.3.1 Global Hawk 10. An attempt is made to calibrate the code against a representative Global Hawk.2 Geometry Characteristics A rendering of Global Hawk is shown in Figure 101. then the external geometry of the vehicle and component volume estimations must be scrutinized for validity. the performance calibration begins. and other considerations.1. the performance changes over time due to weight growth. not necessarily as designed. Propulsion installation losses 3.3. The calibration cases utilized the calibration process outlined above. If this is not sufficient. The fuel mass fraction is held constant at the reference value. The parameters adjusted during the calibration process and any unique methods employed are identified for each of the three cases evaluated here.3 Calibration Cases 10. This system began development in 1994. but the only firm requirements was a fixed Unit Flyaway Price (UFP). 143 . 10. Therefore. and major geometry characteristics are described in Table 101. 10. system modifications. The following items may be adjusted to ensure that the design mission profile is completed: 1. then the volumetric design level must be adjusted.Once the empty weight at the reference fuel mass fraction is calibrated.3. and it has experienced operational use in wartime.
88 ft 3.63 ft Source Janes [1999] Janes [1999] Janes [1999] Janes [1999] Janes [1999] Table 102 Global Hawk Propulsion In addition to the engine.33 lbm/lbh 1. The results were integrated into the detailed geometry input files. Item Thrust (TO S/L) TSFC Weight (Dry) Length Diameter Value 8.1. an additional 50 pounds of propulsion weight was added to account for the engine control electronics and actuators.2 ft 25 5o 44. 10. as an assumption.3 Propulsion The Global Hawk engine is the RollsRoyce 3007H.4 ft Source Janes [1999] Janes [1999] Janes [1999] Janes [1999] Table 101 Global Hawk Geometry Detailed geometry characteristics were found through scaling of 3view drawings.3.581 lbs 8.290 lbs 0.Figure 101 Global Hawk Item Wing span Aspect ratio Sweep (quarter chord) Fuselage Length Value 116. 144 . Major engine characteristics are shown in Table 102.
1. developed from information generated from Altmann [2002] and Janes [1999]. The calculated propulsion performance is shown in Figure 102 and Figure 103. are found in Global Hawk literature and vendor data sheets. The assumed avionics weights use a fragmentary Master Equipment List (MEL). kts 300 400 Alt = 0 kft Alt = 16kft Alt = 33 kft Alt = 49 kft Alt = 65 kft Figure 102. and the thrust specific fuel consumption factor is assumed to be 0. These calibration factors were selected through the performance matching process.The overall propulsion installation thrust installation factor is assumed to be 0. lbs 7000 6000 5000 4000 3000 2000 1000 0 0 100 200 Velocity. Global Hawk Installed Thrust Performance Maximum Thrust. Weights for all of the components are not available.4 Avionics Global Hawk is known to have an extensive electronics suite.65. Figure 104 shows the avionics weights determined for the calibration case. Details of some avionics components.3. 145 .8. as guidance. such as INS and data recorders. Global Hawk Engine Thrust Performance Figure 103 Global Hawk Engine Specific Fuel Consumption Performance 10.
6 pounds each). Autopilot The autopilot weight consists of two Integrated Mission Management Computers (IMMCs). as a bestjudgment assumption. This weight was the bestjudgment assessment of the weight of all avionics for which detailed weight information is not available. Global Hawk Avionics Weights Summary General Avionics The general avionics category is intended to capture all avionics components not covered by specific avionics categories. No weights information is readily available on these units. which is a bestjudgment assumption. lbs 200 150 100 50 0 General Avionics Autopilot INS GPS ADS ATC Recorder Figure 104. 300 pounds of general avionics was assumed. The custom air data system weight equation was used. Air Data System The air data system consists of two air data booms on the tails. The GPS must support Differential GPS (DGPS) functionality. and two Kearfott KN4072 units (10 pounds each). The weight is 40 pounds. The GPS weight was set to 5 pounds. Air Traffic Control 146 . INS The INS weight consists of two LN100 INS units (21.Global Hawk Avionics Weight 350 300 250 Weight. most rugged design. The equation is set to have the largest size. GPS The GPS weight consists of the GPS antenna and a GPS processor. and highest performance settings.
Global Hawk has a robust air traffic control system. Global Hawk is the first UAV to gain a type certification, enabling a file and fly capability. Global Hawk is known to have VHF/UHF ATC radios, and Identify Friend or Foe (IFF) avionics [Altmann 2002]. The ATC weight was set to 50 pounds, which is a bestjudgment assumption. Recorder Global Hawk has a data recorder capability. L3 claims that its S/TAR family of data recorders is installed on the Global Hawk [L3 2003]. These recorders have an advertised weight range of 1933 pounds, and a weight of 25 pounds is selected for the Global Hawk calibration. It is assumed that two of these units are installed, given the significant redundancy on the aircraft.
10.3.1.5 Subsystems
Global Hawk has a complex set of subsystems. A list of known subsystems identified by Altmann and Janes is captured in the simple MEL. Unfortunately, no weights data is available for the subsystems. Therefore, no actual weights were used, only assumed subsystem weights and parametric methods. The resulting subsystems weights are shown in Figure 105.
Global Hawk Subsystems Weight
450 400 350
Weight, lbs
300 250 200 150 100 50 0 FCS ECS Electrical System Fuel System Communications
Figure 105 Global Hawk Subsystems Weights Summary
Flight Control System (FCS) Global Hawk has a redundant flight control system. Per vehicle side, Global Hawk has two ailerons, two spoilers, and two ruddervators. Since the flight control system includes the actuators, the weight was assumed to be twice the parametric flight control system weight to account for the redundancy. The custom FCS weight estimation equation (Flight Control Method 7) was used to estimate this weight.
147
Electrical System Global Hawk has a redundant electrical power system. There are two flight critical buses, two start buses, three batteries, an AC and a DC generator, power converters, and interfaces to ground power. The Raymer [1992] transport electrical system equation (Electrical System Method 6) was used, with two generators input. Fuel System Global Hawk has a sophisticated fuel system that interacts with the environmental control system. The aircraft has two high capacity fuel tanks and one sump tank in the fuselage, and three wing tanks. In addition to feeding the engine, the fuel is routed through various avionics, payload, and other bays to provide thermal management. The fuel is actively pumped through the system. The Raymer [1992] transport fuel system weight equation (Fuel System Method 2) was employed. Environmental Control System (ECS) The environmental control system is linked to the fuel system. Fuel cools the avionics and other components via heat exchangers and other means. The Roskam [1990]General Dynamics lowsubsonic aircraft ECS weight method (ECS Method 2) was employed. The performance model takes no drag penalty or propulsion penalty for the ECS system, since there is no ram air or engine bleed air used for thermal management. Communications System The Global Hawk communication system consists of multiple communications links and associated equipment. There is a high bandwidth satellite communications link using a 48inch parabolic dish for payload data downlink. There are two low bandwidth satellite communications links for command and control. Global hawk has air traffic control links that are accounted for under ATC equipment weights. There is a lineofsight communications link between the aircraft and the launch and recovery element to support launch and recovery operations. The custom dish weight equation is used to estimate the high bandwidth link parabolic dish weight. Additionally, 100 pounds of communications equipment weight was input to estimate the weight of all other communications equipment, which is a bestjudgment assumption used in the absence of better information.
10.3.1.6 Structures
The Global Hawk structure consists of the main wing, tails, fuselage, nacelle, landing gear and installation weight. No direct weights data is available for the structure. However, Altmann [2002] provides useful information to describe the structural design drivers and philosophy. The factor of safety for the structure is 1.25. Altmann provides a
148
VN diagram that indicates that the light weight vertical load is approximately 3.6 G, and the heavy weight vertical load is approximately 2 G. Because the wing weight is calculated at gross weight, the vertical load is assumed to be 2 G. The Global Hawk structural weight is presented in Figure 106.
Global Hawk Structural Weight
2,500
2,000
Weight, lbs
1,500
1,000
500
0 Wing Tail Fuselage Nacelle Gear Installation
Figure 106 Global Hawk Structural Weight Summary
Wing The wing is made of Carbon Fiber Reinforced Plastic (CFRP) [Jane’s 1999]. The piecewise linear beam model is used to estimate the wing weights. A “black aluminum” formulation was used. The base material is the 2017 Aluminum alloy with a minimum gage of 0.02 inches. A composites weight reduction factor of 0.85 was used, as suggested by Raymer [1992]. Tails The tail consists of a single Vtail made of unspecified composites [Jane’s 1999]. The piecewise linear beam model is employed. As with the wing, 2017 aluminum with a minimum gage of 0.02 inches was assumed for the reference material. A composites weight reduction factor of 0.83 was used for the tails, as suggested by Raymer [1992]. Fuselage The fuselage is essentially an aluminum alloy box structure [Jane’s 1999]. The Raymer [1992] General Aviation method was used. No credit was provided for composites. Nacelle A single nacelle houses the Rolls Royce 3007H engine. The Raymer [1992] transport nacelle equation was used for weight estimation. The resulting nacelle weight may be low, though there is no available weights information for comparison. Landing Gear 149
The Global Hawk landing gear consists of a retractable nose gear mounted in the fuselage and a twostrut retractable main gear mounted in the wing root. The custom gear weight equation was used.
10.3.1.7 Payloads
Global Hawk payloads consist of a Synthetic Aperture Radar (SAR), an ElectroOptical/ Infrared (EO/IR) payload, and the supporting electronics. The supporting electronics include an integrated sensor processor, a receiver/exciter/controller unit, transmitter (for SAR, presumably), and a sensor electronics unit. It is unclear if the elements of the communications architecture, INS, or structure are included in the advertised payload weight of 1,900 pounds [Jane’s 1999]. The summation of the listed components comes to 797 pounds [Jane’s 1999]. There is no available source that clarified this discrepancy. To satisfy the sizing mission profile, 1,900 pounds was assumed for the total payload weight, with an even weight division between SAR and EO/IR.
10.3.1.8 Overall Weights
The available information in Jane’s [1999] on Global Hawk weights does not contain any noticeable internal discrepancies. According to Jane’s [1999], the empty weight is 9,200 pounds, the maximum fuel weight is 14,500 pounds, the payload weight is 1,900 pounds, and the maximum takeoff weight is 25,600 pounds. The sum of the components comprising the maximum takeoff weight equals the maximum takeoff weight. There is conflicting information from other sources on Global Hawk weights. Northrop Grumman [2003] lists the gross takeoff weight at 26,700 pounds. Due to typical weight growth of aircraft, both weights may be true for the associated reporting time. The Jane’s [1999] weight is assumed as correct for the purposes of sizing. The payload weight of 1,900 pounds, and the fuel fraction associated with Jane’s [1999] fuel weight and maximum takeoff weight were input. The target design gross weight is assumed to be equal to the maximum takeoff weight. The calibration process sought to match the empty weight and design gross weight. The empty weight matches within 0.2%, and the design gross weight matches within 0.2%. The Global Hawk weight summary is shown in Figure 107.
150
Global Hawk Weight Summary
100,000 10,000
Weight, lbs
1,000 100 10 1
Av io ni cs
St ru ct ur es
G ro wt h
on
ad
Fu el
Su bs ys te m s
Pr op ul si
Pa ylo
Em
Figure 107 Global Hawk Weights Summary
The mass fractions are shown in Figure 108.
Global Hawk Mass Fraction Summary
60% 50%
Mass Fraction
40% 30% 20% 10% 0%
Av io ni cs
St ru ct ur es
G ro wt h
on
ad
Fu el
De sig n
Su bs ys te m s
Pr op ul si
Pa ylo
Figure 108 Global Hawk Mass Fraction Summary
10.3.1.9 Aerodynamics
The predicted Global Hawk incompressible trimmed Lift to Drag (L/D) over the range of Reynolds number per unit length is shown in Figure 109.
151
Em
pt y
G ro ss
pt y
Global Hawk Drag Contributors.014 0. All other aerodynamics methods are standard.6 0.2 Lift Coefficient Re/L = 73.2 Cdprof CD ind CD fuse CD int CD Wave Lift Coefficient Figure 1010 Global Hawk Trimmed Drag Contributors Several aerodynamics methods were used to generate the drag polar.016 0. 152 .Global Hawk Incompressible Aerodynamic Performance 35 30 Lift to Drag Ratio 25 20 15 10 5 0 0 0.547 Re/L = 87.010 0.8 1 1.2 0.6 0.7 Drag Coefficient 0. The Trefftz plane analysis was used to determine the trimmed lift distribution and induced drag.4 0.133 Re/L = 124.012 0.008 0.002 0.006 0.8 1 1. A static margin of 5% was assumed.952 Figure 109 Global Hawk Trimmed Incompressible Aerodynamic Performance The Global Hawk drag contributors by drag type at Mach 0.000 0 0.2 0. A 5% miscellaneous drag factor was applied. Mach = 0.7 at the maximum bounding Reynolds number are detailed in Figure 1010.4 0.250 Re/L = 174.004 0. The parametric airfoil method was used to determine the profile drag.
Global Hawk Altitude Profile 70. and 200 nautical miles from the end of cruise to the final loiter altitude. A halfhour loiter at 5.000 40.000 Altitude. ft 50.200 nautical miles radius [Northrop 2003]. hrs Figure 1011 Global Hawk Altitude Profile Global Hawk Mach Profile 0.3. not the boundaries of the flight envelope.000 10. Figure 1013 shows the same data points as Figure 1011 and Figure 1012.000 30.7 0.10 Performance Altmann [2002] provides useful information on the Global Hawk performance and flight envelope limitations. Range credit was assumed to be 100 nautical miles for the initial climb to 50.000 feet.10. Northrop Grumman advertises the Global Hawk Performance as 24 hours time on station at 1.3 0.1. and Figure 1013. Altitude and Mach characteristics are shown in Figure 1011. and the maximum Mach is approximately Mach 0. This performance estimate was adopted for sizing. Figure 1012.1 0 0 5 10 15 20 25 30 35 Time.000 20.6 0. hrs Figure 1012 Global Hawk Mach Profile 153 . The maximum equivalent airspeed is 175 Keas. Sudden changes in the Machaltitude profile occur during transitions from one mission segment to another.000 feet was assumed for airfield operations.2 0.5 Mach 0.000 0 0 5 10 15 20 25 30 35 Time. Note that Figure 1013 shows the Mach and altitude points flown in the optimized mission profile.4 0. such as changing from loiter to egress.7.000 60.
Global Hawk MachVelocity Envelope Used
70,000 60,000
Altitude, ft
50,000 40,000 30,000 20,000 10,000 0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Mach
Figure 1013 Global Hawk MachAltitude Envelope
10.3.1.11
Design Technology
The weights and performance calibration process resulted design technology levels are shown in Table 103.
Design Item Volume Efficiency Induced Drag Interference Drag Wave Drag Laminar Flow Factor Of Safety Weight Growth Installation Weight Tech Level (01) 0.5 0.31 1.0 0.31 0.31 1.0 0.35 0.5
Table 103 Global Hawk Design Technology Levels
154
10.3.2
Predator
10.3.2.1 Predator Description
Predator is a MediumAltitude Endurance (MAE) UAV designed to provide battlefield surveillance with a beyond line of sight communications capability. This aircraft is an evolution from the General Atomics Gnat UAV. The Predator program began in 1994 as an ACTD. The program transitioned to operational use very early in development.
10.3.2.2 Geometry Characteristics
The Predator key geometry characteristics are shown graphically in Figure 1014, and numerically in Table 104.
Figure 1014 Predator Item Wing span Aspect ratio Sweep (quarter chord) Fuselage Length Value 48.7 ft 19.25 0o 26.7 ft Source Jane’s [1999] Jane’s [1999] Jane’s [1999] Jane’s [1999]
Table 104 Predator Geometry Characteristics
155
10.3.2.3 Propulsion
Predator uses the Rotax 914 reciprocating engine to drive a pusher propeller. Major engine characteristics are presented in Table 105.
Item Maximum Power (S/L) BSFC Weight Value 105 HP 0.5 lbm/HPhr 150.4 lbs Source Jane’s [1999] Assumed Jane’s [1999]
Table 105 Predator Propulsion Characteristics
The predicted installed predator engine performance is shown in Figure 1015 and Figure 1016.
Predator Installed Thrust Performance 350
Maximum Thrust, lbs
Alt = 0 kft Alt = 6.25 kft Alt = 12.5 kft Alt = 18.75 kft Alt = 25 kft
300 250 200 150 100 50 0 0 50
Velocity, kts
100
150
Figure 1015 Predator Installed Thrust Performance
156
Predator Installed Thrust Specific Fuel Consumption
0.5
Alt = 0 kft Alt = 6.25 kft Alt = 12.5 kft Alt = 18.75 kft Alt = 25 kft
TSFC, lbm/lbhr
0.4 0.3 0.2 0.1 0 0
50
100
150
Velocity, kts
Figure 1016 Predator Installed Specific Fuel Consumption Performance
10.3.2.4 Avionics
Predator has a relatively simple avionics suite compared to Global Hawk. Predator is largely a singlesting system with little redundancy. A summary of the Predator avionics weights is presented in Figure 1017.
Predator Avionics Weights
25 20
Weight, lbs
15 10 5 0 General Autopilot INS GPS ADS ATC Recorder
Figure 1017 Predator Avionics Weights Summary
General Avionics The Predator general avionics category is intended to capture nonallocated avionics weights. The general avionics weight allocation was 5 pounds, which was a bestjudgment assumption used in the absence of better information.
157
Autopilot Details of the autopilot physical characteristics are not available. The Predator autopilot must support both autonomous flight and pilot commanded maneuvering. The autopilot weight allocation was 10 pounds, which was a bestjudgment assumption used in the absence of better information. INS Predator has a single Litton LN100G INS, weighing 21.6 pounds. GPS Predator has a GPS capability, and 1 pound was allocated to the receiver. Unlike Global Hawk, Predator does not land autonomously with the aid of DGPS. Air Data System The air data system (ADS) was calculated via the custom method. The ADS is assumed to have mediumlow ruggedness, a mediumhigh level of miniaturization, and mediumlow performance. The resulting ADS weight is 4.28 pounds. Air Traffic Control Predator has a VHF/UHF ATC radio and an IFF capability [Jane’s 1999]. The ATC weight was assumed to be 5 pounds, based on the weight class of simple general aviation aircraft ATC equipment. The ATC weight is assumed to be avionics weight, though it is possible that it may be considered payload weight. The ATC equipment may not be permanent avionics. Recorder The data recording capability of the Predator is unknown. The avionics data recorder weight allocation was 1 pound, which was a bestjudgment assumption. Any additional data recorder weight is assumed to be covered under payload weights.
10.3.2.5 Subsystems
A summary of the Predator subsystems weights is presented in Figure 1018.
158
at minimum. 159 . The Raymer [1992] transport electrical system weights equation (Electrical System Method 6) was employed.81 kW. The performance model includes the impacts of the ram air on aerodynamic efficiency. All control surfaces are electrically actuated. Environmental Control System The Predator ECS system is assumed to be a ramair system. The custom flight control system method (FCS Method 7) was employed to estimate this weight. of a generator and power distribution wiring. The Predator has a satellite communications capability for command and control and imagery downlink. Fuel System Little is known from available literature of the Predator fuel system. The Roskam [1990] General Dynamics lowsubsonic aircraft ECS weight equation (ECS Method 2) was used. Communications Predator has a sophisticated communications system. Electrical System The electrical system consists. lbs 30 25 20 15 10 5 0 FCS ECS Electrical Fuel Sys Comms Figure 1018 Predator Subsystems Weights Summary Flight Control System The Predator flight control system consists of inboard and outboard wing trailing edge control surfaces and allflying ruddervators. It also has the ability to communicate within lineofsight range to a launch and recovery ground station. The Raymer [1992] transport fuel system weight equation (Fuel System Method 2) was used. The calculated nominal power load is 2.Predator Subsystems Weights 45 40 35 Weight.
Predator includes communications equipment. 10. and fuselage was estimated through the calibration process. The wing was assumed to weigh 75% of the equivalent aluminum wing with a minimum gage of 0. Predator Structural Weight 180 160 140 Weight.02 inches. which weighs 18. lbs 120 100 80 60 40 20 0 Wing Tail Fuselage Gear Installation Figure 1019 Predator Structural Weight Summary Wing The piecewise linear beam model was applied to the wing structure. The smaller Gnat UAV in the Predator family is stressed for 6 G maneuvers at an unspecified weight.Predator uses a 30inch parabolic dish for satellite communications. A summary of the Predator structural weights is presented in Figure 1019.2.3.3 pounds. since the wing and tails were probably designed by General Atomics with similar methods. tails.02 inches. The factor of 75% was matched to the wing factor. The 75% composites factor was attained through the calibration process. All or part of the equipment is produced by Loral. The tail was assumed to weight 75% of the equivalent aluminum tail with a minimum gage of 0. since the empty weight could not be sufficiently reduced through other parameters. Absent of further information. 160 .6 Structures The structure is largely made of carbon/epoxy composites [Jane’s 1999]. volumetric. the 6 G loading was applied to the Predator. The custom dish weight estimation method is used to calculate the weight. This equipment is assumed to include the Limited Capability Common Data Link. The 75% composites multiplication factor applied to the wing. Tails The tail weight was estimated with the piecewise linear beam model. and power impacts.
Jane’s lists the maximum takeoff weight at 2. The payload weight incorporated for the mission profile and the weights sizing is the sum of the known EO/IR and SAR payload weights. Again.5% of this value. Landing Gear The landing gear weight was estimated with the custom tricycle landing gear weight method. and these values were applied directly.872 pounds.250 pounds. The predator weight and mass fraction summaries are presented in Figure 1020 and Figure 1021. The target design gross weight is 1. and a maximum payload weight of 450 pounds.3. a fuel weight of 650 pounds. 161 . The sum of these weights comes to 1.Fuselage The fuselage weight was estimated with the Raymer General Aviation fuselage weight equation.6% of the target value. These two payloads were incorporated for the purposes of the sizing mission. A factor of 75% was applied to account for the extensive use of composites in the fuselage. The empty weight and fuel weight from Jane’s are assumed to be valid. respectively. The Versatron Skyball EO/IR payload with supporting electronics weighs 66 pounds.603 pounds.7 Overall Weights Jane’s [1999] lists the Predator Empty weight at 772 pounds. The advertised payload capacity is 450 pounds. which totals 181 pounds. The Lynx SAR payload weighs 115 pounds. Jane’s [1999] does not state what payload is carried for the mission. The calibrated empty weight is within 0. The calibrated Predator analysis design gross weight is within 0.2. The weights of the actual payloads appear to be much less than the maximum payload weight. 10. the 75% factor was identical to that of the wings and tails due to the assumed similarities in the design approach. Payloads Predator is known to carry both EO/IR and SAR payloads.
000 1. 162 G es ig n G ro ss th .3.8 Aerodynamics The predicted aerodynamics characteristics of the Predator are shown in Figure 1022 and Figure 1023.2. lbs ro w G Figure 1020 Predator Weights Summary Predator Mass Fractions 60% 50% Mass Fraction 40% 30% 20% 10% 0% Su bs ys te m s Pr op ul si on St ru ct ur es Av io ni cs pt y Figure 1021 Predator Mass Fraction Summary 10.000 100 10 1 Pa yl oa d Av io ni cs Su bs ys te m s Pr op ul si on St ru ct ur es Fu el pt y Em D Pa yl oa d Fu el ro w th Em Weight.Predator Weight Summary 10.
The descent from the final loiter point to sea level 163 .6 0.000 0 0.2. A climb from sea level to 20.6 0. Note that wave drag is not shown here.000 feet was imposed on the mission performance calculation. An additional onehour loiter at sea level is added to account for recovery operations.4 0.015 0. The EO/IRSAR payload combined weight of 181 pounds was used. A ceiling of 25.4 0.2 0. The Jane’s [1999] mission profile was used here.9 Performance The mission profile for Predator is 24 hours time on station at a 500 nautical mile radius.2 0.8 1 1.2 Lift Coefficient Figure 1022 Predator Trimmed Aerodynamic Performance The Predator drag contributors at the maximum bounding Reynolds number are shown in Figure 1023. The 2003 General Atomics Predator brochure indicates that the performance is 24 hours time on station at a 400 nautical mile radius.020 0. according to Jane’s [1999].010 0. not the maximum payload capacity.000 feet was included in the ingress segment. since this is not a consideration at these flight Mach numbers.8 Re/L = 167K Re/L = 227K Re/L = 391K Re/L = 615K 1 1.005 0.2 CDprof CDind CDfuse CDint Lift Coefficient Figure 1023 Predator Trimmed Drag Contributors 10.Predator Aerodynamic Performance 30 Lift to Drag Ratio 25 20 15 10 5 0 0 0.3.025 Drag Coefficient 0. Predator Drag Contributors 0.
000 25. kts Figure 1025 Predator Velocity Profile 164 . Figure 1026 does not represent the limits of the flight envelope.000 0 0 10 20 Time. only velocityaltitude points flown in the optimized mission.000 Altitude 15. Note that Figure 1026 is a combination of the data points shown in Figure 1024 and Figure 1025.000 5. and Figure 1026. hrs 30 40 Velocity. hrs 30 40 Figure 1024 Predator Altitude Profile Predator Velocity Profile 120 100 80 60 40 20 0 0 10 20 Time.000 10.000 20. The Predator altitude and velocity performance is shown in Figure 1024. Figure 1025.was included in the egress segment. Predator Altitude Profile 30.
ft 15000 10000 5000 0 40 50 60 70 80 90 100 110 Velocity. Design Item Volume Efficiency Induced Drag Interference Drag Wave Drag Laminar Flow Factor Of Safety Weight Growth Installation Weight Tech Level (01) 0.4 1.5 0. 1 (No compressibility impacts) 0. Table 106 Predator Design Technology Levels 165 .3.4 1. kts Figure 1026 Predator VelocityAltitude Envelope 10.10 Design Technology The weights and performance calibration process resulted design technology levels shown in Table 106.75 1.Predator VelocityAltitude Envelope Used 30000 25000 20000 Altitude.2. 0.
3. the wing configuration is driven by a constraint to reuse Pioneer program wing tooling.17 ft Source Jane’s [1999] Jane’s [1999] Jane’s [1999] Jane’s [1999] Table 107 Shadow 200 Geometry Characteristics 166 . Palumbo describes an evolutionary design history beginning in 1990 that has not ended.10.3.3 Shadow 200 10. and does not represent a clean sheet design synthesis. For example. 10. requirements changes.1 Shadow 200 Description Shadow 200 is a small tactical UAV designed to support lineofsight battlefield surveillance missions.3. Figure 1027 Shadow 200 Item Wing span Aspect ratio Sweep (quarter chord) Overall Length Value 12.3.75 ft 7.2 Geometry Characteristics The Shadow 200 geometry characteristics are shown graphically in Figure 1027. the technology year was assumed to be 2000 due to the extended development time. and incorporation of more advanced technologies. The design attempts to leverage legacy components and tooling. and numerically in Table 107. However.3. significant design evolution. Initial development began in 1990. Palumbo [2000] is assumed to be the most authoritative source of Shadow 200 data.07 0o 11.
75 kft Alt = 7.3. Figure 1028 and Figure 1029 show the Shadow 200 predicted installed propulsion characteristics.5 kft Alt = 11. Shadow 200 Installed Thrust Performance 90 80 70 60 50 40 30 20 10 0 0 50 Velocity. Item Maximum Power (S/L) BSFC (Max power.57 lbm/HPhr 28 lbs 2. S/L) Weight Propeller Diameter Value 38 HP 0.3 Propulsion The Shadow 200 uses a UEL AR 741 rotary engine.59 through the calibration process. so the only remaining performance factor for modification is the propulsion losses. kts Alt = 0 kft Alt = 3.33 ft Source Jane’s [1999] Jane’s [1999] Palumbo [2000] Jane’s [1999] Table 108 Shadow 200 Propulsion Characteristics The thrust and efficiency loss factor was found to be 0. The aerodynamics technologies are already very conservative.10. The engine drives a twoblade propeller. The engine weight listed includes the alternator.3. Major characteristics of the Shadow 200 engine are shown in Table 108. lbs 100 150 Figure 1028 Shadow 200 Installed Thrust Performance 167 .25 kft Alt = 15 kft Maximum Thrust.
This autopilot includes an INS and air data measurement equipment.4 Avionics The Shadow 200 uses a relatively simple avionics suite.3 0.Shadow 200 Installed Thrust Specific Fuel Consumption 0. which has an airborne component weighing 3 pounds. communications equipment. and elements of the electrical system [Palumbo 2000]. The avionics.5 kft Alt = 11. General Avionics Palumbo does not provide a detailed breakout of avionics weights. The Shadow 200 currently uses the Athena GS211 GuideStarTM autopilot.3. The Shadow 200 is known to use the Tactical Automatic Landing System (TALS). lbm/lbhr 0. includes the autopilot and a Mode IV IFF transponder. No further information about the avionics suite is provided by Palumbo. 10. There is no indication that any of the avionics components are redundant.3. which includes avionics. The avionics suite apparently has evolved since 2000 to include more modern equipment.3.75 kft Alt = 7.6 0. 168 .3. kts Figure 1029 Shadow 200 Installed Specific Fuel Consumption Performance 10. and the alternator weight was included in the propulsion weight.5 Subsystems An overview of the Shadow 200 subsystems weight results is in Figure 1030. The total weight allocated to avionics is 57 pounds.4 0. due to the year 2000 design year assumption.25 kft Alt = 15 kft 150 Velocity. For the purposes of this analysis. The allocated weight of 30 pounds is applied. The electrical system weight was set to 0 pounds.5 TSFC.1 0 0 50 100 Alt = 0 kft Alt = 3. which weighs 2 pounds.2 0. at minimum. 30 pounds was allocated to avionics and 27 pounds was allocated to communications equipment.
Fuel System The available literature provides little information on the Shadow 200 fuel system. The fuel system numbers appear to be reasonable. due to the weights accounting provided by Palumbo [2000]. Environmental Control System The available literature provides no information on the Shadow 200 Environmental Control System (ECS).Shadow 200 Subsystems Weights 30 25 20 15 10 5 0 FCS ECS Fuel Communications Figure 1030 Shadow 200 Subsystems Weights Summary Flight Control Systems The Shadow 200 Flight Control System (FCS) consists of one aileron per wing and Vtail ruddervators. The Raymer [1992] Transport fuel system weight equation (Fuel System Method 2) was used. despite being outside the applicable range. The Roskam [1990] . Other elements of the electrical system weights are captured in the avionics weight. The electrical system weight is fully accounted for by the engine and avionics weight categories. lbs 169 . A ramair ECS was assumed. The Shadow 200 wing tanks are believed to be selfsealing.General Weight. and the associated performance impacts are included in the flight performance model. Electrical System The electrical system weight was set to 0 to accommodate the weight accounting system differences between this model and the weights described by Palumbo [2000]. The Palumbo [2000] alternator weight is included in the engine weight. The custom FCS weight method (FCS Method 7) was used to estimate FCS weight.
8 G. and they were not included here. since the fuel fraction is relatively low on this UAV. based on the assumed allocation between avionics and communications weights described earlier.6 Structures The Shadow 200 structure is 90% composites.3. This weight method appears to scale well to small UAVs. as suggested by Raymer [1992]. This limit was applied directly at the design gross weight condition. The reference material is the 2017 Aluminum alloy with a minimum gage of 0. Separate communications relay payloads may be integrated. but they are not baseline avionics.8 G was applied here.Dynamics lowsubsonic aircraft ECS weight method (ECS Method 2) was used. No actual weights information is known for this system. Shadow 200 Structural Weights 40 35 30 Weight. The limit load of 3.02 inches. A composites wing weight multiplication factor of 0. but 27 pounds was assumed. which is primarily composed of graphite and Kevlar epoxy [Jane’s 1999].3.6 G. Tails 170 . Jane’s [1999] lists the limit load at 3. and Palumbo [2000] lists the limit load at 3.85 was used. which is a moderate bandwidth lineofsight communications link. 10. The Shadow 200 structural weight estimations are summarized in Figure 1031. lbs 25 20 15 10 5 0 Wing Tails Fuselage Gear Installation Figure 1031 Shadow 200 Structural Weight Summary Wing The piecewise linear beam model was used to determine the wing weight. Communications The Shadow 200 is equipped with the Tactical Common Data Link.
Although this equation is operating outside of its intended application. The Raymer [1992] General Aviation fuselage weight equation was used to estimate the fuselage weight. This weight was assumed to be split evenly between EO/IR and SAR functionality for the purposes of calibration and technology modeling. Fuselage The shadow 200 fuselage consists of a main pod and two booms that connect the tail to the wing. The standard landing gear weight equation is used to account for the fixed landing gear weight.3.83 was used.8 Overall Weights Palumbo [2000] provides group weights for structures.7 Payloads The maximum payload weight is 52 pounds [Palumbo 2000]. This houses the engine. The reference material is the 2017 Aluminum alloy with a minimum gage of 0. 171 . 10. The primary contributor to the weight is the main pod.3.The piecewise linear beam model was used to determine the tail weight. A composites tail weight multiplication factor of 0. A 0. most subsystems. Landing Gear The Shadow has both fixed landing gear and a parachute recovery capability. 27 pounds of avionics weights are moved to subsystems to account for the communications subsystems assumptions. The parachute recovery gear was estimated using the custom parachute recovery system weight method. These parameters as modified for this research. as suggested by Raymer [1992]. The Shadow 200 weight and mass fraction summaries are presented in Figure 1032 and Figure 1033. The gear drag contributions are accounted for in the aerodynamics analysis. the takeoff gross weight (design gross weight here).3. respectively.3. and the corresponding values from the analysis are provided in Table 109.02 inches. The structural weight in Palumbo appears to contain many elements of the subsystems weights. the resulting weight appears to be reasonable when combined with the overall installation weight. as recommended by Raymer [1992]. payload. The installation weight was primarily added to the fuselage weight. 10. avionics. and the avionics. landing gear.9 fuselage weight multiplication factor was applied to account for composites usage. and fuel. payload. propulsion.
1 lbs Difference 1.5 lbs (Includes wt.Structure (struct + subs) Avionics Engine/Alternator Payload/Ballast Fuel Design Gross Weight Palumbo Original/Modified 115/142 lbs 57/30 lbs 28 lbs 52 lbs 64 lbs 316 lbs Analysis 144. growth) 30 lbs 28 lbs 52 lbs 64.77% 0% 0% 0% 1. lbs 100 10 1 Av io ni cs Su bs ys te m s Pr op ul sio n St ru ct ur es Pa ylo ad G ro wt h Fu el Figure 1032 Shadow 200 Weights Summary Shadow 200 Mass Fraction Summary 70% 60% Mass Fraction 50% 40% 30% 20% 10% 0% Av io ni cs Su bs ys te m s Pr op ul sio n St ru ct ur es G ro wt h ad Fu el De sig n Em Pa ylo Figure 1033 Shadow 200 Mass Fraction Summary 172 Em pt y G ro ss pt y .99% Table 109 Shadow 200 Calibration weight comparison Shadow 200 Weight Summary 1000 Weight.63 lbs 319.0% 0.
Shadow 200 Drag Contributors 0.2 CDprof CDind CDfuse CDint CDgear Lift Coefficent Figure 1035 Shadow 200 Trimmed Drag Contributors 10.3. Jane’s [1999] indicates that the Shadow has an operational radius of 43 nautical miles and 6 hours of endurance.04 0.2 Lift Coefficient Re/L = 186K Re/L = 259K Re/L = 461K Re/L = 737K Figure 1034 Shadow 200 Trimmed Aerodynamic Performance Note that gear drag was included in the drag contributor summary. Shadow 200 Lift to Drag Ratio 16 Lift to Drag Ratio 14 12 10 8 6 4 2 0 0 0.4 0. though it is not explicitly states that this occurs on the same flight or that the maximum payload weight is included.05 Drag Coefficient 0.6 0.03 0.01 0.00 0 0.10 Performance The Shadow 200 flight performance is described differently by various sources. The Shadow 200 flies without retracting the landing gear.8 1 1.6 0.8 1 1.3.3.10.2 0.2 0.3.4 0.9 Aerodynamics Figure 1034 and Figure 1035 show the estimated aerodynamic performance of the Shadow 200.02 0. An AAI Shadow 200 brochure presented 173 .
at the Association of Unmanned Vehicle Systems International (AUVSI) conference on or before 2003 states that the performance is 56 hours without further details. The aircraft then performed a cruise climb until reaching the ceiling. Palumbo [2000] and Jane’s [1999] state the Shadow 200 ceiling is 15.000 6. Shadow 200 Altitude Profile 16. The second brochure also states (on the opposite side) that the endurance at 100 kilometers is 4 hours.000 feet.000 12.000 4.000 10.500 feet. The ingress segment climb was assumed to be from sea level to 12. such as transitioning from loiter to egress. Figure 1037. and does not represent the boundaries of the flight envelope. Figure 1036. hrs Figure 1036 Shadow 200 Altitude Profile 174 .000 8. Rapid changes in the velocityaltitude profile occur during changes of mission segment. an additional 30 minutes of endurance was added at the end of the mission to account for recovery operations and mission planning uncertainties. and Figure 1038 show the Shadow 200 altitude and velocity performance. The aircraft descended from the final loiter altitude to sea level in the egress segment. Palumbo [2000] describes the design mission profile as 4 hours time on station at a 200 kilometer radius. ft Time.000 14.000 0 0 2 4 6 8 Altitude.000 2. Note that Figure 1038 is assembled from the data points of the optimized mission profile. A second AAI brochure presented at AUVSI at or before 2003 states that the Shadow 200 can fly 50 ‘klicks’ (kilometers) and remain on station up to 4 hours. Here. which was applied here.
000 4.3.000 0 50 60 70 80 90 100 Altitude. ft Velocity.11 Design Technology The weights and performance calibration process resulted in the design technology levels shown in Table 1010.Shadow 200 Velocity Profile 120 100 Velocity.000 8.000 6.3.000 2.000 12.000 14. kts 80 60 40 20 0 0 2 4 6 8 Time.000 10. kts Figure 1038 Shadow 200 VelocityAltitude Envelope 10. 175 . hrs Figure 1037 Shadow 200 Velocity Profile Shadow 200 VelocityAltitude Envelope Used 16.
1.Design Item Volume Efficiency Induced Drag Interference Drag Wave Drag Laminar Flow Factor Of Safety Weight Growth Installation Weight Tech Level (01) 0.2 0.2 1. (No compressible effects) 0.35 Table 1010 Shadow 200 Design Technology Levels 176 .5 0.5 0.35 0.
The calibrated wings and tails are located and sized based on manual inputs. The reference year calibration and optimization cases are compared in this chapter. Emphasis is placed on the differences between the calibrated and optimized solutions. The avionics.1 Global Hawk 1994 Optimum Design Table 111 shows the Global Hawk active design variables and the associated limiting bounds.Chapter 11 Reference Year CalibrationOptimization Comparisons 11. The technology levels and design levels are the same. and payload inputs are identical. 177 . Details of the parametric geometry are described in Chapter 5. The results of the Global Hawk.1 Overview Optimized designs using the same requirements and same technology levels as the calibration cases were developed. There are more similarities than differences between the calibrated and optimized reference year cases. propulsion. so a specialized parametric geometry formulation is necessary. Predator. and Shadow 200 reference year optimized solutions are presented. and performance analysis. The calibration cases are inappropriate for optimization. 11. There are no differences in analysis methods. so it is not customized for any one case. mass properties. such as aerodynamics. The calibrated fuselage geometry will not reshape to account for wing junctures. subsystems. radome changes. The parametric geometry has similarities among all cases.2 Reference Case Comparisons 11. not parametric relationships. These values were used for all optimized Global Hawk cases without modification. The mission profile and input flight envelope constraints are identical. or other considerations. The comparison between the reference year optimized parametric and calibrated designs is necessary to better understand the progression of design into the future.2. The reference year calibration cases were described in the previous chapter.
The tail moment arm is less on the optimized case.2 0o Upper Bound 0.5% higher than the baseline. The comparison of the calibrated and optimized 1994 Global Hawk cases are shown graphically in Figure 111. The optimized wing has a shorter span. There is no EO/IR housing in the optimized case. Figure 112. Compressibility impacts do not appear to limit either case.17 1. The numeric differences in the geometry between these cases are presented in Figure 114.4 50 0. whereas the 1994 calibration case uses the lower fairing to house the SAR payload. The optimized fuselage has a slightly shorter fuselage length.Design Variable MFFuel bW ARW T/W LFuse t/cW lW ΛW Lower Bound 0. and much reduced sweep than the baseline wing. and Figure 113. so the tail area is greater.65 140 30 0.4 20 10 0. 178 . The overall wing area went down 8. The tail configurations for the two cases are essentially identical. and the optimum Global Hawk is on the right for these figures. The calibrated Global Hawk case is on the left. permitting an efficient higher speed loiter.09 0. 15o Table 111 Global Hawk Design Variables The optimized Global Hawk geometry has significant differences relative to the calibration case. The optimized case uses the parametric tail volume coefficient method. lower aspect ratio. The basic parametric fuselage prior to the addition of the radome and wing fairing has a lower lengthtodiameter ratio than the calibration fuselage.18 15 0. The optimized Global Hawk has a wing loading 3.6%. The parametric fuselage applied to all optimized Global Hawk cases uses the lower fairing to mount the wing.
Figure 111 1994 Global Hawk Geometry Comparison – Top View
Figure 112 1994 Global Hawk Geometry Comparison – Side View
179
Figure 113 1994 Global Hawk Geometry Comparison – Isometric View
Global Hawk Geometry Comparison
120 100 80 60 40 20 0 Fuse Length, ft Wing Span, ft Aspect Ratio Baseline 94 Opt
Figure 114 Global Hawk Geometry Comparison
The primary weights differences between the optimal and baseline cases reside within the structures and propulsion weights. The optimized Global Hawk has a reduced aspect ratio and shorter wingspan, resulting in reduced aerodynamic efficiency but also a reduced structural weight. The avionics and payload weights are identical. The minor subsystems weight differences are caused primarily driven by the flight control system. The differences in the weight and mass fraction summaries are shown in Figure 115 and Figure 116, respectively.
180
Global Hawk Weight Summary
100,000 10,000
Weight, lbs
1,000 100 10 1
Av io ni cs Su bs ys te m s Pr op ul sio n St ru ct ur es Pa ylo ad G ro wt h Fu el Em G ro ss pt y
Baseline 94 Opt
Figure 115 Global Hawk Summary Weights Comparison
Global Hawk Mass Fraction Summary
70% 60%
Mass Fraction
50% 40% 30% 20% 10% 0% Baseline 94 Opt
Su bs ys te m s Pr op ul sio n
Av io ni cs
St ru ct ur es
G ro wt h
ad
Fu el
De sig n
Em pt y
Figure 116 Global Hawk Summary Mass Fraction Comparison
There are no significant differences in the mission profiles. The calibration case achieves the maximum altitude of 65,000 feet shortly before the optimized case. The lift to drag ratio of the optimized case is lower than the calibration case, which is compensated for with the higher fuel fraction. The minor differences in specific fuel consumption are caused by the cruise Mach number and altitude experienced at a given point in the mission. The mission performance differences between the calibrated and optimized 1994 Global Hawk cases are presented in Figure 117, Figure 118, Figure 119, Figure 1110, and Figure 1111.
Pa ylo
181
Global Hawk Altitude Profile Comparison
70,000 60,000
Altitude, ft
50,000 40,000 30,000 20,000 10,000 0 0 5 10 15 20 25 30 35 Time, hrs Baseline 94 Opt
Figure 117 Global Hawk Altitude Profile Comparison
Global Hawk Mach Profile Comparison
0.7 0.6 0.5
Mach
0.4 0.3 0.2 0.1 0 0 5 10 15 20 25 30 35
Baseline 94 Opt
Time, hrs
Figure 118 Global Hawk Mach Profile Comparison
182
Global Hawk MachAltitude Envelope Used
70,000 60,000
Altitude, ft
50,000 40,000 30,000 20,000 10,000 0 0 0.2 0.4 Mach 0.6 0.8 Baseline 94 Opt
Figure 119 Global Hawk MachAltitude Envelope Used
Global Hawk Specific Fuel Consumption Comparison
0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0 5 10 15 20 25 30 35 Time, hrs
SFC, lbm/lbhr
Baseline 94 Opt
Figure 1110 Global Hawk Specific Fuel Consumption Comparison
183
Global Hawk Lift to Drag Ratio Comparison 30 Lift to Drag Ratio 25 20 15 10 5 0 0 5 10 15 20 25 30 35 Time. hrs Baseline 94 Opt Figure 1111 Global Hawk Lift to Drag Comparison 184 .
11.09 0. The tail proportions of the optimized case appear exaggerated because the tail moment arm is short. this is a valid solution.04 5 0. The parametric Predator tail geometry is different for the two cases. The optimization process resulted in a significantly lower lengthtodiameter ratio and shorter fuselage length for the optimized Predator than for the calibration case. Although less visually pleasing than the calibration case. The calibrated Predator is on the left and the optimum Predator is on the right for these figures.2 Upper Bound 0. which drives the vertical component of the tail. whereas the calibrated Predator does not. The geometry differences between these two cases are presented graphically in Figure 1112.2 Predator 1994 Optimum Design Table 112 shows the Predator active design variables and the associated limiting bounds. Key geometry differences are presented numerically in Figure 1115. Although still a Vtail configuration. 185 . The parametric Predator fuselage contains a wing fairing on the underside. Design Variable MFFuel bW ARW P/W LFuse t/cW lW Lower Bound 0. These values were used for all optimized Predator cases without adjustment.17 0. The dihedral angle is higher on the optimized case because the wing aspect ratio is higher. The optimized solution has a slightly shorter span and nearly identical aspect ratio to the calibration case.2. The traditional Vtail was selected to minimize erroneous interference impacts in the Trefftz plane analysis that might be exploited by the optimizer and artificially drive the optimized solutions. The Predator parametric fuselage has distinct differences from the calibration case.1 30 0. which scales the radome dimensions linearly with the parabolic dish diameter. Figure 1113.25 6 10 0. despite the Vtail plane intersecting the wing plane. and the calibrated Predator has an inverted Vtail. and Figure 1114. The parametric Predator and parametric Global Hawk have the same radome sizing relationship. The inverted Vtail was found to be free of interference problems on the baseline case.9 Table 112 Predator Design Variables The Predator 1994 optimal solution exhibits geometry variations relative to the calibration case.45 55 30 0. the parametric Predator has a traditional Vtail.
Figure 1112 1994 Predator Geometry Comparison – Top View Figure 1113 1994 Predator Geometry Comparison – Side View 186 .
ft Wing Span. If a takeoff analysis were performed.Figure 1114 1994 Predator Geometry Comparison – Isometric View Predator Geometry Comparison 50 40 30 Baseline 20 10 0 Fuse Length.5% less than the calibrated case. The primary differences reside in structures and propulsion weights. The avionics and payload weights are identical. ft Aspect Ratio 94 Opt Figure 1115 Predator Geometry Comparison A comparison of the calibrated and optimized 1994 Predator weight and mass fraction summary is presented in Figure 1116 and Figure 1117. The uninstalled power to weight ratio for the optimized 1994 Predator is 15. then a higher power to weight ratio might result. which has significant synergistic effects on the design. The optimized Predator has a lower design gross weight than the calibration case. This directly drives the propulsion weight down. The differences in the subsystems are driven primarily by reduced flight control system and fuel system weights. respectively. 187 . This appears to be a driver for the weight difference between the two cases.
The mission performance characteristics are compared in Figure 1118. the structures weight reduction also appears to be a function of reduced wing. and airspeed.In addition to the reduced propulsion weight influence. which is scaled as a function of engine power. Predator Weight Summary Summary 10.9% higher than the baseline case. The propeller efficiency is a function of the power required.000 1. lbs 100 Baseline 94 Opt 10 1 Su bs ys te m s Pr op ul sio n Av io ni cs St ru ct ur es G ro wt h ad Fu el Pa ylo Em G ro ss pt y Figure 1116 Predator Weights Summary Comparison Predator Mass Fraction Summary 50% 45% 40% 35% 30% 25% 20% 15% 10% 5% 0% Baseline 94 Opt Mass Fraction Av io ni cs St ru ct ur es Su bs ys te m s Pr op ul si Figure 1117 Predator Summary Mass Fraction Comparison The internally optimized mission profiles for the two reference year Predator cases are similar. and Figure 1122. Figure 1121. the altitude. fuselage. 188 Pa ylo G ro wt h on ad Fu el De sig n . The optimized Predator has a wing loading 10. and tail size. Figure 1120. The difference in specific fuel consumption appears to be driven by the propeller. Figure 1119.000 Weight. resulting in higher speeds in some portions of the mission.
000 5. hrs 30 40 Figure 1119 Predator Velocity Profile Comparison 189 . kts 0 10 20 Time.Predator Altitude Profile Comparison 30.000 25. ft 15.000 20. hrs Figure 1118 Predator Altitude Profile Comparison Predator Velocity Profile Comparison 120 100 80 60 40 20 0 Baseline 94 Opt Velocity.000 0 0 10 20 30 40 Baseline 94 Opt Time.000 10.000 Altitude.
15 0.2 0.3 0. hrs 25 30 35 40 Baseline 94 Opt Figure 1121 Predator Specific Fuel Consumption Comparison 190 .000 10.05 0 0 5 10 15 20 Time.000 20.000 25.1 0.000 0 40 50 60 70 80 90 100 110 Velocity. kts Baseline 94 Opt Figure 1120 Predator VelocityAltitude Envelope Used Predator Specific Fuel Consumption Comparison 0. ft 15. lbm/lbhr 0.35 0.000 5.Predator VelocityAltitude Profile Used 30.25 SFC.000 Altitude.
hrs 25 30 35 40 Figure 1122 Predator Lift to Drag Comparison 191 .Predator Lift to Drag Ratio Comparison 25 20 Lift to Drag Ratio 15 Baseline 94 Opt 10 5 0 0 5 10 15 20 Time.
The resulting aspect ratio was much higher than the calibration case.08 3 5 0. which largely drove the wing design. due to an increased wing loading.09 0. A maximum wingspan constraint of 12. The numerical geometry comparison is shown in Figure 1125.2 12 0.2.75 feet. The calibrated Shadow 200 is on the left and the optimum Shadow 200 is on the right in these figures.25 12. and Figure 1125. The optimal case does not have this restriction. 192 . so the aspect ratio and wingspan are free to vary independently.05 3 0. Figure 1124. The optimization process resulted in a higher aspect ratio wing than the calibration case. The optimal Shadow 200 flies faster than the calibrated Shadow 200. The parametric tails are a traditional Vtail configuration attached to the fuselage. Additionally.2 Upper Bound 0. The optimized case does not reach the wingspan constraint.17 1 Table 113 Shadow 200 Design Variables The Shadow 200 reference year optimal solution varies significantly from the calibration case. These values were used for all optimized Shadow 200 cases. was imposed to account for transportation considerations.75 20 0. Future versions of the software could easily integrate a parametric twinboom configuration. The calibration case has an inverted Vtail attached at the tips to the booms. The calibration Shadow 200 case fuselage has a pod and twinboom configuration. Design Variable MFFuel bW ARW P/W LFuse t/cW lW Lower Bound 0. whereas the calibration case has a midwing attachment to the pod fuselage. The Shadow 200 was constrained by AAI to maintain the chord of the available tooling.3 Shadow 200 2000 Optimum Design Table 112 shows the Shadow 200 active design variables and the associated limiting bounds. This configuration is more difficult to integrate parametrically than the parametric fuselage formulation used for the other cases. The geometry comparison of these cases is presented in Figure 1123. The fuselage geometry is noticeably different. the parametric wing is attached to the lower fuselage with a wing fairing.11. the span for the calibration case. The tail geometry is also different between the two cases.
Figure 1123 2000 Shadow 200 Geometry Comparison – Top View Figure 1124 2000 Shadow 200 Geometry Comparison – Side View Figure 1125 2000 Shadow 200 Geometry Comparison – Isometric View 193 .
Shadow 200 Geometry Comparison 14 12 10 8 6 4 2 0 Fuse Length. showing an increase in relative utility. The avionics and payload weights between the two cases are identical. lbs 100 Baseline 00 Opt 10 1 Pa yl oa d Pr op ul si on St ru ct ur es Fu el pt y Em D es ig n Su bs ys te m s Av io ni cs ro w Figure 1127 Shadow 200 Summary Weights Comparison 194 G G ro ss th . The payload mass fraction is higher for the optimized case. ft Aspect Ratio Baseline 00 Opt Figure 1126 Shadow 200 Geometry Comparison The Shadow 200 2000 weight and mass fraction summary comparisons are presented in Figure 1127 and Figure 1128. which is a result of offsetting effects of the improved lift to drag ratio and degraded specific fuel consumption of the optimized case. Shadow 200 Weight Summary 1000 Weight. respectively. The fuel mass fractions are similar. The higher payload mass fraction is made possible by a reduction in structural mass fraction. The degraded specific fuel consumption is caused by propeller efficiency reduction at higher speeds. ft Wing Span. The differences in subsystems weights are largely due to flight controls and fuel system weight variations.
000 14.000 6. Figure 1131.000 0 0 1 2 3 4 5 6 7 Time.000 4. and Figure 1133.000 2.000 12. Figure 1132. Shadow 200 Altitude Profile Comparison 16. Figure 1130.000 Altitude.Shadow 200 Mass Fraction Summary 70% 60% Mass Fraction 50% 40% 30% 20% 10% 0% Baseline 00 Opt Su bs ys te m s Pr op ul sio n Av io ni cs St ru ct ur es ro wt h G ad Fu el Pa ylo Figure 1128 Shadow 200 Summary Mass Fraction Comparison Comparison of the sameyear Shadow 200 flight performance characteristics are shown in Figure 1129.000 8. ft 10. hrs Baseline 00 Opt Figure 1129 Shadow 200 Altitude Profile Comparison 195 Em pt y .
Shadow 200 Velocity Profile Comparison 120 100 80 60 40 20 0 0 1 2 3 4 5 6 7 Time. kts Shadow 200 VelocityAltitude Envelope Used 16.000 0 40 50 60 70 80 90 100 110 120 Velocity.000 Altitude.000 12.000 4. kts Baseline 00 Opt Figure 1131 Shadow 200 VelocityAltitude Envelope Used 196 .000 6. ft 10.000 14.000 2. hrs Baseline 00 Opt Figure 1130 Shadow 200 Velocity Profile Comparison Velocity.000 8.
lbm/lbhr 0.00 0 1 2 3 4 5 6 7 Time.50 0.10 0.30 0.25 0. hrs Baseline 00 Opt Figure 1132 Shadow 200 Specific Fuel Consumption Comparison Shadow 200 Lift to Drag Ratio Comparison 14 12 Lift to Drag Ratio 10 8 6 4 2 0 0 1 2 3 4 5 6 7 Time.40 SFC.Shadow 20 Specific Fuel Consumption Comparison 0.15 0.35 0. hrs Baseline 00 Opt Figure 1133 Shadow 200 Lift to Drag Ratio Comparison 197 .20 0.45 0.05 0.
Global Hawk is the only design class that contends with compressibility effects. For a fixed technology year. Beyond 2010. followed by avionics/ subsystems. The Global Hawk. low aspect 198 . The fuselage length trend is shown in Figure 129. Figure 126. or both were typically selected by the optimizer to delay the onset of compressibility effects. and 2015 cases took the approach using low aspect ratio wings with little sweep and reduced thickness to chord ratios.1 Sizing Overview The Global Hawk optimized cases experienced steady weight reduction through the advancing years.2 Geometry The graphical summary of the Global Hawk technology cases is presented in Figure 121. Two design families. The earliesttolatest date progresses from left to right in these figures. the relatively unvarying EO/IR payload weight limited the sizing improvement potential. The relative improvement per period steadily diminished through 2010. and Shadow 200 technology solutions are presented in this chapter. as shown in Figure 128. Wing geometry trends are shown in Figure 124. and Figure 128. the unswept. Figure 125. The initial advances were mostly driven by the SAR payload weight reduction and propulsion improvements. 2005. taking a significantly different approach than the two preceding cases. Predator. 12.Chapter 12 Technology Case Optimization Applications 12. Figure 122. The Global Hawk cases can ingress and egress at a low transonic speed. emerged from the optimization.2. 12. and 2025 cases had high aspect ratio wings with some sweep and high thickness to chord ratio wings. The 1994. or niches. so wing sweep. The improvement degradation began to stabilize near 2015 through 2020. structures and aerodynamics. Figure 127.1 Overview The technology impacts investigation evaluated the design synthesis of air vehicles with constant requirements and advancing technologies.2. 2010. reduced wing thickness to chord ratio. The 2025 optimum design was an unusually good design. and Figure 123. The 2000. despite advances in other technologies. Optimization aided the design synthesis substantially.2 Global Hawk Technology Impacts 12.
The fuselage length decreased from 1994 through 2010. 2005.717%. highaspect ratio niche has a higher aerodynamic efficiency and a lower fuel mass fraction that the competing niche. The optimizer tended to select wing thickness to chord ratios of 15. 2010. the swept.ratio niche has a relatively low structural mass fraction due to the smaller wing size and more efficient structure. 2005. 2025 199 . The wing span and aspect ratio tended to be more significant drivers than taper ratio and wing sweep. but this comes at the expense of structural weight. 2000. The radome geometry influence on the fuselage outer loft became insignificant by 2020. but then remained relatively constant from 2015 through 2020. 2000. which was well above the minimum value of 9%. For a fixed technology year. Years (Left to Right): 1994. 2020. Figure 121 Global Hawk Geometry Trends – Top View. The wing span for sameniche vehicles consistently decreased over time. 2015. 2020. Years (Left to Right): 1994. but a higher fuel fraction due to the relatively poor aerodynamic performance. 2010. 2025 Figure 122 Global Hawk Geometry Trends – Side View. 2015.
Years (Left to Right): 1994. 2005.Figure 123 Global Hawk Geometry Trends – Isometric View. 2010. 2000. 2020. 2025 Global Hawk Wingspan 120 100 80 60 40 20 0 94 Opt 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt Wingspan. ft Figure 124 Global Hawk Wing Span Trends 200 . 2015.
Global Hawk Aspect Ratio 30 Aspect Ratio 25 20 15 10 5 0 94 Opt 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt Figure 125 Global Hawk Aspect Ratio Trends Global Hawk Wing Sweep Wing Sweep.5% 17.5% 15.0% 15.5% 16.0% 16. Degrees 10 8 6 4 2 0 94 Opt 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt Figure 126 Global Hawk Wing Sweep Trends Global Hawk Thickness to Chord Ratio Thickness to Chord Ratio 17.0% 94 Opt 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt Figure 127 Global Hawk Thickness to Chord Ratio Trends 201 .
Global Hawk Design Variable Values Relative to Limits 100% Relative Value 80% 60% 40% 20% 0% 1990 1995 2000 2005 ARw t/c Sweep 2010 2015 2020 2025 Design Year Figure 128 Global Hawk Wing Parameter Comparisons Global Hawk Fuselage Length 50 Fuselage Length. There was no consistent trend for uninstalled thrust to weight ratio referenced to the design gross weight as a function of design year. The decrease in thrust to weight ratio of the 2025 design relative to its predecessor was a result of more aggressive aerodynamic design. ft 40 30 20 10 0 94 Opt 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt Figure 129 Global Hawk Fuselage Length Trends 12. so the thrust to weight ratio increased from 2010 to 2020. respectively. such as higher aspect ratio and increased sweep. 202 . but then was reduced more gradually and inconsistently from 2000 to 2025. The Global Hawk uninstalled thrust level rapidly shrank from 1994 to 2000.3 Propulsion The global hawk uninstalled thrust and uninstalled thrusttoweight trends are shown in Figure 1210 and Figure 1211.2. The vehicle thrust to weight ratio became less costly to the design as time progressed due to engine thrust to weight improvements.
000 2.4 Weight Summary Weights The Global Hawk weight and mass fraction trends are shown in Figure 1212 and Figure 1213. and avionics were reduced. propulsion.05 0 94 Opt 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt Figure 1211 Global Hawk Thrust to Weight Ratio Trends 12.000 4. lbs 6.000 0 94 Opt 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt Figure 1210 Global Hawk Uninstalled Thrust Trends Global Hawk Thrust to Weight Ratio 0.15 0.25 0.Global Hawk Uninstalled Thrust Uninstalled Thrust. As expected.3 0.2 0.000 3. Note that the weights in Figure 1212 are on a log scale.1 0. The weight improvement trends are shown in Figure 1214. The mass fraction of the payload increased as the mass fractions consumed by fuel. respectively.35 Thrust to Weight Ratio 0. 203 . so weight categories such as the growth weight appear exaggerated despite having a relatively constant mass fraction.2.000 1. the synthesized vehicle weight improved as technology improves.000 5.
Obligated Weights Trends 204 . % 50% 40% 30% 20% 10% 0% 2000 2005 2010 2015 2020 2025 W_DG W_Obligated Design Year Figure 1214 Design Gross . lbs 94 Opt 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt 1000 100 10 1 Su bs ys te m s Pr op ul sio n Av io ni cs St ru ct ur es G ro wt h ad Fu el Pa ylo Em pt y Figure 1212 Global Hawk Weight Trends Global Hawk Mass Fraction Summary 70% 60% Mass Fraction 94 Opt 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt 50% 40% 30% 20% 10% 0% Su bs ys te m s Pr op ul sio n Av io ni cs St ru ct ur es G ro wt h ad Fu el Pa ylo Em pt y Figure 1213 Global Hawk Mass Fraction Trends Global Hawk Improvement Over Previous Period 70% 60% Improvement.Global Hawk Weight Summary 100000 10000 Weight.
lbs 1 Figure 1215 Global Hawk Avionics Weight Trends 205 .0001 0.001 0.01 0. The relative contribution of the communications system was greater in the future cases when using the assumed technology curves.1 0.00001 94 Opt 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt GPS ADS ATC Recorder Subsystems Weights The subsystems weight trends are shown in Figure 1216. The dish diameter and associated dish weight followed a technology trend separate from the subsystems curve. The relative influence of the overall avionics weight contribution to vehicle sizing compared to the payload weight became less significant as time advanced.Avionics Weights The avionics weights improved dramatically through future years. The results can be seen in Figure 1215. Weight. The avionics weight reductions follow the technology trends provided in Chapter 4. This was caused by the parabolic dish weight that was a function of dish diameter. Global Hawk Avionics Weight 1000 100 10 General Avionics Autopilot INS 0.
Relative to the 1994 EO/IR value. The EO/IR payload became the primary vehicle sizing driver in 2025. The installation weight became the most dominant structural weight contributor in 2005.6% in 2025.4% in 2005 and 14.Global Hawk Subsystems Weight 1000 100 FCS ECS Electrical System Fuel System Communications Weight. The largeEO/IR payload technology prediction was applied here. The SAR payload weight saw aggressive improvement in weight relative to the EO/IR payload. The physical 206 . lbs 10 1 94 Opt 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt Figure 1216 Global Hawk Subsystems Weight Trends Payload Weights The Global Hawk payload weight trends are shown in Figure 1217. lbs EO/IR SAR 10 1 94 Opt 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt Figure 1217 Global Hawk Payload Weight Trends Structures Weights The Global Hawk structural weight trends are shown in Figure 1218.9% by 2025. Relative to the 1994 SAR value. Global Hawk Payload Weight 1000 100 Weight. the EO/IR payload weight was reduced by only 7. the SAR payload weight was reduced by 87.2 % in 2005 and 98.
and Figure 1222. The internal mission optimization routine selected slightly different flight profiles for the various solutions.5 Performance The optimized Global Hawk mission profiles and related performance data are shown in Figure 1219.2. 207 . Figure 1221. lbs Tail 100 Fuselage Nacelle Gear Installation 10 1 94 Opt 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt Figure 1218 Global Hawk Structures Weight Trends 12. The installation weight associated with these items was proportional to the component weights. Figure 1220. fuel reduction.size of the structure decreased due to volume reductions of many components. The combined relative weight of the obligated weights did not change as rapidly as the major structural components. Global Hawk Structures Weight 10000 1000 Wing Weight. The 1994 optimum had the highest average Mach number and reached the maximum altitude quickest. and therefore the installation weight did not decrease rapidly. and general sizing phenomena. Future research should fully account for these relevant phenomena. The wing structural weight trends would likely vary if an aerolastic and stiffness analyses were performed.
hrs G 00 G 05 G 10 G 15 G 20 G 25 Figure 1221 Global Hawk Mission Lift to Drag Ratio Trends 208 .Global Hawk Altitude Profile 70.000 30.000 0 0 5 10 15 20 25 30 35 Time.000 10.2 0.000 60. hrs G 15 G 20 G 25 Figure 1220 Global Hawk Mach Profile Trends Global Hawk Lift to Drag Ratio 30 25 Lift to Drag Ratio G 94 20 15 10 5 0 0 5 10 15 20 25 30 35 Time.3 0.4 0.6 0.000 G 94 G 00 G 05 G 10 Altitude. hrs G 15 G 20 G 25 Figure 1219 Global Hawk Altitude Profile Trends Global Hawk Mach Profile 0.000 50.1 0 0 5 10 15 20 25 30 35 Time.7 0.000 20.5 Mach G 94 G 00 G 05 G 10 0. ft 40.
5 0. The total bar is the design gross weight difference between the 1994 optimized Global Hawk and the 2005 optimized Global Hawk with all technologies applied simultaneously in the design. is 6. and the total difference in design gross weight from 1994 to 2005 is 74. hrs G 05 G 10 G 15 G 20 G 25 Figure 1222 Global Hawk Mission Fuel Consumption Trends 12.8 0.5%.6 Technology Sensitivities The Global Hawk technology sensitivities are shown in Figure 1223. The components bar shows the weight improvement for each individual technology group when applied individually. The overall synergy. the combined bar is the multiplication of all of the improvements from each of the technology groups.9%. 209 . The three bars represent different ways of viewing the technology impacts. The combined improvement of all individual technology groups is 67.2.7 0. defined here as the difference of the total and combined technologies. There was overall synergy among the combined technologies.Global Hawk Specific Fuel Consumption 0. Finally.68%.4 0. The avionics/subsystems technologies also had significant influence.1 0 0 5 10 15 20 25 30 35 Time.3 0. The combined bar is found from: Improvement = 1 − Components ∏ (1 − Improvement Component ) Global Hawk was most sensitive to payload and then propulsion technology impacts between 1994 and 2005.2 0.6 SFC lbm/lbhr G 94 G 00 0.
% Combined 80% 60% 40% 20% 0% Total Components Combined Total Structures Propulsion Payload Design Avion/Subs Aerodynamics Figure 1223 Global Hawk Design Gross Weight Sensitivity to Technology 210 .Global Hawk Technology Sensitivities 19942005 100% Weight Reduction.
Figure 1228. subsystems. and fuel volume. permitting the fuselage to assume more visually pleasing and aerodynamically efficient contours. and were not major design drivers.3. the dish diameter and radome size decreased.1 Sizing Overview Like Global Hawk. and Figure 1226. 12. Figure 1229. which define the wing area. given the dominance of the radome. Figure 1225.12.81 for the 2015 optima. The fuselage length trend is provided in Figure 1231. The fuselage length generally decreased with time due to decreasing avionics. propulsion. which was far more aggressive than the Global Hawk large EO/IR trend. The optimizer generally selected more aggressive wing taper for this family of UAV.2 Geometry Graphical representation of the Predator technology case geometry behavior is shown in Figure 1224. The most dominant visual feature of the optimized predator solutions is the radome. The strongest wing geometry contributors to design gross weight were the wing span and aspect ratio.3 Predator Technology Impacts 12. and Figure 1230. The parametric wing fairing length is proportional to the wing root chord. The notable reduction in EO/IR weight throughout the dates considered permitted continuing weight reduction beyond 2005. Predator used the small EO/IR technology trend. Apparently. One possible explanation for this result is the impact of the wing on the fuselage geometry. Wing taper provides wing structural weight and induced drag reductions. Numerical wing geometry trends are presented in Figure 1227. the negative impacts of low taper ratios on the fuselage drag overcame the wing structural and aerodynamic benefits for the 2015 case. The Predator parametric radome strongly affected the aerodynamic performance of the vehicles under consideration. 211 . The wing root chord may have been reduced to manage the fuselage length to diameter ratio for fuselage drag reduction. The taper and thickness to chord ratios had no correlation with design year. Predator sizing was strongly influenced by the payloads and other obligated weights. The optimization process resulted in a relatively low wing taper ratio of 0. The wing area tended to decrease with decreasing vehicle weight. As the assumed satellite communications technology improved.3.
2025 Figure 1225 Predator Geometry Trends – Side View. Years (Left to Right): 1994. 2015. 2020. 2005. 2010. 2015. 2000. 2020. 2000. Years (Left to Right): 1994.Figure 1224 Predator Geometry Trends – Top View. 2010. 2005. 2025 212 .
ft 40 30 20 10 0 94 Opt 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt Figure 1227 Predator Wing Span Trends 213 . Years (Left to Right): 1994. 2020. 2015. 2000. 2005. 2025 Predator Wing Span 50 Wing Span.Figure 1226 Predator Geometry Trends – Isometric View. 2010.
100 80 60 40 20 0 94 Opt 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt Figure 1230 Predator Wing Area 214 .Predator Aspect Ratio 30 25 Aspect Ratio 20 15 10 5 0 94 Opt 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt Figure 1228 Predator Aspect Ratio Trends Predator Thickness to Chord Ratio Thickness to Chord Ratio 0.15 0.05 0 94 Opt 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt Figure 1229 Predator Thickness to Chord Ratio Predator Wing Area 120 Wing Area.ft.1 0.2 0. sq.
respectively. The powertoweight ratio tended to increase as time progressed.3 Propulsion The Predator uninstalled power and uninstalled powertoweight trends are shown in Figure 1232 and Figure 1233. ft 15 10 5 0 94 Opt 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt Figure 1231 Predator Fuselage Length Trends 12. since the weight impacts attributable to the propulsion system were less stressing on future designs. the uninstalled power decreased with vehicle weight as time progresses. Predator Uninstalled Power Uninstalled Power.3. HP 100 80 60 40 20 0 94 Opt 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt Figure 1232 Predator Uninstalled Power Trends 215 .Predator Fuselage Length 20 Fuselage Length. As expected.
which revealed that the other technology improvements were providing benefit.06 0.02 0. Predator Weight Summary 10000 1000 94 Opt 00 Opt 05 Opt 10 Opt 10 1 0.07 0.01 0.4 Weight Summary Weights The Predator technology case weight and mass fraction trends are shown in Figure 1234 and Figure 1235. The Predator weights uniformly decreased with improving technology.3.1 15 Opt 20 Opt 25 Opt Weight.Predator Power to Weight Ratio Power to Weight Ratio. lbs 100 Av io ni cs Su bs ys te m s Pr op ul sio n St ru ct ur es Pa ylo ad G ro wt h Fu el Figure 1234 Predator Weight Summary Trends 216 De sig n Em G ro ss pt y .05 0. but the general trend was for the fuel mass fraction to decrease while the payload mass fraction increased.08 0.04 0. HP/lb 0.03 0. respectively.00 94 Opt 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt Figure 1233 Predator Power to Weight Ratio 12. The behavior of the mass fractions had variation. The Predator overall weight improvement trends are shown in Figure 1236. The overall design gross weight improved faster than the reduction in obligated weights.
% 50% 40% 30% 20% 10% 0% 2000 W_DG W_Obligated 2005 2010 2015 Pa ylo 2020 2025 Design Year Figure 1236 Predator Design Gross – Obligated Weight Trends Avionics Weights The Predator technology case avionics weights trends are presented in Figure 1237. the avionics weights became negligible as technology advanced. As seen on Global Hawk. 217 G ro wt h on ad Fu el .Predator Mass Fraction Summary 60% 50% 94 Opt 00 Opt 05 Opt 30% 20% 10% 0% 10 Opt 15 Opt 20 Opt 25 Opt Mass Fractions 40% Av io ni cs St ru ct ur es Su bs ys te m s Pr op ul si Figure 1235 Predator Summary Mass Fraction Trends Predator Improvement Over Previous Period 70% 60% Improvement.
The relative weight reduction of the SAR payload from 1994 was identical to that on Global Hawk.1 0.001 94 Opt 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt Figure 1237 Predator Avionics Weight Trends Subsystems Weights The Predator technology case subsystems weight trends are shown in Figure 1238. because this weight was a function of the satellite communications dish diameter that followed a technology trend distinct from the other subsystems. The subsystems weights all decreased with time. lbs 1 INS GPS ADS ATC Recorder 0.1 0. lbs ECS 1 Electrical Fuel Sys Comms 0.Predator Avionics Weights 100 10 General Autopilot Weight. The EO/IR saw 218 .01 0.01 94 Opt 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt Figure 1238 Predator Subsystems Weight Trends Payload Weights The Predator payload weight trends are presented in Figure 1239. The slowest to decrease was the communications system weight. Predator Subsystems Weight 100 10 FCS Weight. The payload weights exhibited similar behavior to those seen on Global Hawk.
The structural weights all decreased with time. This weight accounts for the structural mounting of avionics. The slowly decreasing payload weight was the primary contributor to the relatively intractable installation weight. lbs EO/IR SAR 10 1 94 Opt 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt Figure 1239 Predator Payload Weight Trends Structural Weights The Predator technology case structural weight trends are shown in Figure 1240. because the more aggressive small EO/IR technology trend was applied. Predator Payload Weight 1000 100 Weight.more weights improvement than on Global Hawk. Predator Structural Weight 1000 100 Weight. The installation weight became increasingly dominant. and propulsion. subsystems. payload. lbs Wing Tail Fuselage Gear Installation 10 1 94 Opt 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt Figure 1240 Predator Structures Weight Trends 219 .
000 94 Opt Altitude. The Predator flight performance showed some variation.000 10. and Figure 1244. Figure 1242. the lift to drag ratio degraded with time. The culprit was that the size of the EO/IR ball relative to the wing made this very high drag sphere a larger contributor in the future cases. The 1994 Global Hawk also had a faster ingress than its descendants.5 Performance The Predator technology case mission profiles and associated performance information are presented in Figure 1241. Aerodynamic performance improvements can be attained by increasing the aspect ratio and wing span. ft 20. hrs 30 40 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt Figure 1241 Predator Altitude Profile Trends 220 . Figure 1243. The 1994 optimum reached the maximum altitude during loiter much more rapidly than later aircraft in the family. Improved specific fuel consumption in later years made aerodynamic performance less important to achieve the same performance results. The higher flight velocity brought the 1994 Predator design to the loiter point rapidly. Predator Altitude Profile 30.000 25. but both of these options add structural weight that offsets the benefits of reduced fuel consumption.000 5. the high lift to drag ratio permitted this aircraft to climb faster than its less aerodynamically efficient descendents.3. During loiter.12. In general.000 15.000 0 0 10 20 Time.
hrs 30 40 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt Figure 1242 Predator Velocity Profile Trends Predator Lift to Drag Ratio 20 18 16 94 Opt Lift to Drag Ratio 14 12 10 8 6 4 2 0 0 10 20 Time.Predator Velocity Profile 120 100 94 Opt Velocity. kts 80 60 40 20 0 0 10 20 Time. hrs 30 40 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt Figure 1243 Predator Mission Lift to Drag Ratio Trends 221 .
2 0. hrs 30 40 94 Opt 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt Figure 1244 Predator Mission Fuel Consumption Trends 12. which is the highest of the three vehicle classes evaluated here.3. % 100% 80% 60% 40% 20% 0% Total Components Combined Combined Total Structures Propulsion Payload Design Avion/Subs Aerodynamics Figure 1245 Predator Design Gross Weight Sensitivity to Technology 222 .6 Technology Sensitivities The Predator design gross weight sensitivities to various technologies are shown in Figure 1245. Predator Technology Sensitivities 19942005 120% Weight Reduction.35 0.15 0.05 0 0 10 20 Time. The Predator showed 7.1% overall synergy.25 0. followed by avionics/subsystems and propulsion technology categories.Predator Specific Fuel Consumption 0.1 0.3 SFC. Predator is most sensitive to the payload technology category. lbm/lbhr 0.
Two families of solutions. The fuselage length is shown presented in Figure 1251. 2020. 2010. Years (Left to Right): 2000. The 2015 and 2025 cases fall into the latter family. and Figure 1248. No clear aspect ratio and taper ratio relationships are evident. The wing geometry behavior is provided in Figure 1249 and Figure 1250.4. 2020. 2025 223 . Figure 1247.4 Shadow 200 Technology Impacts 12.4.1 Sizing Overview The Shadow 200 family has shown the least improvement over time. Figure 1246 Shadow 200 Geometry Trends – Top View. The second family had a combination of high aspect ratio and low wing loading to improve aerodynamic performance and propulsion efficiency. 2010. The first family sought to minimize physical size to reduce structural weight.12. 2015. emerged through optimization. 2015.2 Geometry The Shadow 200 technology case trends are shown graphically in Figure 1246. 2025 Figure 1247 Shadow 200 Geometry Trends – Side View. 2005. which improved the fuel fraction. 2005. This was caused by the relatively low fuel mass fraction lowering the weight escalation factor. 12. Figure 1252 shows the wing loading and aspect ratio trends for all technology cases. or niches. Years (Left to Right): 2000.
Figure 1248 Shadow 200 Geometry Trends – Isometric View. 2020. 2010. 2025 Shadow 200 Wingspan 14 12 Wingspan. Years (Left to Right): 2000. 2015. 2005. ft 10 8 6 4 2 0 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt Figure 1249 Shadow 200 Wing Span Trends 224 .
ft 7 6 5 4 3 2 1 0 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt Figure 1251 Shadow 200 Fuselage Length Trends 225 .Shadow 200 Aspect Ratio 20 Aspect Ratio 15 10 5 0 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt Figure 1250 Shadow 200 Aspect Ratio Trends Shadow 200 Fuselage Length 8 Fuselage Length.
lb/sq. The maximum power to weight difference among all optimized cases was 17%.Shadow 200 Wing Loading and Aspect Ratio 30 25 20 15 10 5 0 2000 W/S.3 Propulsion The Shadow 200 technology case uninstalled power and uninstalled power to weight tends are shown in Figure 1253 and Figure 1254. AR 2005 2010 2015 2020 2025 Design Year Figure 1252 Shadow 200 Wing Loading – Aspect Ratio Relationships 12. This variation was more a function of the solution niche than the design year. HP Figure 1253 Shadow 200 Uninstalled Power Trends 226 .4.ft. respectively. Shadow 200 Uninstalled Power 35 30 25 20 15 10 5 0 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt Uninstalled Power.
01 Su bs ys te m s Pr op ul sio n Av io ni cs St ru ct ur es G ro wt h ad Fu el Pa ylo Em Figure 1255 Shadow 200 Weight Trends 227 De sig n G ro ss pt y . as expected. Figure 1257 shows the weight improvement trends.1 0. respectively.4 Weight Summary Weights Figure 1255 and Figure 1256 show the Shadow 200 technology case weight and mass fraction trends. The Shadow 200 overall weights decreased with improving technology.02 0 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt Figure 1254 Shadow 200 Power to Weight Ratio Trends 12. including the aerodynamics and propulsion. The higher payload mass fraction translated to a higher utility.4.12 0. The decrease in the avionics and fuel mass fractions allowed an overall increase in payload mass fraction. Shadow 200 Weight Summary 1000 100 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt Weight. The overall design gross weight improvement tracked the obligated weight improvement with an offset.Shadow 200 Power to Weight Ratio Power to Weight Ratio.1 0. lbs 10 1 0.08 0.06 0. HP/lb 0.04 0. The offset represented improvement in structures technology and technologies that affect the fuel fraction.
228 . % 50% 40% 30% 20% 10% 0% 2005 W_DG W_Obligated 2010 2015 Design Year 2020 2025 Figure 1257 Shadow 200 Design Gross – Obligated Weights Trend Avionics Weights The Shadow 200 technology case avionics weights trends are shown in Figure 1258. The avionics weights rapidly became negligible contributors by 2010 using the assumed avionics technology curves.Shadow 200 Mass Fraction Summary 70% 60% Mass Fraction 50% 40% 30% 20% 10% 0% Av io ni cs St ru ct ur es G ro wt h on ad Fu el Su bs ys te m s Pr op ul si Pa ylo Em pt y 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt Figure 1256 Shadow 200 Summary Mass Fraction Trends Shadow 200 Improvement Over Previous Period 60% Improvement.
lbs 1 General Avionics 0. 229 . so the communications subsystem weight did not decrease rapidly. The EO/IR became the primary air vehicle sizing driver in later years. Despite applying the small EO/IR technology curve. Like the previous two cases. Shadow 200 Subsystems Weight 100 10 Weight. The flight control system was a function of control surface size. All subsystems except for the communications system were affected by vehicle sizing.1 Baseline 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt Figure 1259 Shadow 200 Subsystems Weight Trends Payload Weights The Shadow 200 technology case payload weight trends are shown in Figure 1260. so the large area of the 2015 and 2025 solutions resulted in a higher flight control weight than the adjacent cases.1 0. the SAR payload weight decreased much more rapidly than the EO/IR. The subsystems weights all decreased with time.001 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt Figure 1258 Shadow 200 Avionics Weight Trends Subsystems Weights Figure 1259 shows the Shadow 200 subsystems weight trends. the EO/IR weight remained high relative to the SAR.Shadow 200 Avionics Weight 100 10 Weight. lbs FCS ECS Fuel Communications 1 0.01 0.
Figure 1263. There were some variations in the optimized flight profiles. and Figure 1265 show the Shadow 200 flight profile and related performance trends. Shadow 200 Structures Weight 100 10 Wing Tails Fuselage Gear Installation Weight. The installation weight is the dominant contributor in all years because of the small vehicle size and high obligated weight fraction. Structures Weights The Shadow 200 structures weights technology trends are presented in Figure 1261.1 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt Figure 1261 Shadow 200 Structures Weight Trends 12. lbs EO/IR SAR 1 0. Figure 1264.4. As seen on Predator. the low technology cases tended to have a higher lift to drag ratio than the later cases. lbs 1 0.1 00 Opt 05 Opt 10 Opt 15 Opt 20 Opt 25 Opt Figure 1260 Shadow 200 Payload Weight Trends.5 Performance Figure 1262.Shadow 200 Payload Weight 100 10 Weight. The structures weights all decreased with time. The relative contribution of the EO/IR ball drag was less on the 230 .
000 14. ft 10.000 12.000 0 0 1 2 3 4 5 6 7 Time. kts Figure 1263 Shadow 200 Velocity Profile Trends 231 .000 2. The high lift to drag ratio of the early cases facilitated a more rapid climb.earlier optimal Shadow 200 cases because the ratio of ball area to wing area was smaller. Shadow 200 Alitude Profile 16.000 4. hrs Figure 1262 Shadow 200 Altitude Profile Trends Shadow 200 Velocity Profile 120 100 80 60 40 20 0 0 1 2 3 4 5 6 7 Time.000 8.000 6.000 S 00 S 05 S 10 S 15 S 20 S 25 Altitude. hrs S 00 S 05 S 10 S 15 S 20 S 25 Velocity.
Technology that affected the obligated weight directly had the greatest impact.15 0. 232 .4. hrs Figure 1265 Shadow 200 Mission Fuel Consumption Trends 12.10 0.05 0.20 0.30 0. hrs S 00 S 05 S 10 S 15 S 20 S 25 Figure 1264 Shadow 200 Mission Lift to Drag Ratio Trends Shadow 200 Specific Fuel Consumption 0. The strongest technology contributors were payloads and avionics/subsystems groups. which is the lowest of all the cases evaluated here.50 0.40 SFC.6 Technology Sensitivities The Shadow 200 design gross weight sensitivities to various technologies are shown in Figure 1266.22% overall synergy.35 0.45 0.Shadow 200 Lift to Drag Ratio 14 12 Lift to Drag Ratio 10 8 6 4 2 0 0 1 2 3 4 5 6 7 Time.00 0 1 2 3 4 5 6 7 Time.25 0. The Shadow 200 had a 4. given the small fuel fraction and associated low weight escalation factor. lbm/lbhr S 00 S 05 S 10 S 15 S 20 S 25 0.
Shadow 200 Technology Sensitivities 20002010 100% Weight Reduction.The higher fuel fraction of the previous cases created greater sensitivities to propulsion technology. % Combined 80% 60% 40% 20% 0% Total Components Combined Total Structures Propulsion Payload Design Avion/Subs Aerodynamics Figure 1266 Shadow 200 Design Gross Weight Sensitivity to Technology 233 . The more modest Shadow 200 total improvement was only slightly higher than the combined payload and avionics/ subsystems groups.
while simple. weight and power model is flexible and covers the considerations of interest. The AV size. The propulsion analysis methods are wellsuited to the simple propulsion technology trends used here. 234 . adequately cover the propulsion phenomena of interest. The unused complex propeller performance model shows great promise despite the long relative computation time. such as airfoil profile performance. Many models. and semianalytic structures modeling provide high utility conceptual design capabilities beyond parametric equations. and those ultimately selected for the final analysis adequately blend effectiveness and computational speed. The alternative of simply specifying a fixed mission profile is more constraining on the design problem. The specialized avionics methods were unused with the exception of the air data system model. Other aerodynamic models employed proved effective for the problem set. The piecewise linear beam model works well for all AV classes covered. The aerodynamics methods are capable of modeling all cases considered. The methodology used here captures important phenomena affecting UAV design for the AV classes considered. Integration of this MDO code onto a personal computer offers great convenience and severs reliance on more expensive computer resources. The tabular drag polars are convenient for mission analysis lookup requirements and for postdesign evaluations.1 AV Design Methodology Conclusions The goal of achieving an effective and robust aircraft design methodology for unmanned aircraft has been realized. The empirical fuselage weights equations should be improved in future research. The methodology has been applied to three different classes of UAVs. propeller performance. though the lack of aerolastic analysis and stiffnessbased design create limitations. The subsystems models rely largely on manned aircraft parametric equations that generally operate outside the intended application. The applied genetic algorithm is well suited to the unmanned aircraft MDO problem. because every system has specific avionics suites. The propulsion models applied. The parametric airfoil formulation adequately captures driving LRN phenomena. Integration of the selfoptimizing mission profiles enables additional freedom in AV optimization. lift distribution prediction. Several analysis methods are available for many disciplines.Chapter 13 Conclusions 13. The noisy design space caused by the selfoptimizing mission profiles and tabular drag polars necessitate a method as tolerant as a GA.
the propulsion technology strongly affects the high performance Global Hawk and Predator families. 13. or perhaps any other. The process of tailoring existing technology trends to those applied resulted in substantial format modification.2 Calibration Cases Conclusions The overall calibration achieved for the three UAV classes was successful. so the target results are not difficult to obtain. the calibration is based on often inconsistent and incomplete data. All three families show overall technology synergies. Despite the inherent limitations of this process. all calibrations are within the bounds of the formulation. which have high fuel mass fractions relative to the Shadow 200. The code optimized many AVs within a family over many technology years without a need for tailoring the design variables. The technologies change such that the reference technology year is difficult to identify. which lends credibility to the AV analysis methodology. The mission profiles. design gross weight. The design gross weight is sensitive to weights that directly affect the obligated weights for all cases. 13. The EO/IR relative weight trends are assumed to match the EO/IR relative performance 235 . However. The relatively steady EO/IR sensor payload weight limits the achievable size reductions. Existing trends frequently did not suit the needs of this research. The vertical separation of the VTails and the wing ensure that no singularities occur in the lift distribution methods. This calibration process. constraint weighting.The parametric geometry formulation used in the optimization cases is robust.3 Technology Cases Conclusions The technology impact study demonstrated the utility of the AV design methodology. experienced turbulent design histories that are difficult to replicate. This research identified the formats for technology trends information that can be applied to AV trade studies. Additionally. or analysis methods. Unmanned aircraft. The design gross weight improves with advancing technology within each vehicle family. Future AV technology modeling could benefit from following the model data requirements developed here. The calibration process provides the user with many means of achieving the desired results. empty weight. such as the Global Hawk and Shadow 200. The minimal tail length parametric functions effectively prevent the tail size to attain unrealistic proportions without adding additional constraints for the optimizer. and other known data was brought into close agreement with the available information on the UAVs of interest. could not replicate the design history of the UAVs of interest. Requirements and design goals evolve over time.
General Atomics with the GnatPredator family. The resulting designs decreased in weight as technology advanced. Miniaturization as a function of technology improvement has been practically demonstrated through the DARPA Micro Aerial Vehicle (MAV) program. Often. Avionics technology impacts are significant for all families considered here. Technologies enable UAV requirements to improve. and TeledyneRyan (now Northrop Grumman) with the long lineage of reconnaissance UAVs that ultimately led to the Global Hawk. and other relevant payloads are not covered. which often results in larger vehicles over time. Electronics Intelligence (ELINT). The unique ATC equipment trades are not considered in this research. Measurements and Signals Intelligence (MASINT). in this scenario. Technology advancement for fixed requirements promotes miniaturization. previously not applicable to a vehicle class. The SAR performance per weight improved approximately 12 orders of magnitude more than the SAR performance during the timeframe under consideration. This is due to new capabilities. The vehicle sizes that support this role have greatly decreased over time.trends here. This behavior is well represented by historical trends. This is to be expected. Communications Intelligence (COMINT). The MAV program achievements lend credibility to the weight reduction predictions. The technically attainable UAV sizes for this role have decreased from the 9foot wingspan Pointer in the 1980’s to the MAV class of today. the requirement for close range battlefield surveillance has existed for decades. Many MAVs with sizes either achieving or approaching the goal were successfully developed. The manufacturer. which is probably conservative. Rather than wait for the technology to further improve. For example. This scenario has been demonstrated by Aerovironment with solar powered aircraft. become available to potential UAV systems that are slightly larger than the existing capability. historical UAV designs within a manufacturer’s line tend to increase in size as technology improves. For example. Flight performance and payload performance requirements were held constant for this research. there is no equivalent performance per weight prediction available for the EO/IR payload. Not all avionics are likely to see the full potential gains modeled in this research. One of the MAV program objectives is to demonstrate miniaturized technologies in vehicles with wingspans less than 6 inches. and is very slow to adapt to change. the UAV manufacturer will develop a new system to accommodate the new mission capability. The avionics contribution to vehicle weight becomes negligible after approximately 2010 with the technology modeling used. has an established capacity for UAV development and gains approval to develop a larger version of the existing family. the ATC equipment is largely driven by general aviation and commercial aviation. because the avionics category includes a significant portion of the obligated weights. Any one of these potential payloads could significantly impact the vehicle design. The General 236 . Only a limited set of potential payloads are considered. Unfortunately.
and other considerations may be drivers for UAVs equipped with conventional landing gear. despite its poor quality. the EO/IR payload weight technology trend became a limitation to the ultimate size reductions of all cases considered here. UAV design tools that capture technology insertion can facilitate requirements evolution. Alternatively.4 Recommendations Technology trends that are directly applicable to UAV design should be developed and refined. this AV design methodology could be used to help define future requirements. Gnat) 13. time progresses from right to left. Predator A. Takeoff and landing speeds. Future UAV design code enhancements should incorporate launch and recovery performance and system models. The greatest resources should be placed on technologies that have the greatest impact on AV design. Tomorrow’s UAV systems will probably not be designed against today’s requirements. 237 . The constraints associated with this analysis will likely drive the configuration. climb angles.note that time progresses from right to left. These requirements could be applied to the AV design methodology to assess the potential solution space more realistically. Improved UAV subsystems weight methods are not possible without knowledge of the applicable design considerations. The detailed propeller analysis method should be evaluated for gains in computational efficiency. The methodology should incorporate offdesign propulsion models for reciprocating engines and turbofans. The performance models can be improved.Atomics GnatPredator family evolution is shown in Figure 1 . As an example. UAV subsystems often operate beyond typical limitations of manned aircraft. The Teledyne Ryan UAV family evolution is shown in Figure 13 again. field length. Electric and fuel cell propulsion system models should be added so that UAVs incorporating these options can be evaluated. Performance requirements for future UAV systems that leverage technology advancements should be further explored. Figure 131 General Atomics GnatPredator family evolution (Left to Right: Predator B. such as the EO/IR payload weight trend. Detailed UAV subsystems definition and design methodology development should be conducted.
The model should be more highly tailored to composites materials. As the vehicle size is reduced. incorporating the poststall airfoil models developed in this research. Improvements should include higher quality secondary structure and control surface analysis. The nonlinear behavior of the airfoils prior to and post stall creates limitations on the fidelity of the matrix methods for lift distribution calculations. Key structural phenomena of interest include flutter analysis and stiffnessbased design capabilities. The wing structural model should be improved to capture more physical phenomena. Parachute recovery systems take up volume that scales as a function of the vehicle weight and descent requirements. and infrared signature levels. and potential decrease in radar cross section will reduce the probability of detection. Potential emphasis can be placed on generating physics and historybased technology trends. Future research efforts could focus on survivability benefits gained though integration of advanced technologies. This research helps define useful formats for these technology predictions. The current limitation that all surfaces operate in the approximately linear region prior to stall imposes limitations on the maximum permissible lift coefficient of the vehicle. A similar numeric method could be applied to the vortex lattice method. In reality. the survivability will likely increase. the overall vulnerability of the AV will decrease due to smaller presented area of flight critical components to threat systems. 238 . As discussed previously. An entire graduatelevel research effort could be dedicated exclusively to defining technology trends. Additionally. Future model should capture local poststall behavior on the configuration. but not a lifting system. Decreased visual. acoustic. existing technology trends do not always lend themselves to integration into aircraft design and optimization codes.Pneumatic and rocket launch system integration will have structural impacts. The numerical lifting line theory presented in Anderson [1991] addresses these considerations for a wing. local stall conditions on some portion of the vehicle typically occur prior to the vehicle achieving the maximum lift coefficient.
SF = η tip (1) ⋅ (1 + ct c r. w ⋅ SF 2 2 bw Sw = ARw 1 S w = ⋅ ARw ⋅ c r2. the index i is dropped from this discussion.w (1) ) + ∑ (η tip ( j ) − η tip ( j − 1) ) ⋅ (ct c r . Figure A1 Wing Geometry Parameters First. and the index for the number of segments for a given wing is j. The index for the number of wings is i. Note that if a single panel wing with a taper ratio of 1 is applied. parameter must be defined. w ⋅ SF 2 4 Sw = The wing root chord equations are: 239 . SF is analogous to a complex taper parameter. w ( j ) − ct c r . w ( j − 1) ) npan j =2 [ ] The wing area equations are: 1 ⋅ bw ⋅ c r . SF.Appendix A Geometry Methods Wing Geometry This section provides the trapezoidal geometry formulation for the wing segments. then SF is equal to 2. For clarity. a nondimensional wing shape factor. Some general geometry parameters of interest for the trapezoidal wing elements are shown in Figure A1.
the following equivalent taper ratio equation is used: λeq = 2 ⋅ Sw −1 bw ⋅ c r . The segment dihedralindependent semispan. for a wing or horizontal tail is found by: s seg (1) = η tip (1) ⋅ bw if j=1 2 240 . quarter chord sweep. w ⋅ SF 2 bw = ARw ⋅ S w The wing aspect ratio equations are: ARw = 2 ⋅ bw c r . w ⋅ SF 2 To convert from the complex wing geometry to an equivalent single taper. leading edge sweep. It is necessary to know the leading edge Cartesian coordinates of the root and tip.w ⋅ SF 2 bw ARw = Sw 4 ⋅ Sw ARw = 2 c r . w ⋅ SF 1 ⋅ ARw ⋅ c r . root and tip chords. w The segment paneling can now begin. sseg.w cr . and the root and tip incidence angles of all wing panels.w The wing span equations are: 1 ⋅ ARw ⋅ SF 2 Sw = 1 ⋅ bw ⋅ SF 2 Sw 2⋅ ARw = SF bw = bw = 2 ⋅ Sw c r .cr .w = bw cr .
w ( j ) = inc w − ε w ( j ) The segment quarter chord sweep for a wing or horizontal tail is: (x / c Λ . w. seg ( j ) = Λ x.25) Λ c / 4. w ( j ) − 0. w ( j ) = inc w − ε w ( j − 1) if j>1 The segment tip incidence angle is found by: inct . w ( j ) ⋅ c r . w ( j ) The segment root incidence angle is found by: inc r .seg ( j ) = c r c r . w ( j ) ⋅ c r . w. seg ( j ) = ct c r . the dihedralindependent segment tip location relative to the wing root for a wing or horizontal tail is: s tip ( j ) = η tip ( j ) ⋅ bw 2 sseg (1) = ηtip (1) ⋅ bw if j=1 and for a vertical tail: stip ( j ) = ηtip ( j ) ⋅ bw The segment root and tip chords are found by: c r . w (1) = inc w (1) if j=1 inc r .w ( j ) + tan −1 ⋅ (c r . seg ( j ) − ct . w. w. w ( j ) ct . seg ( j ) ) bw / 2 and for a vertical tail: 241 .s seg ( j ) = bw ⋅ (η tip ( j ) − η tip ( j − 1) ) if j>1 2 and for a vertical tail: sseg ( j ) = bw ⋅ (ηtip ( j ) − ηtip ( j − 1) ) if j>1 Similarly.
seg ( j ) ⋅ x / cε ( j ) ⋅ sin (inc r . seg ( j ) − ct . w ( j ) )) δy = s seg ( j ) δz = s seg ( j ) ⋅ tan (Λ LE . seg ( j ) ⋅ x / cε ( j ) ⋅ sin (inct . w.w. w. w. w ( j ) ) − c r . seg ( j ) = tan −1 tan (Λ x . w ( j ) ) + ⋅ (cr . seg ( j ) = Λ x. seg ( j ) ⋅ x / cε ( j ) ⋅ (1 − cos(inc r . w ( j − 1) Next. w. w ( j ) ) + ct . w ( j ) ) + ct .w ( j ) + tan −1 ⋅ (cr . w ( j ) ) ∆y = δy ⋅ cos(Γw ( j ) ) ∆z = δz + δy ⋅ sin (Γw ( j ) ) ⋅ cos 2(inc r . the wing root coordinates are: Pr . w. the difference in leading edge coordinates between the segment root and tip if dihedral is neglected is calculated: δx = s seg ( j ) ⋅ tan (Λ LE . Pr . w ( j ) = Pt . seg ( j ) − ct . w. seg ( j ) ) ⋅ cos(inc r . seg ( j ) ⋅ x / cε ( j ) ⋅ (1 − cos(inct . seg ( j ) ) bw The segment leading edge sweep for a wing or horizontal tail is: x / cΛ . w ( j ) ) ⋅ cos(inc r . seg ( j ) = tan −1 tan (Λ x . w ( j ) )) − c r . seg ( j ) ) bw / 2 and for a vertical tail: x / cΛ . w. seg ( j ) − ct . w ( j ) ) Now the effects of dihedral can be taken into consideration: ∆x = δx + δy ⋅ sin (Γw ( j ) ) ⋅ sin (inc r . w.w ( j ) ) + ⋅ (c r . w ( j ) Λ LE . w. seg ( j ) ) bw Now the process of finding the coordinates for the segment root and tip leading edges can begin.w ( j ) Λ LE . seg ( j ) ) ⋅ sin (− inc r . w ( j ) ) 242 . (x / cΛ . w ( j ) − 0.25) Λ c / 4. Starting with the most inboard segment (j=1). w (1) = Pw otherwise (j>1).
The segment wing tip locations are then found by: X t . The second method panels the entire wing by blending segments. w. seg ( j ) ) ⋅ s seg ( j ) ⋅ cos(Γw ( j ) ) ] The overall projected reference span for a wing and horizontal tail is: bref = and for a vertical tail: bref = npan j =1 npan j =1 ∑ [2 ⋅ s seg ( j ) ⋅ cos(Γw ( j ) ) ] ∑ [s seg ( j ) ⋅ cos(Γw ( j ) ) ] Lift Distribution Paneling There are two primary methods of paneling the wing. w ( j ) + ∆y Z t . w ( j ) = Z r . w ( j ) = Yr . seg ( j ) − ct . For the first method. are: ys1 = s seg ( j ) ⋅ k −1 n pan k n pan root tip ys 2 = s seg ( j ) ⋅ The sinusoidal spacing equivalents are: 243 . The spanwise position of the segment root and tip before dihedral considerations is determined. The even spacing root and tip panel spanwise locations.w ( j ) + ∆z The overall projected planform reference area for a wing is: S ref = npan j =1 ∑ [(c r . the following procedure is followed. each wing segment is broken into npan panels. w. w ( j ) = X r . The sinusoidal spacing places more panels towards the tip to more fully capture the rapidly changing behavior at the wing tip. respectively. The first method panels each segment individually. w ( j ) + ∆x Yt . For the index k values from 1 to npan. Either even or sinusoidal spacing may be used.
w ( j ) ) ⋅ The cartesian coordinates for the quarter chord of the panel tip are: 244 . w. seg ( j ) − c r . and the information does not need to be retained. w. w ( j ) + (Yt . w ( j ) + ( X t . ∆ys(k). w ( j ) − inc r . seg ( j ) ) ⋅ ys1 s seg ( j ) ys 2 s seg ( j ) root chord tip chord c 2 = c r . will be used for other calculations. w. w ( j ) ) ⋅ ys1 s seg ( j ) ys 2 s seg ( j ) root incidence tip incidence inc 2(k ) = inc r . w ( j ) + (inct . w ( j ) − inc r .w. w ( j ) ) ⋅ Z1(k ) = Z r . This is because each of these parameters is calculated for each panel. seg ( j ) − c r . seg ( j ) + (ct . w ( j ) − Z r . w ( j ) − Yr . w ( j ) − X r . The difference between ys1 and ys2. w ( j ) ) ⋅ The cartesian coordinates for the quarter chord of the panel root are: X 1(k ) = X r . seg ( j ) + (ct . k −1 π ys1 = s seg ( j ) ⋅ sin ⋅ root n 2 pan k π ys 2 = s seg ( j ) ⋅ sin ⋅ tip n 2 pan Notice that no index is associated with ys1 or ys2. w ( j ) + (inct . w. w ( j ) + (Z t . The nondimensional panel spanwise station is: s tip ( j ) − s seg ( j ) + bw ys1 + ys 2 2 η (k ) = 2 ⋅ The root and tip chords of this panel are: c1 = c r . w ( j ) ) ⋅ ys1 1 + ⋅ c1(k ) s seg ( j ) 4 ys1 s seg ( j ) ys1 s seg ( j ) Y 1(k ) = Yr . w. seg ( j ) ) ⋅ The incidence angles of the root and tip of this panel are: inc1(k ) = inc r .
w ( j ) + (Yt . w ( j ) − X r . w ( j ) − Z r . w ( j ) ) ⋅ ys 2 1 + ⋅ c 2( k ) s seg ( j ) 4 ys 2 s seg ( j ) ys 2 s seg ( j ) Y 2(k ) = Yr . The even spacing root and tip panel spanwise locations are: ys1 = bw ⋅ k −1 n pan k n pan root tip ys 2 = bw ⋅ The sinusoidal spacing equivalents are: k −1 π ys1 = bw ⋅ sin ⋅ root n pan 2 k π ⋅ tip ys 2 = bw ⋅ sin n pan 2 The nondimensional panel spanwise station is: η (k ) = ys1 + ys 2 bw The parameter θ(k) is used for the lifting line method. w ( j ) + ( X t . which is now dependent upon the value of ys1 and ys2. This method is primarily used for the lifting line theory. w ( j ) ) ⋅ Z 2(k ) = Z r . 245 . w ( j ) ) ⋅ The second wing paneling method blends the wing segments such that the panels are distributed from wing root to wing tip rather than from segment root to tip. w ( j ) + (Z t .X 2( k ) = X r . The one major difference is the selection of segment index j. It can be thought of as an angular equivalent of the spanwise station. though the vortex lattice method and Trefftz plane analysis can use it as well. w ( j ) − Yr . It is defined as: θ (k ) = cos −1 ys1 + ys 2 bw Calculation of the other parameters uses the same equations as shown above.
Regardless of which paneling method is used. is found by: φ (k ) = tan −1 The panel incidence angle is: inc(k ) = Z 2(k ) − Z1(k ) Y 3(k ) − Y 1(k ) inc1(k ) + inc 2(k ) 2 246 . except for the xcoordinate. φ(k). The equivalent panel dihedral angle. The average panel chord is: cave (k ) = c1 + c 2 2 The 3/4chord panel control point used in the vortex lattice method and. other parameters can now be determined. in the Trefftz plane analysis is: Xm(k ) = X 1(k ) + X 2(k ) 1 + ⋅ c ave (k ) 2 2 Y 1(k ) + Y 2(k ) Ym(k ) = 2 Z1(k ) + Z 2(k ) Zm(k ) = 2 The point used to simulate the Trefftz plane (infinitely downstream) for induced drag calculations in the vortex lattice method is: Xi (k ) = 100 ⋅ bw Yi(k ) = Ym(k ) Zi (k ) = Zm(k ) where bw is the span of the main wing.
Adkins and Liebeck [1994] does not discuss the airfoil data lookup method in detail. The code was also evaluated for a threebladed propeller under the same design conditions. These methods bring optimal propeller design into agreement with analysis. the results of this code correspond to the sample case with suitable repeatability. which is an improvement on earlier blade element methods. Third. (The airfoil drag affects the circulation distribution radially across the propeller blade). The chord distribution agreement is much better than that of the twist distribution.1 Overview The propeller design and evaluation codes were developed from the work of Charles Adkins and Robert Liebeck described in their paper entitled. viscous terms are included in the induced velocity calculations. Second. but the form of the chord distribution remains the same.(AB) indicate that the local chord size is sensitive to the blade count. The differences between the code and the sample case can be partially explained by the use of different airfoil data methods. Other differences may be explained by subtleties of two different codes. 247 . First. small angle approximations are not necessary. The methods used here differ from those of previous optimal propeller design codes in three ways.2 Blade Element Model Propeller Design Design Results A sample design and analysis case for a general aviation propeller was provided in Liebeck [1994]. The results in Figure B. The blade twist is very insensitive to blade count.Appendix B Blade Element Propeller Methods B. there is no light disk loading requirement on the propeller. In general. Design of Optimum Propellers [1994]. B. There is little difference between the two airfoil interpolation methods.
The bladeelement propeller offdesign analysis is much more realistic at predicting offdesign conditions than the disk actuator method. FT 0. A special model was created as part of the airfoil interpolation subroutines to calculate poststall lift and drag values from 0 to 360 degrees angle of attack. An added advantage of this method is the ability to predict static thrust. By using the angle of attack as a lookup variable. DEG 40 30 20 10 0 0 1 2 3 Local Radius. 248 . Note that the angle of attack and drag coefficients were determined in a similar way in the design code.3 0.1 0 0 1 2 3 Local Radius. Both airfoil interpolation methods are presented here as well.5 Local Chord. rather propellers experience a range of operational conditions. FT 3Blade 2Blade Beta Comparison of 2 and 3 Blade Props 70 60 50 Beta. The agreement with Liebeck [1994] is good in all cases. except these parameters were functions of the lift coefficient and Reynolds number.Chord Comparison of 2 and 3 Blade Props 0. Propellers never act as point designs.2 0. FT (A) (B) Figure B1 Comparison Between Two and Three Bladed Propeller Designs B. The lift and drag coefficients are found as a function of angle of attack and Reynolds number through the airfoil interpolation subroutine. it is possible to find feasible solutions of the lift and drag coefficients even beyond stall.3 Blade Element Model Propeller OffDesign Analysis The offdesign analysis is necessary to determine how the propeller performs during the majority of operation.4 0.6 0. These poststall airfoil methods are described in the aerodynamics chapter.
The pilot typically controls the vehicle by issuing control commands in response to video and other data from the AV. RC systems are generally driven by the model aircraft hobby industry. There are four categories considered: radio control receivers. autonomous autopilots. at minimum. Most autonomous autopilots permit pilots to take control of the AV. RPV autopilots are becoming displaced by small autonomous autopilots. Pilots may or may not be in the flight control loop for launch and recovery operations. the weight increases by 8 ounces. 249 . the weight increases by 8 ounces. The average RC receiver weight is 3 ounces. These numbers are based on typical commercially available RC systems. RPV Autopilots RPV autopilots contain. Autopilots are defined here as electronic systems that control AV functions.1 ounces.1 Autopilot Autopilots permit UAV systems to fly without pilots.Appendix C Selected Avionics and Subsystem Weights Methods C. an inner loop autopilot and a means of determining vehicle angular rates. These weight increases are causes by the additional components plus the weight of the processors. the weight increases by approximately 4.5 pounds. The average RPV autopilot is assumed to weigh 1. If a low grade INS system is included. If a commercially available GPS system is included. Radio Control (RC) Receivers RC receivers have a broad range of sizes and capabilities. Autonomous Autopilots Autonomous autopilots have the ability to perform most flight operations without pilot intervention. The minimum weight of a RC receiver is assumed to be 0. RPV autopilots. If a GPS system is included. and intelligent autopilots. the weight increases by approximately 8 ounces. If an INS system is included. The minimum weight of a RPV autopilot is assumed to be 2 ounces.4 ounces. and operate in a sophisticated RPV mode.
2 Inertial Navigation System (INS) Inertial Navigation Systems (INS) are used for almost all UAVs. performance. optimize collection. After corrections. assist group UAV operations. The difference between autonomous and intelligent autopilots primarily lies in the processing. these correction factors provide an approximation where none previously existed.0 1. Intelligent autopilots can determine vehicle health.5 Table C1 Autopilot Factors A limited database of autopilot systems was established to provide guidance on the range of autopilot solutions. The average autonomous autopilot weight is assumed to weigh 2 pounds. The corrections are valid for avionics boards and boxes. General Autopilot Considerations The autopilot weight is corrected for ruggedness. All UAVs 250 . While not exact. which is based on Nicolai [1984]. and have adaptable characteristics. The reference density is assumed to be 45 pounds/cubic foot. The addition of a GPS system is assumed to weigh 8 ounces. Some unsophisticated systems that are little more than radiocontrolled aircraft are the exception. Driver Ruggedness Performance Miniaturization Minimum (0) 0. and miniaturization. This research influenced the above factors. The factors are shown in Table C1.The minimum weight of an autonomous autopilot is assumed to be 4 ounces.0 1. The volume for an autopilot is calculated. The weights are assumed to be the same as for the autonomous autopilots.4 0. Intelligent Autopilots Intelligent autopilots provide additional functionality over autonomous autopilots. C. because little additional hardware is required. The autopilot power is assumed to be 4 Watts/pound.4 0.5) 1. The actual installed volume occupied by the autopilot is assumed to be 300% of the autopilot alone. then the uninstalled autopilot weight will be set to the minimum autopilot weight.0 1. The addition of an INS system is assumed to add 1 pound.0 Maximum (1) 1. if the uninstalled autopilot weight is less than the minimum autopilot weight.4 Mid (0.5 2. which is driven by software.
Rate based INS systems provide angular rates and sometimes linear accelerations.5 Table C2 INS Factors A limited database of INS systems was established to provide guidance on the range of INS solutions. performance. because GPS is necessary if the INS sensors are crude.5 2. The factors are shown in Table C2.under consideration in this study use an INS. The minimum weight for a navigation INS system is assumed to be twice the weight of a C/A code GPS. as RPVs. The minimum weight for a rate based INS system is assumed to be 2 ounces. A typical navigation INS system is assumed to weight 5 pounds. Rate based INS systems are typically used by the inner loop of the flight control system. which is based upon the popular LN100G INS.0 1. While not exact. and miniaturization. The reference density is assumed to be 70 pounds/cubic foot.1 Mid (0. The INS weight is corrected for ruggedness. if the uninstalled INS weight is less than the minimum INS weight. it was necessary to fly unmanned aircraft as radio controlled models. Prior to the advent of GPS. C. The actual installed volume occupied by the INS is assumed to be 200% of the INS alone. The INS power is assumed to be 1. then the uninstalled INS weight will be set to the minimum INS weight. This research influenced the above factors.3 Global Positioning System (GPS) GPS is an essential element of modern UAV systems. Often times the drift is contained by corrections enabled by GPS or air data sources. A typical rate based INS system is assumed to weigh 2 pounds.4 0. Driver Ruggedness Performance Miniaturization Minimum (0) 0. The volume for an INS is calculated. The two categories of INS covered here are rate based and navigation grade INS systems. or to accept 251 . After corrections.0 1. Navigation INS systems integrate the angular rates to find the angles and integrate the linear accelerations to find velocity.5) 1.0 1. these correction factors provide an approximation where none previously existed.73 Watts/pound. The installed weight of the INS is assumed to be 150% of the uninstalled INS. which is also based on the Litton LN100G.4 0.0 Maximum (1) 1.
While autopilots typically take care of flight controls and managing flight critical systems. C. The GPS power. A typical processor is assumed to weigh 3 pounds. cost. because there are no commercial applications to drive the weight down. and installation factors are also found using boards and boxes functions. or preprocessing sensor data prior to sending to the autopilot. Infrared (IR).4 Processors Processors are typically necessary to perform a variety of tasks. This weight is corrected for ruggedness. payload management. performance.5 0. The weight and volume installation factors are calculated using the INS functions.0 1.0 Table C3 Camera Factors 252 . and miniaturization. The weight of a typical EO camera is assumed to be 0. This can be to aid in takeoff in landing operations or to provide general situational awareness. The power.5 Mid (0. C. Three types of cameras are considered: Electrooptical (EO). Some processor tasks may involve communications management. These cameras are considered distinct from cameras that are part of the payload. and lowlight. and miniaturization according to the boards and boxes functions described earlier. when coupled to an autopilot or other navigation system. performance.5 Cameras Cameras are often needed on UAVs.5 0. Driver Ruggedness Performance Miniaturization Minimum (0) 0.4 ounces based upon the TracPac GPS receiver. The basic GPS weight for a commercially available (C/A) code GPS is assumed to be 4. processors are often required. GPS. volume. allows a UAV to navigate long distances accurately. The weight of a typical IR camera is assumed to be 2 pounds.0 1. The weight of a typical lowlight camera is assumed to be 1 pound.5 pounds.0 Maximum (1) 2. and volume are calculated based upon the weight using the functions for the INS.inaccurate navigation.0 2.5) 1. A precision (P) code GPS is assumed to be 50% heavier.0 4. The factors are shown in Table C3. The camera weight is corrected for ruggedness.
First. Cost: $97.5 pounds Volume: 0. Third. the AV may generate a voice signal to ATC and interpret the ATC voice via voice recognition software.7 Air Traffic Control (ATC) Several ATC options are covered in the code. C. The Filser ATR 720 A [Aircraft Spruce and Specialty 2002] ATC radio is used as a reference.5 pound per gigabyte.56 pounds Volume: 51 in3 Power: 4 Watts output (Assume 16 Watts input – 25% efficiency) Cost: $1095 (2002) The ATC radio requires an antenna. ATC equipment makes it possible for UAVs to operate in the same airspace as manned aircraft. and any other data of interest.The power and volume are found using the boards and boxes functions. the voice communications between the ATC personnel and the UAV ground personnel may be relayed through the AV. The recorder is assumed to weigh 0. While the FAA and foreign ATC services have evolving UAV airspace operability requirements. the equipment covered here will be relevant. The weight and volume installation factors are also found using the boards and boxes functions. C. so the third option is dropped from consideration. The volume and power is found using the associated boards and boxes formulas. the UAV system ground personnel may communicate directly with the local ATC with ground station ATC radios.50 (2002) 253 .6 Recorders Recorders are sometimes necessary for recording payload data. ATC Radios UAV systems may use ATC radios in one of three ways. Second. The radio characteristics are shown below. The weight and installation factors are from the boards and boxes formulas. (Almost all external) Power: 0. flight data. Weight: 1. A typical VHF ATC antenna has the following characteristics: Weight: 0. Only the first two options affect the AV weight.
4 pounds Volume: 33. Factors are applied when entered directly as avionics. Mode S is incorporated into TCAS. It is conceivable that a UAV program could modify the equipment. but this may introduce FAA acceptance issues. The hardware is largely driven by commercial and general aviation. Weight: 0.1 pounds Volume: 83. The reference transponder is the Garmin GTX327 [Aircraft Spruce and Specialty Co. Mode C Mode C transponders have all of the features of the Mode 3A. The AmeriKing altitude encoder [Aircraft Spruce and Specialty Co 2002] is used as a reference. This requires an altitude encoder. C. 2002]. Mode S uses all of the features of Mode C.Mode 3A Mode 3A transponders provide an identification code to ATC when interrogated. ATC equipment is relatively mature.5 in3 Power: 15 Watts Cost: $1895 (2002) The weight of the Mode 3A ATC also includes the ATC antenna weight. Weight: 2. The characteristics are shown below. This is a relatively new mode that is not widely adopted. 254 . This is used for modes 3A. General ATC Considerations No technology factors are applied to the ATC equipment for this formulation. The ATC determines the aircraft position as the ATC radar makes the 360o sweeps.25 in3 Power: 1 Watt (Guess) Cost: $500 (The cost is altitude dependent) All other components and weights from the Mode 3A are included. The characteristics are shown below. and S. Mode S Mode S transponders enable pilots to determine the position of other Mode S aircraft relative to their aircraft. but also add an altitude reporting capability. and the weights are considered identical for the purposes of this code. which are slow to adopt new technologies. an aircraft collision avoidance system.
255 .Installation weight and installation volume factors are applied to the ATC equipment. The functions for avionics boards and boxes are used.
Second Edition. Electric Motor Handbook. David L. Adkins. 2002. Compass Cope Airframe Design History.. 1959. and Von Doenhoff. Band 2. Ira H.. Unmanned Combat Aerial Vehicles: Evolution or Potential Revolution?. A Study of a Reconnaissance Surveillance Vehicle. AIAA 21st Aerospace Sciences Meeting.. McGrawHill. Boucher.. Herman. John D. Inc. and Smith. Troy E. NJ.. Carmichael. Research Paper Presented to Air Force 2025. Stuart F. 1997. 1991. August 1996. Althaus. Anderson.. Inc. Englewood Cliffs. Aircraft Spruce and Specialty Co. GA170 Code Listing. Robert J. Albert E. Adkins. Brown. Aircraft Spruce and Specialty Co. Los Angeles. AIAA830190. SeptOctober 1994.. AIAA Journal (NFI). Charles N. December 1980. Air University. January 1983.. John J.. NeckarVerlag GMBH. Robert B.. December 1998. Dover Publications.. DeVine.. Prentice Hall. Design of Optimum Propellers... Germany. New York. al. April 1998. Altmann. Number 5. Carroll. UAV Study. Robert H. Robert H. Dieter.. 1996. Fundamentals of Aerodynamics. Theory of Wing Sections. Air Force Scientific Advisory Board (SAB). AIAA Journal of Aircraft." Fortune. 256 ... Including a Summary of Airfoil Data. Aerodynamics for Engineers. 1989. August 2002. Michael L. Brown. Strikestar 2025. United States of America. Brown. et. University of Illinois at Urbana Champaign. Course Materials Presented at the UCLA UAV Aircraft Design Short Course. October 27. Design of Optimum Propellers.. 1994. Volume 17. Liebeck. "Batteries are Getting Charged Up. Charles N... Boucher. Bruce W. Second Edition. Bertin. 1985. Profilpolaren fur den Modellflug. Number 12. David R.. Inc.. Volume 10. Air University. Robert J. 20022003 Catalog. Liebeck.Bibliography Abbot.
UAV Annual Report FY 1997. Etkin. Vol 12. David W. Thomas J.. Virginia Tech AOE 5104 Graduate Incompressible Aerodynamics Course Notes.. Charles D. XFOIL 6. Jeffrey A. R. Innovative Development. NagshinehPour. Bernard.. USA. Introduction to UAV Systems. and Gleason. Gundlach. Foch. Drela. VPIAOE255. F. Multidisciplinary Design Optimization and Industry Review of a 2010 StrutBraced Wing Transonic Transport. Addison Wesley Longman. Robert S. AIAA 990003.. December 2000. 1997. Inc. Maryland. Genetic Algorithms in Search.. John F. Mason. Tetrault. R. Defense and Aerospace. Fahlstrom. Global Hawk and Darkstar. 2.. UAV Systems Inc. Virginia Tech..T. Gundlach. et. Volume 37.Davenport. and Machine Learning. P.. Optimization. David E. 1998. Low Reynolds Number Long Endurance Aircraft Design. Dynamics of Flight. Columbia. and Reid. Canada. John F. AIAA. A preliminary Study of Solar Powered Aircraft and Associated Power Trains. April 1995. Gundlach. NASA CR 3699. Drezner. Joel. Fortenbach.. Defense Airborne Reconnaissance Office (DARO). Conceptual Design Studies of a StrutBraced Wing Transonic Transport. Gern. B. and Leonard.A. Stability and Control. Defense Airborne Reconnaissance Office (DARO). A... W..H. John F. 1983.. John Wiley and Sons. 2000. Inc. Paul G. Grasmeyer. 1998. 1998.A. Technology Foresight – Progress Through Partnerships...K. Grossman.K.. Number 6. William.. Office of Science and Technology for HMSO. Development of the Virginia Tech Forestry Unmanned Aerial Vehicle. Haftka. al. Ailinger. 1. J.H. AIAA 921263.. Virginia Tech. 1996. 1992. TrussBraced Wing Code Description and User's Manual. A.. Ko. RAND Project Air Force... 1989.9 User Guide. UAV Annual Report FY 1996. Goldberg. Third Edition. Second Edition. 6 November 1996. 2002. AIAA Journal of Aircraft. MIT Aero and Astro. Kevin G. 1999.H. Lloyd D.. Hall. Kapania. RAND... 1999.. Executive Summary and Vol 13. Masters Thesis. 257 . Richard J. Schetz. Mark.
CONDOR – High Altitude Long Endurance (HALE) Automatically Piloted Vehicle (APV). Design Visualization Using the DXF..F. Libeau.. New York. 1999. Philip G. 1995. Brochure presented at AUVSI 2003. AIAA 903279. Hoerner Fluid Dynamics.F Subroutine. Massachusetts. Jane’s Unmanned Aerial Vehicles and Targets.. China.. David. Kenneth. AIAA. D.. Mattingly. Richard V. 1992. Carl R. 1997. Hall. Col.. S/TAR Strategic/Tactical Airborne Recorder. Structural Sizing of a Solar Powered Aircraft. Top Secret History of Remote Controlled Airborne Weapons. New York. AddisonWesley Publishing Company.. 1990..Hall. An Illustrated History.. Paducah. Unpublished documentation supporting the Virginia Polytechnic and State University StrutBraced Wing (SBW) transonic transport research. Stan A. Arntz. SemiAnalytical Method for Predicting Wing Structural Mass.. Bill. Hoerner. Et. David W. and Oliver. 1988. Peterson. London. Irwin. Inc. Jr. Hill. Kenneth. Italy. Johnstone. 54th Annual Conference of the Society of Allied Weights Engineers. H. Macci. UK. 1992. 1995. AIAA Education Series.. John Wiley and Sons. Turner Publishing Co. Kentucky. Munson. Mechanics and Thermodynamics of Propulsion.. Holder.. Issue Thirteen. McDaid. Mises.. Jane’s Information Group Limited. 2000. S. Al. David W. 2003. WA. R. Barnes and Noble. Aerodynamics Aeronautics and Flight Mechanics. Michael. 258 . Huntsville Alabama. FluidDynamic Drag. AIAA. McCormick. Reconnaissance is Black. Schiffer Publishing Inc.. Inc. S. 1959. L3 Communications. Aircraft Engine Design. Theory of Flight.. Smart Weapons. Munson. Dover Publications. Unmanned Vehicles. World Unmanned Aircraft. NASA Contractor Report 172313. 2001. Inc. Toronto.. Barnes W. Second Edition. Jane’s Publishing Company Limited.. 1987. Reading. May 1999. Hugh. J.. 1984. N. Inc. Vancouver. Second Edition.
Reston. Ahmed K. The National Academy of Sciences. Robert C. 2003. AIAA. Virginia Polytechnic Institute and State University. Review of ONR’s Uninhabited Combat Air Vehicles Program. Palumbo. Future Aeronautical and Space Systems. 1984. Office of the Secretary of Defense (OSD). The Aerothermodynamics of Gas Turbine and Rocket Propulsion. Progress in Astronautics and Aeronautics.. 2000. G. C. Volume 188. and Beno. UCAVs and Commercial Satellites: "The Missing Link". Washington DC. National Academy Press. Unmanned Systems 2000. 1989. Fundamentals of Aircraft Design. William C.. 2000. Washington DC. Air War College. New York. November 2002. Air University. 20022027. Washington DC. Enabling Science for Military Systems. Air University. The National Academy of Sciences. 259 .. Andrew M. RQ4A Global Hawk. Amir H. Uninhabited Air Vehicles . Inc. McGraw Hill. 2000. VPIAOE256. Office of the Secretary of Defense (OSD). Thomas G. Oates. 1. Volume 172.. USA. 2. Draft Version 7. 1997. Mark J. Nelson. METS. Noor. Air Command and Staff College. Ahmed K. High Altitude Endurance Unmanned Aerial Vehicle. April 1998. Washington DC. Nicolai. O'Reilly. Department of Aerospace and Ocean Engineering. Nichols. and Martel. Dominic J. Reston. VA. AIAA. Mexico.... Flight Stability and Automatic Control. Structures Technology for Future Aerospace Systems. 1998. Preliminary Structural Analysis of a StrutBraced Wing.. Northrop Grumman 452AS4209_06. inc..03.. April 2001.Critical Leverage Systems for Our Nation's Defense in 2025. Noor. AIAA Education Series. Brochure presented at AUVSI 2003.. AIAA.. Northrop Grumman. National Academy Press. Unmanned Aerial Vehicles Roadmap 20002025. 2000. Leland M. Association of Unmanned Vehicle Systems International. April 1999. Uninhabited Air Vehicles.NaghshinehPour. Unmanned Aerial Vehicles Roadmap. Design Evolution of the Shadow 200 Tactical Unmanned Aerial Vehicle. Progress in Astronautics and Aeronautics.. VA. 1984.
. 1992.. Donovan. Selig. Daniel P.C. UAV Aircraft Design. 1995. Daniel P. C. John F.. Ottawa. Summary of LowSpeed Airfoil Data. P. Stinton. 1992.. SoarTech Publications. Soartech 8.Power Computing Solutions. William P. Virginia Beach. Aircraft Design: A Conceptual Approach. Fallbrook. California. AIAA Education Series. Dan M. Aerodynamic Characteristics of Seven Airfoil Sections Through 180 Degrees Angle of Attack for Use in Aerodynamic Analysis of Vertical Axis Wind Turbines. 1985. Lightning Bugs and other Reconnaissance Drones.. Jeffery. James F. Sloan. 260 . Michael S. Roskam. Col. Al. NY. Sullivan. NASA/CR97206337. VA. 2nd Edition.. The cando story of Ryan’s unmanned ‘spy planes’. D. Washington. Inc. Van Nostrand Reinhold Company. Virginia Beach. Gregory. E. Long Island. Texas. Michael S. Albuquerque. Jan. Introduction to Aircraft Weight Engineering. Joles. VA. Design and Packaging of Electronics Equipment.. New York. Oxford. Roskam Aviation and Engineering Corporation. SAND802114. The SM701 Airfoil. Et.. and Fraser. An Airfoil for World Class Sailplanes. Lt. William. 1982. Joel. Selig. July 1998. Raymer. 1989.. 1998 Army After Next Unmanned Aerial Vehicle Studies. Volume 1. Mark D. Sheldahl. SAWE (Society of Allied Weights Engineers). The Design of the Aeroplane. William. Somers. Fireflies and other UAVs (Unmanned Aerial Vehicles). SAWE Inc. and Klimas. Airfoils at Low Speed. Arlington. Wagner. R. Aerofax Inc. Operations Research Center. New Mexico. 1997. Stokely. Brouillette. Maj. Parts IVIII. Circa 1992. 1996.. D.. Los Angeles. 1990. Ritchie. Armed Forces Journal International. David B. Blackwell Sciences Ltd. Airpane Design. Wagner. Kansas. Electric Power System for High Altitude UAV Technology Survey. Maughmer. Maj.A. Sloan. 1968. Sandia National Laboratories. The Pennsylvania State University. Darrol.. Raymer.. Aerodynamics Course Notes. H.. 1995. March 1981. (PCS). August 2002.. UCLA Short Course Materials. Academy of Aeronautics.
261 . UK. September 1985. Airlife Publishing Ltd. The Evolution of the Cruise Missile. R. Alabama.Werrell. 2000. Maxwell AFB.. Fundamentals of Fighter Design. Kenneth P.. Whitford. Air University Press..
Navigation and Control Global Positioning System High Altitude Long Endurance High Altitude Powered Platform Identify Friend of Foe Inertial Navigation System Infrared LeftHandSide Lifting Line Theory Line of Sight Low Reynolds Number Medium Altitude Endurance Measurements and Signals Intelligence Micro Aerial Vehicle MultiDisciplinary Design Optimization RightHandSide Remotely Operated Aircraft Remotely Piloted Vehicle Satellite Communications Society of Allied Weights Engineers StrutBraced Wing Sea Level Static Conditions Size. Weight and Power Unmanned Aerial Vehicle 262 .List of Acronyms AAI AC ACTD AIAA ATC AUVS AUVSI AV BLOS CFD COMINT DGPS ECS ELINT EO EO/IR FCS FOM FOS GN&C GPS HALE HAPP IFF INS IR LHS LLT LOS LRN MAE MASINT MAV MDO RHS ROA RPV SATCOM SAWE SBW SLS SWaP UAV Company name Aerodynamic Center Advanced Concept Technology Demonstrator American Institute of Aeronautics and Astronautics Air Traffic Control Association of Unmanned Vehicle Systems Association of Unmanned Vehicle Systems International Air Vehicle Beyond Line of Sight Computational Fluid Dynamics Communications Intelligence Differential Global Positioning System Environmental Control System Electronics Intelligence ElectroOptical ElectroOptical/Infrared Flight Control System Figure of Merit Factor of Safety Guidance.
UCAS UCAV UFP UIUC VLM Uninhabited Combat Aircraft System Uninhabited Combat Aerial Vechicle Unit Flyaway Price University of Illinois at Urbana Champaign Vortex Lattice Method 263 .
264 . minor subroutines are not listed when there are significant differences in level. The consolidated nomenclature definition list is defined initially within the document.Formatting Notes The code architecture descriptions use capitalized and italicized letters to name subroutines. Some variables with limited usage are defined within the text of the document and not captured within the consolidated nomenclature definition list. When a single variable name represents two or more variables. The names presented may differ from the source code names when clarity is gained from doing so. then the variable name is used once and is followed by all definitions. The units of variables are covered in the nomenclature list to streamline the rest of the document. When practical. The functional architecture is accurate.
won first prize in the AIAA 1998 MidAtlantic Regional Student Conference. and received several other honors and awards. was chairman of the Virginia Tech AIAA. At 17. published three AIAA papers. a Mars atmospheric flight vehicle. His SBW research resulted in four published papers. He also developed radiocontrolled flying engineering models and conducted Multidisciplinary Design Optimization (MDO) work for Aurora Flight Sciences on Jason. he worked at Naval Research Laboratory on over a dozen UAV programs. led the NASA Forestry UAV project. 265 . At age 15. John went on to complete his Masters degree in Aerospace Engineering from Virginia Tech in 1999. He graduated Suma Cum Laude in undergraduate Aerospace Engineering at Virginia Tech in 1998. John works for the United States Government today. While an undergraduate. Gundlach IV has worked on several manned and unmanned aircraft research and development projects. with many original designs. working on the StrutBraced Wing (SBW) MDO project. including a Journal of Aircraft article. where he helped build the fullscale mockup of the Theseus highaltitude UAV. During the undergraduate years.Vita John F. was the team leader of the 2nd place winner of the 1997 AIAA Design/Build/Fly UAV Competition. Andes National Award. John joined Aurora Flight Sciences. John helped build a wind tunnel and static display model of a 5060 passenger turboprop design for a fledgling aircraft company. John has built over 50 radio controlled model aircraft starting at the age of 13. John received the 1998 Sigma Gamma Tau National Aerospace Honor Society Ammon S.